Aquaculture research, tập 41, số 4, 2010

In vitro exposure of Atlantic salmon foregut to brio anguillarum at two concentrations [6 104 Vi-and 6 106colony-forming units CFU mL 1] resulted in clear changes in the intestinal epith

REVIEW ARTICLE

Lactic acid bacteria vs pathogens in the

gastrointestinal tract of fish: a review

Einar Ring1,2, Lisbeth Lvmo1, Mads Kristiansen1,Yvonne Bakken1w, Irene Salinas3,

Reidar Myklebust4, Rolf Erik Olsen2& Terry M Mayhew5

1 Department of Marine Biotechnology, Norwegian College of Fishery Science, University of Troms,Troms, Norway

2 Institute of Marine Research, Bergen, Norway

3 Fish Innate Immune System, Department of Cell Biology, University of Murcia, Murcia, Spain

4 Institute of Anatomy and Cell Biology, University of Bergen, Bergen, Norway

5 School of Biomedical Sciences, Queen’s Medical Centre, University of Nottingham, Nottingham, UK

Correspondence: E Ring, Department of Marine Biotechnology, Norwegian College of Fishery Science, University of Troms, N-9037 Troms, Norway E-mail: einar.ringo@nfh.uit.no

Present address: Lisbeth Lvmo, Gransveien 34,7048 Trondheim, Norway.

w

Present address: Yvonne Bakken, Skretting, 8450 Storkmarknes, Norway.

Abstract

Intensive ¢sh production worldwide has increased

the risk of infectious diseases However, before any

infection can be established, pathogens must

pene-trate the primary barrier In ¢sh, the three major

routes of infection are the skin, gills and

gastrointest-inal (GI) tract The GI tract is essentially a muscular

tube lined by a mucous membrane of columnar

epithelial cells that exhibit a regional variation in

structure and function In the last two decades, our

understanding of the endocytosis and translocation

of bacteria across this mucosa, and the sorts of cell

damage caused by pathogenic bacteria, has

in-creased Electron microscopy has made a valuable

contribution to this knowledge In the ¢sh-farming

industry, severe economic losses are caused by

fur-unculosis (agent, Aeromonas salmonicida spp

salmo-nicida) and vibriosis [agent, Vibrio (Listonella)

anguillarum] This article provides an overview of

the GI tract of ¢sh from an electron microscopical

perspective focusing on cellular damage (speci¢c

at-tack on tight junctions and desmosomes) caused by

pathogenic bacteria, and interactions between the

‘good’ intestinal bacteria [e.g lactic acid bacteria

(LAB)] and pathogens Using di¡erent in vitro

meth-ods, several studies have demonstrated that

co-incu-bation of Atlantic salmon (Salmo salar L.) foregut

(proximal intestine) with LAB and pathogens can

have bene¢cial e¡ects, the cell damage caused by thepathogens being prevented, to some extent, by theLAB However, there is uncertainty over whether ornot similar e¡ects are observed in other species such

as Atlantic cod (Gadus morhua L.) When discussingcellular damage in the GI tract of ¢sh caused bypathogenic bacteria, several important questionsarise including: (1) Do di¡erent pathogenic bacteriause di¡erent mechanisms to infect the gut? (2) Doesthe gradual development of the GI tract from larva toadult a¡ect infection? (3) Are there di¡erent infectionpatterns between di¡erent ¢sh species? The presentarticle addresses these and other questions

Keywords: probiotics, pathogenic bacteria, gutintegrity, ¢sh

IntroductionWith the development of commercial aquaculture, ithas become apparent that diseases can be a signi¢-cant limiting factor Major bacterial pathogens of ¢shinclude the Gram-negative species, Aeromonas salmo-nicida,Vibrio (Listonella) anguillarum,Vibrio (Aliivibrio)salmonicida and Yersinia ruckeri, the aetiologicalagents of furunculosis, vibriosis, cold-water vibriosisand red mouth disease respectively In addition, Aero-monas hydrophila may cause infections in ¢sh and is

generally associated with small surface lesions,

sloughing of scales, local haemorrhage and

septicae-mia All these diseases are common worldwide and

produce considerable economic losses during

inten-sive aquaculture of trout and salmon (Austin &

Aus-tin 1999)

The best way to avoid disease problems in a system

seems to be through e¡ective management practices,

i.e management of stock, soil, water, nutrition and

environment A number of other approaches have

been applied in an attempt to address the disease

pro-blem including sanitary prophylaxis, disinfection

and chemotherapy, with particular emphasis on the

use of antibiotics The application of antibiotics and

other chemicals to pond culture is also quite

expen-sive and undesirable and might lead to antibiotic

re-sistance In Norway, the use of antimicrobial drugs

has decreased from approximately 50 metric tonnes

in 1987 to 746.5 kg in 1997, measured as active

com-ponents (Verschuere, Rombaut, Sorgeloos &

Ver-straete 2000) In 2007, use was still at the same low

level as in 1997 (Norwegian Scienti¢c Committee for

Food Safety 2009) and, recently, vaccinations against

speci¢c diseases have been used

Intensive ¢sh production has increased the risk of

infectious diseases worldwide (Press & Lillehaug

1995; Karunasagar & Karunasagar 1999) but, to

pre-vent microbial entry, ¢sh deploy protective

mechan-isms to hinder translocation of pathogens across the

primary barriers These include production of mucus

by goblet cells, the apical acidic microenvironment of

the intestinal epithelium, cell turnover, peristalsis,

gastric acidity, lysozyme and antibacterial activity of

epidermal mucus (Birkbeck & Ring 2005) At the

same time, pathogenic microorganisms have evolved

mechanisms to penetrate these barriers

Probiotics may reduce the incidence of disease or

lessen the severity of outbreaks One of the proposed

de¢nitions of probiotics used in aquaculture is ‘live

microbial cultures added to feed or environment

(water) to increase viability (survival) of the host’

(Gram & Ring 2005) Probiotic mechanisms include

the production of inhibitory substances against

pathogens, competition for essential nutrients and

enzymes resulting in enhanced nutrition in the host

and the modulation of interactions with the

environ-ment and developenviron-ment of bene¢cial immune

re-sponses (Ring & Gatesoupe 1998; Verschuere et al

2000; Balcazar, de Blas, Ruiz-Zarzuela, Cunningham,

Vendrell & Muzquiz 2006; Gomez & Balcazar 2008)

Today, it is generally accepted that lactic acid

bac-teria (LAB) form part of the normal intestinal

micro-biota of ¢sh from the ¢rst few days of life (Ring &Gatesoupe 1998; Ring 2004; Ring, Schillinger &Holzapfel 2005) One of the most important goals formicrobiologists has been to obtain a stabile indigen-ous microbiota in ¢sh The practical e¡ect of thisactivity is the exclusion of invading populations ofnon-indigenous microorganisms, including patho-gens that attempt to colonize the gastrointestinal(GI) tract (Ring et al 2005) The antagonistic e¡ects

of gut microbiota against pathogens and other isms are possibly mediated by competition for nutri-ents and adhesion sites, formation of metabolitessuch as organic acids and hydrogen peroxide andproduction of bacterocins (for a recent review, de-voted to antimicrobial activity of LAB isolated fromaquatic animals, see Ring et al 2005) A fundamen-tal question arises when discussing the protectiverole of the GI tract microbiota: can the GI tract of ¢shserve as a port of entry for pathogens? During the last

organ-25 years, numerous papers have suggested that thealimentary tract is involved in Aeromonas and Vibrioinfections (Gro¡ & LaPatra 2000; Birkbeck & Ring2005; Harikrishnan & Balasundaram 2005; Ring,Salinas, Olsen, Nyhaug, Myklebust & Mayhew 2007;Ring, Myklebust, Mayhew & Olsen 2007; Salinas,Myklebust, Esteban, Olsen, Meseguer & Ring 2008).Therefore, one can hypothesize that LAB and otherbene¢cial bacteria colonizing the GI tract by produ-cing, for example, bacterocins may o¡er protectionagainst invading ¢sh pathogens

The aim of this review is to present information onthe interaction between bene¢cial bacteria, in ourcase LAB, and pathogenic agents in the digestivetract of ¢sh using in vitro methods

Pathogens and cell damageHistorically, Aeromonas salmonicida ssp salmonicida(A salmonicida), the causative agent of furunculosis,has been recognized as one of the most importantbacterial salmonid pathogens because of its severeeconomic impact, especially on the aquaculture in-dustry (Olivier 1997; Bricknell, Bron & Bowden2006) As early as 1930, the Furunculosis Committeesuggested the intestine as a valuable site for isolatingthe bacteria (Mackie, Arkwright, Pyrce-Tannatt,Mottram, Johnston & Menzies 1930) Since then, con-troversy has existed as to whether or not the gut canfunction as an infection route for this and otherpathogenic bacteria The presence of A salmonicida

in the intestine of some ¢sh species (Ring, Olsen,

verli & Lvik 1997; Ldemel, Mayhew, Myklebust,

Olsen, Espelid & Ring 2001; Pedersen & Dalsgaard

2003), together with evidence that salmonids fed

diets containing probiotic bacteria or soybean meal

showed changes in mortality rate after cohabitant

challenge with A salmonicida (Krogdahl,

Bakke-McKellep, Rd & Bverfjord 2000; Robertson,

O’Dowd, Burrels, Williams & Austin 2000), indicates

that the intestine can be an important route of

infec-tion It is known that pathogenic bacteria produce a

wide array of virulent factors (including

haemoly-sins, cytotoxins, enterotoxins, endotoxins and

adhe-sins), which can a¡ect intestinal barrier function and

facilitate translocation (Chopra, Xu, Ribardo,

Gonza-lez, Kuhl, Peterson & Houston 2000) Translocation

mechanisms include increased receptor-mediated

endocytosis (Skirpstunas & Baldwin 2002),

in-creased paracellular permeability mediated by e¡ects

on junctional complexes and the cytoskeleton, and

direct damage to the intestinal cells (Fasano 2002)

In ¢sh, bacterial pathogens can enter the host by

one or more of three di¡erent routes: (a) skin, (b) gills

and (c) GI tract (Birkbeck & Ring 2005; Ring,

Myk-lebust, et al 2007) If the GI tract is involved as an

in-fection route, mucosal adhesion is considered to be a

critical early phase in all infections caused by

patho-genic bacteria (Knudsen, Srum, McLPress & Olafsen

1999; Namba, Mano & Hirose 2007) When the

bac-teria are able to colonize the intestinal mucus, they

can cross the GI tract lining by transcellular or

para-cellular routes However, it is suggested that

translo-cation of bacteria across the intestine, an essential

and prerequisite step for bacterial invasion, cannot

be studied e¡ectively using in vivo models

Table1pre-sents an overview of in vivo and in vitro studies of cell

damage in the GI tract of ¢sh Two di¡erent in vitro

methods, the Ussing chamber and intestinal sac, have

been used to evaluate translocation and cell damage

caused by pathogenic bacteria (Ring, Jutfelt,

Kana-pathippillai, Bakken, Sundell, Glette, Mayhew,

Mykle-bust & Olsen 2004; Jutfelt, Olsen, Glette, Ring &

Sundell 2006; Lvmo 2007a, b; Ring, Salinas, et al

2007; Salinas et al 2008)

The translocation of viable bacteria from the

diges-tive tract into enterocytes has been reported in

sev-eral investigations (for a review, see Ring, Olsen,

Mayhew & Myklebust 2003; Ring, Myklebust, et al

2007) The phenomenon has mainly been observed

for non-pathogenic indigenous gut bacteria, but not

for pathogenic bacteria, and does not normally

com-promise cellular integrity The situation for

patho-genic bacteria is completely di¡erent, as severe

damage with loss of cellular integrity has been noted

in an in vitro study where the foregut of Atlantic mon was exposed toA salmonicida (Ring et al 2004)

sal-as well sal-as in vivo infection of Atlantic salmon (Bakken2002) and spotted wol⁄sh fry (Anarhichas minor Olaf-sen) by V anguillarum (Ring, Mikkelsen, Kaino,Olsen, Mayhew & Myklebust 2006) As no cell da-mage was observed in control ¢sh (not exposed topathogenic bacteria), we concluded that the indigen-ous bacteria do not a¡ect cellular integrity Ring

et al (2004) used the Ussing chamber technique andobserved detached, but almost intact, enterocytes inthe foregut lumen after exposure to A salmonicida invitro A similar result has been reported in the pyloriccaeca of Atlantic salmon in an in vivo challenge ex-periment (Fig 1) and in an in vitro intestinal sac pre-paration (Ring, Salinas, et al 2007; Salinas et al.2008) However, in the in vitro experiment of Ring

et al (2004), a quite di¡erent situation seems to occur

in the hindgut region (distal intestine) as no intactenterocytes were found in the lumen, the microvilliwere disintegrating and some damage to intercellu-lar tight junctions and desmosomes was observed.The di¡erences between foregut and hindgut havenot been elucidated, but it is probable that entero-cytes in di¡erent regions of the GI tract vary in theirsusceptibility to pathogen-induced damage This may

be linked to di¡erent regional rates of epithelial over or di¡erent mechanisms of enterocyte loss byapoptosis or necrosis (Mayhew, Myklebust,Whybrow

turn-& Jenkins 1999) Apoptosis-dependent processes tend

to preserve junctional integrity while necrosis-likeprocesses tend to be associated with junctional com-plex disruption and loss of microvillous morphology(T M Mayhew, pers comm.)

In vitro exposure of Atlantic salmon foregut to brio anguillarum at two concentrations [6 104

Vi-and

6 106colony-forming units (CFU) mL 1] resulted

in clear changes in the intestinal epithelium pared with samples exposed only to Ringer solution(control) (Ring, Salinas, et al 2007) At the highestdose there was an in£ammatory response of gut-as-sociated lymphoid tissue involving leucocytes mi-grating from the lamina propria towards the lumen.The di¡erence in bacterial translocation betweenindigenous intestinal bacteria and pathogens might

com-be due to the production, by pathogens, of variousvirulence factors such as extracellular enzymes, out-

er surface components such as S-layer or secretoryproteins and pore-forming toxins (Fivaz & van derGoot 1999) These factors could result in severe celldamage as demonstrated in the foregut of Atlanticsalmon (Ring et al 2004)

Human studies have shown that di¡erent bacteria

species colonize di¡erent parts of the GI tract

(Madi-gan, Martinko & Parker 2000), and that di¡erent

pathogenic bacteria adhere to, and infect, di¡erent

parts of the GI tract Ring et al (2004) suggested

that the foregut of Atlantic salmon is an infection site

for A salmonicida However, as loosening of cell

junc-tions was observed in the hindgut, this region is also

probably involved in A salmonicida infection but to a

lesser extent than the foregut

When discussing intestinal cellular damages, the

results presented for the foregut by Ring et al

(2004) are quite similar to the severe epithelial

da-mage associated with intracellular fat accumulation

from dietary linseed oil (Olsen, Myklebust, Ring &

Mayhew 2000) In these studies, cell debris was also

observed in the lumen, providing free access to the

epithelial basal membrane As no information per se

is available about cell damage and dietary

compo-nents and pathogens, this topic should be given high

priority in future studies, especially as there is

in-creasing interest in substituting ¢sh oils and ¢sh

meal with vegetable products

GI microbiota and probiotics in fish

Generally, probiotic strains have been isolated from

indigenous and exogenous microbiota of aquatic

ani-mals Earlier publications stated that bacteria

belong-ing to genus Photobacterium, Pseudomonas and Vibrio

were retrieved among the dominant genera in the

in-testine of marine ¢sh (Cahill 1990; Sakata 1990;Ring Strm & Tabachek 1995) while the indigenousmicrobiota of freshwater ¢sh species tend to bedominated by members of the genera Aeromonas,Plesiomonas, representatives of the family Enterobac-teriaceae, and obligate anaerobic bacteria of the gen-era Bacteroides, Fusobacterium and Eubacterium(Cahill 1990; Sakata 1990; Ring et al 1995) Recentpublications allow one to question whether this istrue, particularly the knowledge and experiencegained from bacteriological studies using, for exam-ple, polymerase chain reaction-denaturing gradientgel electrophoresis (PCR-DGGE) (Gri⁄th, Melville,Cook, Vincent, St Pierre & Lanteigne 2001; Jensen,

vres, Bergh & Torsvik 2004; Pond, Stone & man 2006; Hovda, Lunestad, Fontanillas & Rosnes2007; Kim, Brunt & Austin 2007; Liu, Zhou,Yao, Shi,

Alder-He, Benjamisen Hlvold & Ring 2008; Zhou, Liu,Shi, He,Yao & Ring 2009)

It is important to note that the population ofendogenous microbiota may depend on genetic, nu-tritional and environment factors However, microor-ganisms present in the immediate environment ofaquatic species have a much larger in£uence onhealth status than is the case with terrestrial animals

or humans The gut microbiota of aquatic animalsprobably comprise indigenous microbiota togetherwith arti¢cially high levels of allochthonous bacteriamaintained by their constant ingestion from the sur-rounding water (Ring & Birkbeck 1999)

While several studies on probiotics have been lished during the last decade, the methodological andethical limitations of animal studies make it di⁄cult

pub-to understand the mechanisms of action of tics, and only partial explanations are available.Nevertheless, possible bene¢ts linked to administer-ing probiotics have been suggested They include: (i)competitive exclusion of pathogenic bacteria (Mor-iarty 1997; Gomez-Gil, Roque & Turnbull 2000; Bal-caŁzar, de Blas, Ruiz-Zarzuela, Vendrell & Muzquiz2004; Vine, Leukes & Kaiser 2004; Ring et al 2005;Baccazar, Vendrell, de Blas, Ruiz-Zarzuela, Girones &Muzquiz 2007); (ii) source of nutrients and enzymaticcontribution to digestion (Prieur, Nicolas, Plusquellec

probio-& Vigneulle 1990; Sakata 1990; Ring probio-& Birkbeck1999); (iii) direct uptake of dissolved organic materialmediated by the bacteria (Moriarty 1997); (iv) en-hancement of the immune response against patho-genic microorganisms (Andlid, VaŁzquez-JuaŁrez &Gustafsson 1995; Scholz, Garcia-Diaz, Ricque, Cruz-Suarez, Vargas-Albores & Latchford 1999; Rengpipat,Rukpratanporn, Piyatiratitivorakul & Menasaveta

Figure 1 Transmission electron microscopy micrograph

of the pyloric caeca from Atlantic salmon (Salmo salar L.)

challenge with Aeromonas salmonicida ssp salmonicida

The enterocytes are damaged and their organelles are

ex-posed to the gut lumen Several detached enterocytes are

seen in the gut lumen (arrows) After Bakken (2002) MV,

microvilli

2000; Gullian & Rodr|¤guez 2002; Irianto & Austin

2002a, b; Balcazar 2003; BalcaŁzar et al 2004); and

(v) antiviral e¡ects (Kamei,Yoshimizu, Ezura & Kimura

1988; Girones, Jofre & Bosch 1989; Direkbusarakom,

Yoshimizu, Ezura, Ruangpan & Danayadol 1998)

Most probiotics proposed as biological control

agents in aquaculture belong to the LAB group

(Lac-tobacillus, Lactococcus, Carnobacterium, Pediococcus,

Enterococcus and Streptococcus), genus Vibrio (Vibrio

alginolyticus), genus Bacillus, genus Pseudomonas,

genus Roseobacter although other genera or species

have also been mentioned (Aeromonas, Alteromonas

and Flavobacterium) Readers with special interest in

the di¡erent aspects of the use of probiotics in

aqua-culture are referred to the comprehensive reviews of

Ring and Gatesoupe (1998), Gatesoupe (1999),

Ring and Birkbeck (1999), Skjermo and Vadstein

(1999), Gomez-Gil et al (2000), Irianto, Roberwen,

Austin and Pandalai (2000), Olafsen (2001),

Verschuere et al (2000), Irianto and Austin (2002a),

Ring (2002, 2004), Burr, Gatlin and Ricke (2005),

Gram and Ring (2005), Hong, Duc and Cutting

(2005), Ring et al (2005), Balcazar, de Blas, et al

(2006), Balcazar, Decamp,Vendrell, de Blas and

Ruiz-Zarzuela (2006), Farzanfar (2006), Vine, Leukes and

Kaiser (2006), Gatesoupe (2007, 2008),

Kesarcodi-Watson, Kaspar, Lategan and Gibson (2008) and

Tinh, Dierckens, Sorgeloos and Bossier (2008)

LAB

It is well documented that LAB constitute a part of

the indigenous gut microbiota of several ¢sh species

(Ring & Gatesoupe 1998; Ring 2004; Ring et al.,

2005; Balcazar, de Blas, Ruiz-Zarzuela,Vendrell, Calvo,

Marquez, Girones & Muzquiz 2007; Michel, Pelletier,

Boussaha, Douet, Lautraite & Tailliez 2007) In most

of these studies, di¡erent species of Carnobacterium

have been isolated, but Lactobacillus species have also

been isolated from the digestive tract of ¢sh (for a

re-view, see Ring & Gatesoupe 1998; Hagi,Tanaka,

Iwa-mura & Hoshino 2004; Ring 2004; Ring et al 2005;

Balcazar et al 2007; Michel et al 2007; Liu et al 2008)

Carnobacterium divergens vs A salmonicida

andV anguillarum in Atlantic salmon foregut

In two reviews devoted to LAB in ¢sh and ¢sh

farm-ing, Ring (2004) and Ring et al (2005) suggested

that LAB and other bene¢cial intestinal bacteria

might be involved in the primary defence system

against pathogenic colonization and adherence ofpathogenic bacteria in the GI tract Several LAB iso-lated from ¢sh and aquatic animals display antago-nistic activity against ¢sh pathogenic agents (Ring

et al 2005; Ring 2008) To test the hypothesis thatLAB can prevent pathogen-induced damage, Ring,Salinas, et al (2007) used C divergens strain 6251, ori-ginally isolated from the foregut of the Artic charr(Salvelinus alpinus L.) (Ring & Olsen 1999) Thisstrain has growth-inhibitory e¡ects against both

A salmonicida and V anguillarum (Ring, Sepploa,Berg, Olsen, Schillinger & Holzapfel 2002; Ring2008) The aim was to investigate, by means of lightand electron microscopy, the structural changes thatAtlantic salmon intestine underwent following invitro exposure to A salmonicida, V anguillarum and

C divergens at two di¡erent doses (Table 2) In thestudy by Ring et al (2004), only foregut sampleswere examined as this portion of the gut seems to be

a more likely infection route for pathogenic bacteria

in Atlantic salmon Furthermore, the potentially tective role of C divergens against pathogen-induceddamage has been evaluated by simultaneously ex-posing intestinal mucosa to one of the pathogenic

pro-Table 2 Experimental treatments applied to Atlantic mon (Salmo salar L.) intestine during in vitro exposure to var- ious bacterial strains (colony-forming units, CFU)

sal-Treatment number Bacterial strain and dose (CFU mL 1)

After Ring et al (2007).

The estimate bacterial exposure of the foregut was measured by plate counts Stock solution was diluted in sterile 0.9% saline, and 0.1mL volumes of appropriate dilutions were spread on the surface of brainheart infusion agar (Merck, Darmstadt, Ger- many) (BHI) (Aeromonas salmonicida) and tryptic soy agar (TSA;

TSA plates (V ibrio anguillarum and Carnobacterium divergens).

strains and to the probiotic strain Both pathogens

used in the study of Ring, Salinas, et al (2007) have

been shown previously to enter ¢sh through the GI

tract barrier (O’Toole, Hofsten, Rosqvist, Olsson &

Wolf-Watz 2004; Ring et al 2004)

As shown in Table 3, clear di¡erences were

ob-served between the pathogenic (A salmonicida and

V anguillarum) and the non-pathogenic (C divergens)

strains Treatment of the foregut with pathogens, at

both assayed concentrations, results in various

da-maging e¡ects: epithelial cells with altered microvilli

(Fig 2), damaged tight junctions, protruding

epithe-lial cells sloughing into the lumen and the presence

of cell debris in the gut lumen Similar ¢ndings were

not observed following exposure to C divergens The

di¡erent morphological changes caused by A

salmo-nicida orV anguillarum co-incubated with C divergens

also suggest di¡erent invasive and pathogenetic

me-chanisms between the two pathogenic bacterial

strains The foregut exposed to C divergens was

histo-logically similar to control samples showing an

in-tact epithelial barrier This observation is in

agreement with the results of Ring et al (2004),

who suggested that the commensal microbiota do

not a¡ect gut cellular integrity

Surprisingly, exposure to V anguillarum at

3 104

mL 1results in the presence of

phagolyso-some-like vesicles containing degraded bacteria in

the cytoplasm of enterocytes An chemical study with anti-V anguillarum antibodymight support the nature of these degraded bacteria,together with the observation of bacteria-like struc-tures seen close to the tight junctions between enter-ocytes treated with V anguillarum and C divergens.Whether these particles are the pathogenic or pro-biotic bacteria remains unclear, and identi¢cation ofthe unknown bacteria-like particles by immunogoldlabelling techniques, green £uorescence protein and

immunohisto-in situ £uorescence hybridization might elucidatesome of these questions and will be the subject offurther work The ¢nding of phagolysosome-likestructures suggests that the enterocytes process thebacteria and act as phagocytic cells or as antigen-presenting cells in the same way as M-cells in highervertebrates

The histological changes described above werepresent in most samples, and so C divergens seems to

be unable to completely prevent tissue damage by

A salmonicida in Atlantic salmon foregut when thepathogen was present at the highest concentration

In the case of V anguillarum at the lowest tion (3 104CFU mL 1), reduced tissue damage wasobserved when Carnobacterium was present at both ahigh and a low concentration, indicating a possibleprotective e¡ect of C divergens However, this obser-vation might also be attributable to the low concen-tration of the pathogen in groups in which theintestine was exposed to both the pathogen and theprobiont, compared with foregut exposed only to thepathogen To clarify this hypothesis, additional stu-dies are necessary

concentra-Table 3 Morphological description of Atlantic salmon

foregut treated with various types of bacteria (1^11; see

After Ring et al (2007).

Damage and tissue changes were assessed as follows: 0 5 not

observed, 1 5low frequency, 2 5 moderate frequency, 3 5 high

E Ring and R Myklebust (unpubl obs.)

The partially reduced tissue damage observed

when the foregut was exposed to C divergens in

com-bination withV anguillarum at the lowest population

level may be explained by C divergens inhibiting the

growth of the pathogens in vitro (Ring et al 2005;

Ring 2008) and, thereby, colonization However,

the probiotic bacteria do not seem to reduce the

tis-sue-damaging e¡ect when the gut was exposed to a

high concentration of V anguillarum The reason for

this has not been elucidated, although it may be

attri-butable to the pathogen out-competing the probiotic

bacteria

When C divergens was present at the highest dose,

and the pathogen at the lowest dose, less damage was

seen (Table 4) To some extent, this observation might

be due to the antagonistic activity of the probiotic

bacteria against the pathogens, resulting in fewer

live pathogens being available to colonize the foregut

Based on this hypothesis, more information on the

administration of appropriate levels of probiotic

bac-teria is needed

Lactobacillus delbrueckii ssp lactis vs

A salmonicida in Atlantic salmon foregut

It is generally accepted that Lactobacillus delbrueckii is

a heterogeneous group of bacteria that includes

three subspecies: ssp delbrueckii, ssp bulgaricus and

ssp lactis (Weiss, Schillinger & Kansdler 1983)

Jacob-sen, Rosenfelt NielJacob-sen, Hayford, Mller, MichaelJacob-sen,

Paerregaard, Sandstr˛m, Tvede and Jakobsen (1999)

suggested that L delbrueckii ssp lactis is a good

pro-biotic candidate according to its bacteriological

prop-erties in vitro From an aquaculture point of view,

L delbrueckii ssp lactis has been administered in vivo

in the diet to gilthead seabream (Sparus aurata L.) andincreased the cellular innate immune responses (Sal-inas, Cuesta, Esteban & Meseguer 2005) The fate ofprobiotic bacteria in the GI tract of ¢sh is less wellknown, including the histological changes thatmight take place in the gut following incubation withLAB like L delbrueckii ssp lactis

The aims of the work of Salinas et al (2008) were,

by means of microscopy techniques, to (i) determinewhether an L delbrueckii ssp lactis (CECT 287,Valen-cia, Spain) of non-aquaculture origin was capable ofcolonizing the Atlantic salmon GI tract during in vitroincubation, (ii) describe the morphological changesand cellular responses occurring in the intestinalepithelium after in vitro exposure to this probioticstrain and (iii) study the possible protective role of

L delbrueckii ssp lactis against tissue-damaging fects caused byA salmonicida in the foregut of Atlan-tic salmon This information is highly relevant as it iswell known that probiotic bacteria increase the hosthealth status and protect mucosal tissue againstpathogen-caused damage in mammalian models(Fung, Woo, Wan-Abdullah, Ahmad, Easa & Liong2009) Using confocal microscopy, their ¢ndingsshowed that, following short-term in vitro incubation

ef-of Atlantic salmon foregut with mine isothiocyanate (TRITC)-labelled L delbrueckiissp lactis, the probiont was able to colonize the enter-ocyte surface.When L delbrueckii ssp lactis were ob-served in the lumen, the bacteria were found ingroups or clumps (Fig 3a) Moreover, labelled bacter-

tetramethylrhoda-ia were also found at the villus surface, inside the cosal epithelium or even in the lamina propria (Fig.3b) As the intestines were thoroughly washed threetimes before samples were taken, only those bacteriaable to adhere to the mucus or to the epithelial cellswere present Moreover, foregut exposed to the pro-biotic bacteria only resulted in a healthy intestinalbarrier whereas exposure to A salmonicida disruptedits integrity However, pre-treatment of salmon intes-tine with L delbrueckii ssp lactis prevented Aeromo-nas damaging e¡ects (Table 5) These results arepromising in the context of the use of non-auto-chthonous probiotic bacteria as prophylactic agentsagainst ¢sh bacterial infections in the GI tract.Although Nikoskelainen, Salminen, Bylund andOuwehand (2001) studied the adhesion capacity ofdi¡erent LAB strains to rainbow trout (Oncorhynchusmykiss Walbaum) mucus in vitro, probiotic bacteriahad never been shown to cross the ¢sh gut before

mu-Table 4 Outcome of comparison of in vitro e¡ects of

experi-mental treatments on Atlantic salmon intestine (seeTable 2)

ranked by order of severity from 1 5control (Ringer) to

11 5V ibrio anguillarum at 6  10 6 (Page’s L test, P o0.01)

Ranking Bacterial strain and dose (CFU mL 1)

Whereas some authors (Fuller 1973; Myr-Mkinen,

Manninen & Gyllenberg 1983) have postulated that

probiotic bacteria are host speci¢c (and their e¡ects

are limited to their natural hosts), three studies have

reported the absence of speci¢city in LAB when

bind-ing host intestinal mucus (Gildberg & Mikkelsen

1998; Rinkinen, Westermarck, Salminen &

Ouwe-hand 2003; Salinas et al 2008) It was also evident

that the LAB species used by Salinas et al (2008)

ex-erted a pronounced local e¡ect on the GI tract lining.The authors’ qualitative ¢nding, that the concentra-tion of immune cells in the gut epithelium increasedfollowing incubation with the probiotic strain, re-sembled the response that occurred when the intes-tine was exposed to A salmonicida (Ring, Salinas,

et al 2007) This is consistent with a recent study thatfound increased numbers of acidophilic granulocytes

in the gut of seabream (S aurata L.) larvae after a biotic, a mixture of Lactobacillus fructivorans and Lac-tobacillus plantarum, was delivered through the diet(Picchietti, Mazzini, Taddei, Renna, Fausto, Mulero,Carnevali, Cresci & Abelli 2007)

pro-Some information is available demonstrating that

A salmonicida causes cell damage in the foregut ofAtlantic salmon (Ring et al 2004; Ring, Salinas,

et al 2007; Salinas et al 2008), but these studies usedthe pathogen without washing the culture superna-tant The e¡ects observed in these studies might bedue either to toxins, to the bacteria themselves or toboth toxins and bacterial cells However, as a block-ing e¡ect was observed when the intestine was ex-posed to the lactobacilli (Salinas et al 2008), it can

be speculated that the lactobacilli compete with thepathogen in adherence to the mucus layer, by produ-cing bacteriocins, organic acid, H2O2, etc., whichwould inhibit pathogenic colonization The hypoth-esis proposed by Salinas et al (2008) is in accord withthe results of Chabrillon, Ouwehand, Diaz-Rosales,Arijo, Martinez-Manzanares, Balebona and Morinigo(2006), who demonstrated that attachment of thepathogenic bacteria (V anguillarum, Photobacteriumdamselae ssp piscicida,V alginolyticus and Vibrio har-veyi) to intestinal mucus of Gilthead seabream wassigni¢cantly reduced by Lactobacillus rhamnosus and

L V

V LP L

N

40µ(b)

Figure 3 Confocal microscopy of the fate of

tetra-methylrhodamine isothiocyanate-labelled Lactobacillus

delbrueckii ssp lactis in the foregut of Atlantic salmon (a)

Fluorescent image of Atlantic salmon foregut cryosection

after incubation for 30 min with L debrˇeckii subsp lactis

[107 colony-forming units (CFU) mL 1] labelled with

TRITC L, lumen; M, mucus Note the red £uorescence of

labelled bacteria associated with mucus in the gut lumen

Scale bar 5 20mm (b) Fluorescent image of Atlantic

sal-mon foregut cryosection after incubation for 30 min with

L debrˇeckii subsp lactis (107CFU mL 1) labelled with

TRITC Note the red £uorescence of labelled bacteria

lo-cated inside the villi (V) at the level of the lamina propria

(LP) Scale bar 5 40mm After Salinas et al (2008) L,

lu-men; N, nuclei of enterocytes

Table 5 Morphological changes undergone byAtlantic salmon foregut following exposure to Aeromonas salmonicida ssp monicida, Lactobacillus delbrˇeckii ssp lactis or both bacterial strains

sal-Morphological observations Control

Aeromonas only

Lactobacillus only

Lactobacillus1 Aeromonas

Tissue changes were assessed as follows: (0) not observed; (1) low frequency; (2) moderate frequency; (3) high frequency.

After Salinas et al (2008).

Bi¢dobacterium lactis intended for human use These

results, together with those of Vine, Leukes, Kaiser,

Baya and Baxter (2004) working on the topic of

at-tachment competition of probiotics and pathogenic

bacteria to ¢sh intestinal mucus, should stimulate

bacteriologists to carry out further studies on this

topic

In spotted wolf ¢sh (A minor Olafsen) fry exposed

toV anguillarum, paracellular bacterial translocation

seems to occur as a secondary e¡ect of detachment

and loss of enterocytes that create large intercellular

spaces (Ring et al 2006) In contrast to these results,

Salinas et al (2008) reported paracellular bacteria

be-tween the enterocytes when the foregut was exposed

to L delbrueckii ssp lactis This translocation

oc-curred in virtually intact gut epithelia, which means

that translocation of the probiotic bacteria was not

associated with the creation of oedematous spaces, a

characteristic of pathogenic bacteria

Carnobacteria vs.Vibiro (Listonella)

anguillarum in the GI tract of Atlantic cod

Atlantic cod was recently introduced into

aquacul-ture and is common at the ¢shmonger and popular

across Northern Europe (Bowden, Adamson &

Brick-nell 2003) There are many reports on diseases in

Atlantic cod, on wild stocks and in aquaculture (for

a review, see Egidius 1987; Toranzo, Mamarinos &

Romalde 2005; Bricknell et al 2006; Samuelsen,

Ner-land, Jrgensen, Schrder, Svsand & Bergh 2006)

The major pathogenicVibrio of Atlantic cod in culture

is V anguillarum serotype 02b, and there is evidence

that current salmonid vaccines, which contain

sero-types 01 and 02a and V ordalii, are not e¡ective in

protecting against infection with the 02b serotypes

(J Bgvald, pers comm.) As culture of Atlantic cod

has become more common, the 02b serotype appears

to have become the dominant isolate (Santos, Pazos,

Bandin & Toranzo 1995)

Based on the available information that V

anguil-larum serotype 02b is an important bacterial

patho-gen in gadoid culture, Lvmo (2007a, b) conducted

studies using the intestinal sac method to evaluate

the potential protective role of carnobacteria against

V anguillarum serotype 02b in the GI tract of Atlantic

cod The GI tract was divided into six di¡erent parts:

stomach, pyloric caeca, foregut, midgut and hindgut

(distal intestine) and the hindgut chamber

(fermen-tation chamber) Lvmo (2007a) used the same

car-nobacteria strain previously used by Ring, Salinas,

et al (2007) and exposed the foregut to 2 106

V

an-guillarum mL 1or 6.4 106

C divergens mL 1 Theaims of this in vitro study were to investigate whetherexposure of the intestinal mucosa of the Atlantic codtoV anguillarum a¡ects the morphology of the intest-inal epithelium in the foregut, and to compare theseresults with those observed by exposing intestines to

a probiotic bacterium and another segment not posed to bacteria As shown in Table 6, some di¡er-ences were apparent when the foregut was exposed

ex-to bacteria compared with the non-exposed controlgroup All three experimental groups showed a nor-mal-looking mucosa with an intact epithelium, con-sisting of undamaged enterocytes with numerousnormal-looking microvilli However, the foregut ex-posed to bacteria showed paler epithelial cell nuclei,disorganized microvilli and budding from the apices

of microvilli Goblet cells were normally ¢lled whenthe intestine was exposed to bacteria, both C diver-gens and V anguillarum, while the goblet cells wereempty when the intestine was exposed to the Ringersolution Based on the results that the foregut ex-posed to V anguillarum was histologically similar tothe foregut exposed to C divergens, it was suggestedthat the foregut is not a major infection site for V an-guillarum The result of Carnobacterium exposure is ofhigh importance as translocation and cell damagehave been proposed as important criteria when eval-uating the use of probiotics in endothermic animals

as well as in ¢sh (Salminen, Deighton, Benno & bach 1998; Ring, Myklebust, et al 2007)

Gor-Table 6 Morphological description of Atlantic cod (Gadus morhua L.) foregut exposed to no bacteria (control),Vibrio an- guillarum and Carnobacterium divergens

Morphological observations Control V anguillarum C divergens

Intraepithelial lymphocytes

Loosening of enterocytes from basal membrane

ob-In a later study, Lvmo (2007b) investigated the

ef-fect of bacterial exposure in both the hindgut and the

hindgut chamber Here, only results obtained from

the hindgut chamber are presented, as this part of

the digestive tract seems to be more important, at

least from electron microscopic and bacteriological

perspectives A detailed description of the di¡erent

experimental groups exposed to bacteria in the

hind-gut chamber is shown in Table 7 An adherent

Carno-bacterium originally isolated from the hindgut

chamber of Atlantic cod (Seppola, Olsen, Sandaker,

Kanapathippillai, Holzapfel & Ring 2006) and V

an-guillarum serotype 02b was used Previously,

Carno-bacterium has been shown to inhibit di¡erent

bacteria (Listeria monocytogenes, Enterococcus faecalis

and Staphylococcus aureus) in vitro (M Seppola and

E Ring unpubl obs.), and the carnobacteria were

identi¢ed by 16S rRNA to Carnobacterium

maltaroma-ticum (Lvmo 2007b)

Log total viable counts detected in the hindgut

chamber of treatment groups 1, 4 and 5 yielded log

values of 4.6, 4.6 and 4.8 bacteria g 1respectively

These results showed that the total bacterial

popula-tion level was not a¡ected by bacterial exposure A

total of 42, 44 and 39 isolates were identi¢ed by 16S

rRNA from experimental groups 1, 4 and 5,

respec-tively, and these results demonstrated that the

auto-chthonous bacterial communities were a¡ected (Fig

4a^c) The dominant bacterial species identi¢ed in

the hindgut chamber of the control ¢sh (not exposed

to bacteria) wasVibrio wodanis (42%), followed

byVi-brio logei (25%) and VibyVi-brio ¢scheri (24%) (Fig 4a)

Photobacterium phosphoreum and Staphylococcus dermis were also identi¢ed.When the hindgut cham-ber was ¢rst exposed to C maltaromaticum, washedthree times and then exposed toV anguillarum,V wo-danis was still the dominant autochthonous speciesand accounted for 44% of the identi¢ed strains,while V anguillarum and C maltaromaticum ac-counted for 22% and 20% respectively (Fig 4b) Incomparison, when the hindgut chamber was ¢rst ex-posed toV anguillarum, washed three times and thenexposed to C maltaromaticum for 30 min, C maltaro-maticum accounted for 54% of the autochthonousspecies but V wodanis was still a dominant auto-chthonous species and accounted for 35% of theidenti¢ed strains In this experimental group,V angu-illarum was not identi¢ed (Fig 4c)

epi-Based on the results of Lvmo (2007b), it can beconcluded that V wodanis belongs to the autochtho-nous microbiota in the hindgut chamber of Atlanticcod and that its population level is not a¡ected by ex-posing the hindgut chamber to bacteria such as

C maltaromaticum and V anguillarum As C maticum is present in the hindgut chamber when thechamber was ¢rst exposed to C maltaromaticum andthereafter toV anguillarum, or ¢rst exposed toV angu-illarum and thereafter to C maltaromaticum, this in-dicates that the carnobacteria are able to colonize thehindgut chamber and out-compete V anguillarum.This is clearly demonstrated in experimental group

maltaro-5, where Lvmo (2007b) was not able to isolate V guillarum from the hindgut chamber post exposure

an-to the pathogen and thereafter an-to C maltaromaticum

Table 7 Experimental treatments applied to Atlantic cod hindgut chamber during in vitro exposure to various bacterial strains

Treatment Bacterial strain and dose

Exposure time (min) Washed 

Exposure time (min) Washed 

Before sampling the hindgut chamber was washed times with 3 mL sterile saline.

wThe hindgut chamber was ¢rst exposed to Carnobacterium maltaromaticum (30 min), washed (three times with 3 mL sterile saline) and thereafter exposed to V anguillarum for 30 min.

zThe hindgut chamber was ¢rst exposed to Vibrio anguillarum (30 min), washed (three times with 3 mL sterile saline) and thereafter exposed to C maltaromaticum for 30 min.

The estimate bacterial exposure of the hindgut chamber was measured by plate counts Stock solution was diluted in sterile 0.9% saline, and 0.1mL volumes of appropriate dilutions were spread on the surface of and tryptic soy agar plates.

Moreover, the potential of this carnobacteria is also

illustrated by the fact the strain did not a¡ect

mor-phology in the hindgut of Atlantic salmon However,

as exposure of the hindgut chamber of Atlantic cod by

V anguillarum did not cause cell damage, it is

con-cluded that this part of the digestive tract is not

in-volved as an infection route Based on the results of

Ring et al (2004), Ring, Salinas, et al (2007), Ring,

Myklebust, et al (2007) and Salinas et al (2008) on

Atlantic salmon and the results of Lvmo (2007a, b)

on Atlantic cod, we conclude that the intestinal tion route is di¡erent in these two species

infec-Future perspectives

In the studies of Ring, Myklebust, et al (2007) andSalinas et al (2008), light and electron microscopywere used Lvmo (2007a, b) also used light and elec-tron microscopy, and isolation of culturable bacteria,

to evaluate cell damage and the ability to adhere tothe di¡erent parts of the digestive tract As the GItract in some ¢sh species might be an important in-fection route (Gro¡ & LaPatra 2000; Birkbeck &Ring 2005; Harikrishnan & Balasundaram 2005),future studies on this subject have to include charac-terization using modern molecular biological DNA-based methods In these studies, both microbesattached to the epithelium and microbes associatedwith intestinal digesta have to be assessed Samplesshould be subjected to isolation of bacteria from thedigesta and mucus material, lysing of the isolatedbacteria and DNA extraction using mechanical, che-mical and enzymatic breakdown as described by Hol-ben, Srkilahti, Williams, Saarinen and Apajalahti(2002) In the exploratory phase, the microbial com-munity should be assessed by amplifying the DNAwith universal primers for sequencing of the taxono-mically important16S rRNA gene Once the composi-tion of the microbial community is covered, and themost important microbiological clusters in relation tothe nutritional and health status of the salmonare discovered, rapid tools based on quantitativereal-time PCR can be developed to routinely monitorthe important indicator organisms However, thethreshold for bacterial cDNA may be a limiting fac-tor in gut samples from ¢sh, except in acute-phaseinfection In addition to DNA-based techniques, im-portant gut bacteria isolated by cultivation should

be investigated for their metabolic capabilities.Furthermore, the metabolic pro¢les in the digestasamples should be studied using gas and liquidchromatography

Future studies should also include di¡erent ing methods such as immunogold, green £uorescentprotein (GFP) and quorum sensing (QS) The intro-duction of GFP as an endogenous £uorescent tag pro-vides a means of rendering the bacteria visible andtracing their activity in living host cells (Valdivia,Hromockyj, Monack, Ramakrishnan & Falkow 1996;Lun & Willson 2004; Yin, Zhou, Li, Li, Hou & Zhang2007) Green £uorescent protein is a small protein

stain-Fermentation chamber Control fish

Fermentation chamber Fish first exposed to

C maltaromaticum and thereafter to

Vibrio anguillarum Vibrio fischeri Carnobacterium maltaromaticum

Fermentation chamber Fish first exposed to

V anguillarum and thereafter to

Figure 4 Relative abundance of adherent microbiota in

the hindgut chamber of Atlantic cod exposed to (a) sterile

physiological salt water, (b) Carnobacterium

maltaromati-cum and Vibrio anguillarum and (c) V anguillarum and

C maltaromaticum A detailed description of the di¡erent

treatment is given in Table 7 After Lvmo (2007b)

(27 kDa) found in the jelly¢sh, Aequorea victoria It

£uoresces when excited by ultraviolet (UV) light

(Chal¢e, Tu, Euskirchen, Ward & Prasher 1994)

Bac-teria tagged with the GFP gene can be easily

identi-¢ed as green £uorescing colonies under UV light

The GFP £uorescence is independent of cofactors or

additional gene products, is sensitive, stable, speci¢c,

non-toxic and does not interfere with cell growth and

function (Chal¢e et al 1994; Valdivia et al 1996)

Green £uorescent protein-marked ¢sh pathogens

have been constructed to study the invasion

path-ways in vivo and in vitro (Ling, Xie, Lim & Leung

2000; O’Toole et al 2004) Based on this information,

we recommend using GFP as a biomarker to illustrate

the infection kinetics and tissue localization of both

probiotic and pathogenic bacteria in future in vitro

studies

In recent years, the ability of bacteria to

com-municate with one another using chemical signal

molecules has received considerable attention in

Gram-negative ¢sh pathogenic bacteria (Defoirdt,

Boon, Bossier & Verstraete 2004; Bruhn, Dalsgaard,

Nielsen, Buchholtz, Larsen & Gram 2005; Buchholtz,

Nielsen, Milton, Larsen & Gram 2006; Defoirdt, Boon,

Sorgeloos, Verstraete & Bossier 2008) The ability to

send, receive and process information allows

unicel-lular organisms to act as multicelunicel-lular entities and

increases their chances of survival in complex

envir-onments Quorum sensing is commonly associated

with adverse health e¡ects such as bio¢lm formation,

bacteria pathogenicity and virulence, and it may

pro-vide a way of controlling infection in aquaculture

(Defoirdt et al 2004, 2008) Recently, it was

demon-strated that probiotics (Lactobacillus acidophilus La-5)

could a¡ect virulence-related gene expression in

Escherichia coli O157:H7 by secreting a molecule(s)

that either acts as a QS inhibitor or directly interacts

with genes involved in colonization (Medellina-Pena,

Wang, Johnson, Anand & Gri⁄ths 2007) As this

in-formation is not available in ¢sh, we recommend that

such studies are carried out

It is generally accepted that probiotics block

patho-genic bacterial e¡ects by producing bactericidal

sub-stances and competing with pathogens and toxins

for adherence to the intestinal epithelium They also

regulate immune responses by enhancing innate

im-munity and modulating pathogen-induced

in£am-mation via Toll-like receptor-regulated signalling

pathways and regulate intestinal epithelial

homeos-tasis by promoting intestinal epithelial cell survival,

enhancing barrier function and stimulating

protec-tive responses (Vanderpool, Yan & Polk 2008) As no

information is available in ¢sh, these topics should

be given high priority in future studies

A fundamental question arises when discussingthe protective role of the GI tract microbiota: Can thegradual development of the GI tract from larva toadult a¡ect infection and are there di¡erent infectionpatterns between di¡erent ¢sh species? Again, as noinformation is available on these topics, scientistsmight pro¢tably investigate the interactions betweenLAB and pathogens in the digestive tract of ¢sh

ReferencesAndlid T., VaŁzquez-JuaŁrez R.V & Gustafsson L (1995) Yeast colonizing the intestine of rainbow trout (Salmo gairdneri) and turbot (Scophthalmus maximus) Microbial Ecology 30, 321^334.

Austin B & Austin D.A (1999) Bacterial Fish Pathogens: eases of Farmed and Wild Fish, 3rd edn Springer ^ praxis, London, 457pp.

Dis-Baccazar 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.

Bakken Y (2002) Histological studies of pyloric caeca of tic salmon (Salmo salar L.) fed diets containing linseed, soy- bean, and marine oils E¡ects of challenge with Aeromonas salmonicida ssp salmonicida MSc thesis, Norwegian Col- lege of Fishery Science, University of Troms, Norway Balcazar J.L (2003) Evaluation of probiotics bacterial strains in Litopenaeus vannamei Final report, National Center for Marine and Aquaculture Research, Guayaquil, Ecuador BalcaŁzar J.L., de Blas I., Ruiz-Zarzuela I.,Vendrell D & Muz- quiz J.L (2004) Probiotics: a tool for the future of ¢sh and shell¢sh health management Journal of Aquaculture in the Tropics 19, 239^242.

Atlan-Balcazar J.L., de Blas I., Ruiz-Zarzuela I., Cunningham D., Vendrell D & Muzquiz J.L (2006) The role of probiotics in aquaculture.Veterinary Microbiology 114, 173^186 Balcazar J.L., Decamp O., Vendrell D., de Blas I & Ruiz-Zar- zuela I (2006) Health and nutritional properties of pro- biotics in ¢sh and shell¢sh Microbial Ecology in Health and Diseases 18, 65^70.

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 probiotics administration in brown trout (Salmo trutta) British Journal of Nutrition 97, 522^ 527.

Birkbeck T.H & Ring E (2005) Pathogenesis and the trointestinal tract of growing ¢sh In: Microbial Ecology

gas-in Growgas-ing Animals (ed by W Holzapfel & P Naughton),

pp 208^234 Elsevier, Edinburgh, UK.

Bowden T.J., Adamson K & Bricknell I.R (2003) Diseases of

relevance to haddock In: Early Rearing of Haddock State of

the Art (Special Publication), Vol 7 (ed by D.E Aitkin), pp.

113^114 Aquaculture Association of Canada, Halifax,

Canada.

Bricknell I.R., Bron J.E & Bowden T.J (2006) Diseases of

ga-doid ¢sh in cultivation: a review ICES Journal of Marine

Science 63, 253^266.

Bruhn J.B., Dalsgaard I., Nielsen K.F., Buchholtz C., Larsen J.L.

& Gram L (2005) Quorum sensing signal molecules

(acy-lated homoserine lactones) in Gram-negative ¢sh

patho-genic bacteria Diseases of Aquatic Organisms 65, 43^52.

Buchholtz C., Nielsen K.F., Milton D., Larsen J.L & Gram L.

(2006) Pro¢ling of acylated homoserine lactones of Vibrio

anguillarum in vitro and in vivo: in£uence of growth

con-ditions and serotype Systematic and Applied Microbiology

29, 433^445.

Burr G., Gatlin D & Ricke S (2005) Microbial ecology of the

gastrointestinal tract of ¢sh and the potential application

of prebiotics and probiotics in ¢n¢sh aquaculture Journal

of World Aquaculture Society 36, 425^436.

Cahill M.M (1990) Bacterial £ora of ¢shes: a review

Micro-bial Ecology 19, 21^41.

Chabrillon M., Ouwehand A.C., Diaz-Rosales P., Arijo S.,

Martinez-Manzanares E., Balebona M.C & Morinigo

M.A (2006) Adhesion of lactic acid bacteria to mucus of

farmed gilthead seabream, and interactions with ¢sh

pathogenic microorganisms Bulletin of European

Associa-tion of Fish Pathology 26, 202^210.

Chabrillon M., Arijo S., Diaz-Rosales P & Balebona M.C.

(2006) Interference of Listonella anguillarum with

poten-tial probiotics microorganisms isolated from farmed

gilt-head seabream (Sparus aurata, L.) Aquaculture Research

37, 78–86.

Chal¢e M., Tu Y., Euskirchen G., Ward W.W & Prasher D.C.

(1994) Green £uorescent protein as a marker for gene

ex-pression Science 263, 802^805.

Chopra A.K., Xu X.J., Ribardo D., Gonzalez M., Kuhl K.,

Pe-terson J.W & Houston C.W (2000) The cytotoxic

entero-toxin of Aeromonas hydrophila induces proin£ammatory

cytokine production and activates arachidonic acid

meta-bolism in macrophages Infection and Immunology 68,

2808^2818.

Defoirdt T., Boon N., Bossier P & Verstraete W (2004)

Disrup-tion of bacterial quorum sensing: an unexplored strategy

to ¢ght infection in aquaculture Aquaculture 240,69^88.

Defoirdt T., Boon N., Sorgeloos P., Verstraete W & Bossier P.

(2008) Quorum sensing and quorum quenching inVibrio

harveyi: lessons learned from in vivo work Mini review.

The ISME Journal 2, 19^26.

Direkbusarakom S., Yoshimizu M., Ezura Y., Ruangpan L &

Danayadol Y (1998) Vibrio spp the dominant £ora in

shrimp hatchery against some ¢sh pathogenic viruses.

Journal of Marine Biotechnology 6, 266^267.

Egidius E (1987) Vibriosis: pathogenicity and pathology A

review Aquaculture 67, 15^28.

Farzanfar A (2006) The use of probiotics in shrimp ture FEMS Immunology and Medical Microbiology 48, 149^158.

aquacul-Fasano A (2002) Toxins and the gut: role in human disease Gut 50, 9^14.

Fivaz M & van der Goot F.G (1999) The tip of a molecular syringe Trends in Microbiology 7, 341^343.

Fuller R (1973) Ecological studies of lactobacillus £ora ciated with the crop epithelium of the fowl Journal of Ap- plied Bacteriology 36, 131^139.

asso-Fung W.Y., Woo Y.P, Wan-Abdullah W.N., Ahmad R., Easa A.M & Liong M.T (2009) Bene¢ts of probiotics: beyond gastrointestinal health Milchwissenschaft Milk Science In- ternational 64, 17^21.

Gatesoupe F.J (1999) The use of probiotics in aquaculture Aquaculture 180, 147^165.

Gatesoupe F.J (2007) Live yeast in the gut: natural rence, dietary introduction, and their e¡ects on ¢sh health and development Aquaculture 267, 20^30 Gatesoupe F.J (2008) Updating the importance of lactic acid bacteria in ¢sh farming: natural occurrence and probio- tics treatments Journal of Molecular Microbiology and Bio- technology 14, 107^114.

occur-Gildberg A & Mikkelsen H (1998) E¡ects of supplementing the feed to Atlantic cod (Gadus morhua) fry with lactic acid bacteria and immuno-stimulating peptides during a chal- lenge trial with Vibrio anguillarum Aquaculture 167, 103^ 113.

Girones R., Jofre J.T & Bosch A (1989) Isolation of marine bacteria with antiviral properties Canadian Journal of Mi- crobiology 35, 1015^1021.

Gomez G.D & Balcazar J.L (2008) A review on the tions between gut microbiota and innate immunity of

interac-¢sh FEMS Immunology and Medical Microbiology 52, 145^154.

Gomez-Gil B., Roque A & Turnbull J.F (2000) The use and selection of probiotic bacteria for use in the culture of lar- val aquatic organisms Aquaculture 191, 259^270 Gram L & Ring E (2005) Prospects of ¢sh probiotics In: Microbial Ecology in Growing Animals (ed by W Holzapfel

& P Naughton), pp 379^417 Elsevier, Edinburgh, UK Gri⁄th S., Melville K., Cook M., Vincent S., St Pierre M & Lanteigne C (2001) Pro¢ling of bacterial species asso- ciated with haddock larviculture by PCR ampli¢cation of 16S rDNA and denaturing gradient gel electrophoresis Journal of Aquatic Animal Health 13, 355^363.

Gro¡ J & LaPatra S (2000) Infectious diseases impacting the commercial culture of salmonids Journal of Applied Aquaculture 10, 17^90.

Gullian M & Rodr|¤guez J (2002) Immunostimulant qualities

of probiotic bacteria Global Aquaculture Advocate 5, 52^ 54.

Hagi T.,Tanaka D., IwamuraY & Hoshino T (2004) Diversity and seasonal changes in lactic acid bacteria in the intest- inal tract of cultured freshwater ¢sh Aquaculture 234, 335^346.

Harikrishnan R & Balasundaram C (2005) Modern trends

in Aeromonas hydrophila disease management with ¢sh.

Reviews in Fisheries Science 13, 281^320.

HolbenW.E., Srkilahti L.K.,Williams P., Saarinen M &

Apa-jalahti J.H.A (2002) Phylogenetic analysis of intestinal

micro£ora indicates a novel Mycoplasma phylotype in

farmed and wild salmon Microbial Ecology 44, 175^185.

Hong H.A., Duc L.H & Cutting S.M (2005) The use of

bac-terial spore formers as probiotics FEMS Microbiology

Re-views 29, 813–835.

Hovda M.B., Lunestad B.T., Fontanillas R & Rosnes J.T.

(2007) Molecular characterisation of the intestinal

micro-biota of farmed Atlantic salmon (Salmo salar L.)

Aquacul-ture 272, 581^588.

Irianto A & Austin B (2002a) Probiotics in aquaculture.

Journal of Fish Diseases 25, 633^642.

Irianto A & Austin B (2002b) Use of probiotics to control

furunculosis in rainbow trout, Oncorhynchus mykiss

(Walbaum) Journal of Fish Diseases 25, 333^342.

Irianto A., Roberwen P.A.W., Austin B & Pandalai S.G.

(2000) The use of probiotics in aquaculture Recent

Re-search Developments in Microbiology 4, 557^567.

Jacobsen C.N., Rosenfelt NielsenV., Hayford A.E., Mller P.L.,

Michaelsen K.F., Paerregaard A., Sandstr˛m B., Tvede M.

& Jakobsen M (1999) Screening of probiotic activities of

forty-seven strains of Lactobacillus spp by in vitro

techni-ques and evaluation of the colonization ability of ¢ve

se-lected strains in humans Applied and Environmental

Microbiology 65, 4949^4956.

Jensen S., vres L., Bergh  & Torsvik V (2004)

Phyloge-netic analysis of bacterial communities associated with

larvae of the Atlantic halibut propose succession from a

uniform normal £ora Systematic and Applied Microbiology

27,728^736.

Jutfelt F., Olsen R.E., Glette J., Ring E & Sundell K (2006)

Translocation of viable Aeromonas salmonicida across the

intestine of rainbow trout Journal of Fish Diseases 29,

255^262.

Kamei Y.,Yoshimizu M., Ezura Y & Kimura T (1988)

Screen-ing of bacteria with antiviral activity from fresh water

salmo-nid hatcheries Microbiology and Immunology 32, 67^73.

Karunasagar I & Karunasagar I (1999) Diagnosis,

treat-ment and prevention of microbial diseases of ¢sh and

shell¢sh Current Science 76, 387^399.

Kesarcodi-Watson A., Kaspar H., Lategan M.J & Gibson L.

(2008) Probiotics in aquaculture The need, principles

and mechanisms of action and screening processes.

Aquaculture 274, 1^14.

Kim D.H., Brunt J & Austin B (2007) Microbial diversity of

intestinal contents and mucus in rainbow trout

(Oncor-hynchus mykiss) Journal of Applied Microbiology 102,

1654^1664.

Knudsen G., Srum H., McLPress C & Olafsen J.A (1999) In

situ adherence of Vibrio spp to cryosections of Atlantic

salmon, Salmo salar L., tissue Journal of Fish Diseases 22,

409^418.

Krogdahl —., Bakke-Mckellep A.M., Rd K.H & Bverfjord

G (2000) Feeding Atlantic salmon, Salmo salar L., bean products: e¡ect on disease resistance (furunculosis), and lysozyme and IgM levels in the intestinal mucosa Aquaculture Nutrition 6,77^84.

soy-Ldemel J.B., Mayhew T.M., Myklebust R., Olsen R.E., Espelid

S & Ring E (2001) E¡ect of three dietary oils on disease susceptibility in Arctic charr (Salvelinus alpinus L.) during cohabitant challenge with Aeromonas salmonicida ssp sal- monicida Aquaculture Research 32, 935^945.

Lvmo L (2007a) In vitro eksponering av torsketarm giske forandringer som flge av Carnobacterium divergens og Vibrio anguilllarum Project thesis (in Norwegian), Univer- sity of Troms, Norway, p 55.

Lvmo L (2007b) In vitro eksponering av torsketarm giske og bakteriologiske forandringer som flge av Carnobac- terium maltaromaticum og Vibrio anguilllarum Master thesis (in Norwegian), p 56.

Histolo-Ling S.H.M., Xie L., Lim T.M & Leung K.Y (2000) Use of green £uorescent protein (GFP) to track the invasion pathways of Edwardsiella tarda in in vivo and in vitro ¢sh models Microbiology 146,7^19.

Liu Y., Zhou Z.,Yao B., Shi P., He S., Benjamisen Hlvold L & Ring E (2008) E¡ect of intraperitoneal injection of im- munostimulatory substances on allochthonous gut mi- crobiota of Atlantic salmon (Salmo salar L.) determined using denaturing gradient gel electrophoresis Aquacul- ture Research 39, 635^646.

Lun S & Willson P.J (2004) Expression of green £uorescent protein and its application in pathogenesis studies of ser- otype 2 Streptococcus suis Journal of Microbiological Meth- ods 56, 401^412.

Mackie T.J., Arkwright J.A., Pyrce-Tannatt T.E., Mottram J.C., Johnston W.D & Menzies W.J (1930) Interim Report of the Furunculosis Committee HMSO, Edinburgh, UK Madigan M.T., Martinko J.M & Parker J (2000) Brock Biol- ogy of Microorganisms, 9th edn Prentice Hall Interna- tional, Upper Saddle River, NJ, USA.

Mayhew T.M., Myklebust R., Whybrow A & Jenkins R (1999) Epithelial integrity, cell death and cell loss in mam- malian small intestine Histology and Histopathology 14, 257^267.

Myr-Mkinen A., Manninen M & Gyllenberg H (1983) The adherence of lactic acid bacteria to columnar epithe- lial cells of pigs and calves Journal of Applied Bacteriology

55, 241^245.

Medellina-Pena M.J.,Wang H., Johnson R., Anand S &

Grif-¢ths M.W (2007) Probiotics a¡ect virulence-related gene expression in Escherichia coli O157:H7 Applied and Envir- onmental Microbiology 73, 4259^4267.

Michel C., Pelletier C., Boussaha M., Douet D-G., Lautraite A.

& Tailliez P (2007) Diversity of lactic acid bacteria associated with ¢sh and the ¢sh farm environ- ment, established by ampli¢ed rRNA gene restriction analysis Applied and Environmental Microbiology 73, 2947^2955.

Moriarty D (1997) The role of microorganisms in

aquacul-ture ponds Aquaculaquacul-ture 151, 333^349.

Namba A., Mano N & Hirose H (2007) Phylogenetic

analy-sis of intestinal bacteria and their adhesive capability in

relation to the intestinal mucus of carp Journal of Applied

Microbiology 102, 1307^1317.

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

(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.

Norwegian Scienti¢c Committee for Food Safety (2009)

Cri-teria for Safe Use of Plant Ingredients in Diets for

Aquacul-tured Fish ISBN 978-82-8082-299- 4 Norwegian

Scienti¢c Committee for Food Safety, Oslo, Norway.

Olafsen J.A (2001) Interactions between ¢sh larvae and

bac-teria in marine aquaculture Aquaculture 200, 223^247.

Olivier G (1997) Getting to know your enemy In:

Furunculo-sis Multidisciplinary Fish Disease Research (ed by E.-M.

Bernoth, A.E Ellis, P.J Midtlyng, G Olivier & P Smith),

pp 233^234 Academic, San Diego, CA, USA.

Olsen R.E., Myklebust R., Ring E & Mayhew T.M (2000)

The in£uence of dietary linseed oil and saturated fatty

acids on caecal enterocytes in Arctic charr (Salvelinus

al-pinus L.): a quantitative ultrastructural study Fish

Phy-siology and Biochemistry 22, 207^216.

O’Toole R., Hofsten J.V., Rosqvist R., Olsson P.E & Wolf-Watz H.

(2004) Visualisation of zebra¢sh infection by GFP-labelled

Vibrio anguillarum Microbial Pathogenesis 37, 41^46.

Pedersen A & Dalsgaard A (2003) Antimicrobial resistance

of intestinal Aeromonas spp and Enterococcus spp in ¢sh

cultured in integrated broiler-¢sh farms in Thailand.

Aquaculture 219,71^82.

Picchietti S., Mazzini M.,Taddei A.R., Renna R., Fausto A.M.,

Mulero V., Carnevali O., Cresci A & Abelli L (2007) E¡ects

of administration of probiotic strains on GALTof larval

gilt-head seabream: immunohistochemical and

ultrastruc-tural studies Fish and Shell¢sh Immunology 22, 57^67.

Pond M.J., Stone D.M & Alderman D.J (2006) Comparison of

conventional and molecular techniques to investigate the

intestinal micro£ora of rainbow trout (Oncorhynchus

my-kiss) Aquaculture 261, 193^203.

Press C.M & Lillehaug A (1995) Vaccination in European

salmonid aquaculture: a review of practices and

pro-spects BritishVeterinary Journal 151, 45^69.

Prieur G., Nicolas J.L., Plusquellec A & Vigneulle M (1990)

Interactions between bivalves molluscs and bacteria in

the marine environment Oceanographic Marine Biology

Annual Reviews 28, 227^352.

Rengpipat S., Rukpratanporn S., Piyatiratitivorakul S &

Me-nasaveta P (2000) Immunity enhancement in black tiger

shrimp (Penaeus monodon) by a probiont bacterium

(Bacil-lus S11) Aquaculture 191, 271^288.

Ring E (2002) Probiotics in aquaculture In: Disease

Control in Fish and Shrimp Aquaculture in South Asia ^

Diagnosis and Husbandry Techniques (ed by Y Inui & E.R.

Cruz-Lacierda), pp 107^135 South Asian Fisheries opment Centre, Iloilo, Philippines.

Devel-Ring E (2004) Lactic acid bacteria in ¢sh and ¢sh farming In: Lactic Acid Bacteria (ed by S Salminen, A Ouwehand

& A von Wrigth), pp 581^610 Marcel Dekker, New York,

NY, USA.

Ring E (2008) The ability of carnobacteria isolated from

¢sh intestine to inhibit growth of ¢sh pathogenic ia: a screening study Aquaculture Research 39, 171^180 Ring E & Birkbeck T.H (1999) Intestinal micro£ora of ¢sh larvae and fry Aquaculture Research 30,73^93.

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

Ring E & Olsen R.E (1999) The e¡ect of diet on aerobic bacterial £ora associated with intestine of Arctic charr (Salvelinus alpinus L.) Journal of Applied Microbiology 86, 22^28.

Ring E., Strm E & Tabachek J.-A (1995) Intestinal

micro-£ora of salmonids: a review Aquaculture Research 26, 773^789.

Ring E., Olsen R.E., verli  & Lvik F (1997) E¡ect of dominance hierarchy formation on aerobic microbiota associated with epithelial mucosa of subordinate and dominant individuals of Arctic charr, Salvelinus alpinus (L.) Aquaculture Research 28, 901^904.

Ring E., Sepploa M., Berg A., Olsen R.E., Schillinger U & Holzapfel W (2002) Characterization of Carnobacterium divergens strain 6251 isolated from intestine of Arctic charr (Salvelinus alpinus L.) Systematic and Applied Micro- biology 25, 120^129.

Ring E., Olsen R.E., Mayhew T.M & Myklebust R (2003) Electron microscopy of the intestinal micro£ora of ¢sh Aquaculture 227, 395^415.

Ring E., Jutfelt F., Kanapathippillai P., Bakken Y., Sundell K., Glette J., Mayhew T.M., Myklebust R & Olsen R.E (2004) Damaging e¡ect of the ¢sh pathogen Aeromonas salmonicida ssp salmonicida on intestinal enterocytes of Atlantic salmon (Salmo salar L.) Cell and Tissue Research

318, 305^311.

Ring E., Schillinger U & Holzapfel W (2005) Antibacterial abilities of lactic acid bacteria isolated from aquatic ani- mals and the use of lactic acid bacteria in aquaculture In: Microbial Ecology in Growing Animals (ed by W Hol- zapfel & P Naughton), pp 418^453 Elsevier, Edinburgh, UK.

Ring E., Mikkelsen H., Kaino T., Olsen R.E., Mayhew T.M & Myklebust R (2006) Endocytosis of indegenous bacteria and cell damage caused byVibrio anguillarum in the fore- gut and hindgut of spotted wol⁄sh (Anarhichas minor Olafsen) fry An electron microscopical study Aquaculture Research 37, 647^651.

Ring E., Salinas I., Olsen R.E., Nyhaug A., Myklebust R & Mayhew T.M (2007) Histological changes in Atlantic sal- mon (Salmo salar L.) intestine following in vitro exposure

to pathogenic and probiotic bacterial strains Cell and sue Research 328, 109^116.

Tis-Ring E., Myklebust R., Mayhew T.M & Olsen R.E (2007)

Bacterial translocation and pathogenesis in the digestive

tract of larvae and fry Aquaculture 268, 251^264.

Rinkinen M.E.,Westermarck S., Salminen A.C & Ouwehand

A.C (2003) Absence of host speci¢city for in vitro

adhe-sion of probiotic lactic acid bacteria to intestinal mucus.

Veterinary Microbiology 9, 55^61.

Robertson P.A.W., O’Dowd C., Burrels C., Williams P &

Aus-tin B (2000) Use of Carnobacterium sp as a probiont for

Atlantic salmon (Salmo salar L.) and rainbow trout

(Oncor-hynchus mykiss,Walbaum) Aquaculture 185, 235^243.

Sakata T (1990) Micro£ora in the digestive tract of ¢sh and

shell-¢sh In: Microbiology in Poecilotherms (ed by R

Le-sel), pp 171^176 Elsevier, Amsterdam, the Netherlands.

Salinas I., Cuesta A., Esteban M.A & Meseguer J (2005)

Dietary administration of Lactobacillus delbrueckii and

Ba-cillus subtilis, single or combined, on gilthead seabream

cellular innate immune responses Fish and Shell¢sh

Im-munology 19, 67^77.

Salinas I., Myklebust R., Esteban M.A., Olsen R.E., Meseguer

J & Ring E (2008) In vitro studies of Lactobacillus

del-brueckii subsp lactis in Atlantic salmon (Salmo salar L.)

foregut: tissue responses and evidence of protection

against Aeromonas salmonicida subsp salmonicida

epithe-lial damage.Veterinary Microbiology 128, 167^177.

Salminen S., Deighton M.A., BennoY & Gorbach S.L (1998)

Lactic acid bacteria in health and disease In: Lactic Acid

Bacteria Microbiology and Functional Aspects (ed by S.

Salmonen & A von Wright), pp 221^253 Marcel Dekker,

New York, NY, USA.

Samuelsen O.B., Nerland A.H., Jrgensen T., Schrder M.B.,

Svsand T & Bergh  (2006) Viral and bacterial diseases of

Atlantic cod Gadus morhua, their prophylaxis and treatment:

a review Diseases of Aquatic Organisms 71, 239^254.

Santos Y., Pazos F., Bandin I & Toranzo A.E (1995) Analysis

of antigens present in the extracellular products and

cell-surface of Vibrio anguillarum serotypes O1, O2, and O3.

Applied and Environmental Microbiology 61, 2493^2498.

Scholz U., Garcia-Diaz G., Ricque D., Cruz-Suarez L.E.,

Var-gas-Albores F & Latchford J (1999) Enhancement of

vi-briosis resistance in juvenile Penaeus vannamei by

supplementation of diets with di¡erent yeast products.

Aquaculture 176, 271^283.

Seppola M., Olsen R.E., Sandaker E., Kanapathippillai P.,

Holzapfel W & Ring E (2006) Random ampli¢ed

poly-morphic DNA (RAPD) polymerase chain reaction

analy-sis of carnobacteria isolated from hindgut chamber and

large intestine of Atlantic cod (Gadus morhua L.)

Systema-tic and Applied Microbiology 29, 131^137.

Skirpstunas R.T & Baldwin T.J (2002) Edwardsiella ictaluri

invasion of IEC-6, Henle 407, fathead minnow and

chan-nel cat¢sh enteric epithelial cells Diseases of Aquatic

Or-ganism 51, 161^167.

Skjermo J & Vadstein O (1999) Techniques for microbial

control in the intensive rearing of marine larvae

Aquacul-ture 177, 333^343.

Tinh N.T.N., Dierckens K., Sorgeloos P & Bossier P (2008) A review of the functionality of probiotics in the larvicul- ture food chain Marine Biotechnology 10, 1^12 Toranzo A.E., Mamarinos B & Romalde J.E (2005) A review

of the main bacterial ¢sh diseases in mariculture systems Aquaculture 246, 37^61.

Valdivia R.H., Hromockyj A.E., Monack D., Ramakrishnan

L & Falkow S (1996) Applications for green £uorescent protein (GFP) in the study of host^pathogen interactions Gene 173, 47^52.

Van der Marel M., Schroers V., Neuhaus H & Steinhagen D (2008) Chemotaxis towards, adhesion to, and growth in carp gut mucus of two Aeromonas hydrophila strains with di¡erent pathogenicity for common carp, Cyprinus carpio

L Journal of Fish Diseases 31, 321^330.

Vanderpool C.,Yan F & Polk D.B (2008) Mechanisms of biotics action: implication for therapeutic applications in in£ammatory bowel diseases In£ammatory Bowel Dis- eases 14, 1585^1596.

pro-Verschuere L., Rombaut G., Sorgeloos P & Verstraete W (2000) Probiotic bacteria as biological control against in aquaculture Microbiology Molecular Biology Reviews 64, 655^671.

Vine N.G., Leukes W.D & Kaiser H (2004) In vitro growth characteristics of ¢ve candidate aquaculture probiotics and two ¢sh pathogens grown in ¢sh intestinal mucus FEMS Microbiological Letters 231, 145^152.

Vine N.G., Leukes W.D., Kaiser H., Baya S & Baxter J (2004) Competition of attachment of aquaculture probiotic and pathogenic bacteria to ¢sh intestinal mucus Journal of Fish Diseases 27, 319^326.

Vine N.G., Leukes W.D & Kaiser H (2006) Probiotics in marine larviculture FEMS Microbiology Reviews 30, 404^427.

Weiss N., Schillinger U & Kansdler O (1983) Lactobacillus lactis, Lactobacillus leichmannii and Lactobacillus bulgar- icus, subjective synonyms of Lactobacillus delbrueckii, and description of Lactobacillus delbrueckii subsp lactis comb nov and Lactobacillus delbrueckii subsp bulgari- cus comb nov Systematic and Applied Microbiology 4, 552^557.

Yin J.F., Zhou X.Y., Li J.Q., Li Y.H., Hou H.L & Zhang W.H (2007) A study on the survival dynamics of Bacil- lus subtilis in water using green £uorescent protein labelling Chinese Journal of Analytical Chemistry 35, 1405^1409.

Ying C., Lei W., Jiazhong L., Zhantao S & Liguo A (2007) Identi¢cation and puri¢cation of a novel adhesion-asso- ciated protein in a new strain of Lactobacillus, L 15, from

£ounder (Paralichthys olivaceus) Veterinary Microbiology

122, 116^122.

Zhou Z., Liu Y., Shi P., He S., Yao B & Ring E (2009) Molecular characterization of the autochthonous mi- crobiota in the gastrointestinal tract of yellow grouper (Epinephelus awoara) cultured in cages Aquaculture

286, 184^189.

Changes to the histological gill structure and

haemolymph composition of early blue swimmer

ammonia-N exposure and the post-exposure recovery

Nicholas Romano & Chaoshu Zeng

School of Marine and Tropical Biology, James Cook University, Townsville, Qld, Australia

Correspondence: N Romano, School of Marine and Tropical Biology, James Cook University, Townsville, Qld 4811, Australia E-mail: nicholas.romano@jcu.edu.au

Abstract

It is yet unclear whether sub-lethal ammonia-N

levels cause irreparable damage to aquatic

crusta-ceans, or if recovery is possible, the potential factors

involved The aim was to investigate the e¡ect of

0.706 and 2.798 mmol L 1ammonia-N exposure on

the haemolymph osmolality, Na1, K1, Ca21, pH,

am-monia-N, total haemocyte counts (THC) and gill

his-topathology of Portunus pelagicus juveniles at 0, 3, 6,

12, 24 and 48 h respectively Following 48 h, crabs

were transferred to pristine seawater allowing a

re-covery period up to 96 h and similarly measured In

addition moribund crabs, induced from lethal

ammo-nia-N levels of 7.036 and 10.518 mmol L 1, were

mea-sured for haemolymph osmolality/ions and pH levels

The results demonstrate that despite severe gill

da-mage within 6- and 1h of 0.706 and 2.798 mmol L 1

ammonia-N exposure, respectively, no signi¢cant

change (P40.05) in the haemolymph osmolality,

Na1, K1, Ca21or pH levels occurred or by

ammo-nia-N-induced morbidity Although the gills can

completely recover within 24 and 48 h post exposure

to 0.706 and 2.798 mmol L 1ammonia-N,

respec-tively, likely facilitated by signi¢cant haemocyte

in-creases (Po0.05) within the haemolymph and gill

lamellae, dependent factors were the previous

am-monia-N concentration and recovery duration while

individual variability was also noticed

Keywords: gill histology, haemolymph ammonia,

osmoregulation, Portunus pelagicus, recovery,

sub-lethal ammonia-N,THC

IntroductionAmmonia is the ¢rst stage of the nitri¢cation cycleand among the nitrogenous pollutants of nitriteand nitrate, ammonia is often the most toxic toaquatic animals (Meade & Watts 1995; Romano &Zeng 2007a^c) On closed aquaculture systems thatoften utilize high stocking densities with intensivefeeding, periodic spikes of sub-lethal ammonia-Nlevels are a ubiquitous concern for aquaculturefarmers (Timmons, Ebeling, Wheaton, Summerfelt

& Vinci 2002; Kir & Kumlu 2006) Furthermore,anthropogenic runo¡ and discharges of e¥uentwater from aquaculture systems also makes ammo-nia an ecologically relevant pollutant to aquaticcrustaceans (Biao, Zhuhong & Xiaorong 2004; Dave

& Nilsson 2005)

It has been previously demonstrated that elevatedammonia-N can rapidly cause severe damage to thegill structure of crustaceans including necrosis,hyperplasia, epithelial damage, pillar cell (PC) dis-ruption and lamellae collapse (Rebelo, Rodriguez,Santos & Ansaldo 2000; Romano & Zeng 2007a).The relative high vulnerability of the gills to poten-tial pollutants may possibly be explained by theirconstant contact with the external medium Thiscan be of particular concern because the gills ofaquatic animals are multi-functional organs re-sponsible for many crucial physiological processes,including ion exchange, acid/base balance, ammo-nia-N excretion and respiration (Harrison & Humes1992; Pe¤queux 1995; Freire, Onken & Mcnamara2008) Indeed previous investigations detected

signi¢cant decreases to haemolymph osmolality

and/or Na1 ions at elevated ammonia-N levels

(Young-lai, Charmantier-daures & Charmantier

1991; Chen & Chen 1996; Harris, Coley, Collins &

Mccabe 2001; Romano & Zeng 2007d), which may

be linked to gill damage

Because the gills of crustaceans clearly have

im-portant roles, are particularly vulnerable to

exter-nal pollutants and the damage has been linked to

their mortality (Rebelo et al 2000; Romano & Zeng

2007a), correlating such damage in sequence with

the physiological responses of crustaceans may

yield important information on the modes and

pro-gression of toxicity However, to date, histological

examination of ammonia-N-induced gill damage,

and the associated physiological changes linked

with their function, have focused on a single

sam-pling point of exposure (e.g Rebelo et al 2000;

Romano & Zeng 2007a; Miron, Moraes, Becker,

Crestani, Spanevello, Loro & Baldisserotto 2008)

Furthermore, it is yet unclear if sub-lethal

am-monia-N levels will cause irreparable damage to

the gills of aquatic animals, or if healing is

possi-ble, the potential factors involved One of these

healing processes may include the presence of

haemocytes, known to facilitate necrotic tissue

removal (Battistella, Bonivento & Amirante 1996;

Johansson, Keyser, Sritunyalucksana & S˛derhll

2000), within the haemolymph and gills of aquatic

crustaceans

The blue swimmer crab, Portunus pelagicus,

commonly inhabits marine and estuarine systems,

although they are characterized as being weak

os-moregulators (Romano & Zeng 2006) Increasingly

this species is becoming important to the ¢sheries

industry throughout the Indo-Paci¢c region and is

an emerging aquaculture species for pond and

re-circulating systems (Walker 2006; Romano & Zeng

2006) Because of the concerns of ammonia-N on

aquaculture systems, the experiment was hence

designed to continuously monitor the haemolymph

osmolality, Na1, K1, Ca21, pH, ammonia-N, total

haemocyte counts (THC), as well as the gill

histo-pathology of early P pelagicus juveniles over 48 h of

exposure to sub-lethal ammonia-N levels and the

subsequent post-exposure recovery period up to

96 h in pristine seawater (no added ammonia-N)

In addition, two lethal ammonia-N levels were used

to induce crab morbidity to determine whether

hae-molymph osmolality, Na1, K1, Ca21and pH levels

were disrupted to potentially explain causes for

300 L tanks at temperature 29 1 1C and salinity of

25 1% Larvae were initially fed rotifers nus sp.) with the daily addition of microalgae Nanno-chlorposis sp However, from the Zoea II onwards,newly hatched and enriched Artemia were added atincreasing densities of 1^5 Artemia mL 1until thelarvae ¢nal settled as juvenile crabs after approxi-mately 2 weeks

(Branchio-Upon metamorphosis to the ¢rst stage crabs (C1),they were transferred outdoors and further cultureduntil they reached the crab 5 (C5) stage, with meancarapace width of 18.76 0.26 mm) This size waschosen because it permitted a reliable and adequatemount of haemolymph extraction The juvenile crabculture method is described by Romano and Zeng(2007a) and the process took approximately 3 weeks.When a su⁄cient number of crabs reached the C5stage, similar-sized crabs were brought indoors forthe commencement of the experiment

Experimental design and set-upTest solutions

A 10 000 mg L 1 ammonia-N stock solution wasmade by adding 38.2 g of NH4Cl to 1 L of distilledwater This stock solution was then diluted in sea-water to create the desired ammonia-N concentra-tions of the test solutions used for experiments 1 and

2 The salinity of the source seawater was 36%,which was reduced to 30% using de-chlorinatedfreshwater and the ammonia-N, nitrite-N and ni-trate-N concentrations were measured and all werebelow 0.01mg L 1, which were previously analysed

at Australian Centre for Tropical Freshwater search (ACTFR) according to APHA (1989) The pH

Re-of each test solution was maintained at 8.1 throughthe addition of sodium hydroxide pellets and the pH

measured using a pH digital meter (WP-80; TPS,

Brisbane, Australia)

Experiment 1

A total of 288 crabs (C5 stage; mean weight 5

0.225 0.012 g), each individually kept within a

sepa-rate container thus acting as a replicate, were used in

the experiment Among these, 140 replicate crabs

were used for each of the two sub-lethal ammonia-N

treatments of 0.714 and 2.857 mmol L 1(or 10 and

40 mg L 1ammonia-N) respectively (based on values

from Romano & Zeng 2007a), while eight replicate

crabs were used for the control (no ammonia-N added)

Each crab was individually placed within a 5 L

con-tainer, ¢lled with 2 L of the desired test solution and

gently aerated via a ¢ne-tipped pipette Each container

received a daily100% water exchange according to the

‘static renewal method’described by the APHA (1989)

The majority of the crabs were daily fed formulated

crumble feed designed for the tiger prawn, Penaeus

monodon (43% protein,6% fat,3% ¢bre) in the morning

during the exposure and recovery phase However,

each crab was starved 1 day before haemolymph

sam-pling, which was facilitated by marking the containers

at di¡erent time intervals, to ensure1day starved crabs

were used To prevent the decomposition of food

a¡ect-ing the water ammonia-N levels, the feeds were

si-phoned out after 3 h, which was su⁄ciently long

enough for the crabs to cease eating All containers

were held within six 1000 L freshwater baths and the

temperature was maintained at 28 0.5 1C through

submersible heaters and air conditioning

At time intervals of 0,1, 3,6,12, 24, 36 and 48 h, the

haemolymph from eight intermoult replicate crabs

exposed to either 0.714 or 2.857 mmol L 1

ammo-nia-N treatments (10 and 40 mg L 1ammonia-N

respectively) were obtained via a syringe inserted

through the proximal arthropodal membrane at the

base of the right second walking leg Haemolymph

samples from ¢ve crabs were then measured for

hae-molymph osmolality and Na1, K1and Ca21levels

while haemolymph samples from the other three

crabs were used to measure the haemolymph pH,

THC and ammonia-N levels To determine

haemo-lymph Na1, K1and Ca21levels a haemolymph

sam-ple of 20mL was immediately diluted with 2 mL of

distilled water and analysed on £ame photometer

(Sherwood 410, Cambridge, UK) To determine the

haemolymph osmolality, an aliquot of haemolymph

(50mL) was immediately analysed on a cryoscopic

osmometer (Osmomat 030; Gonotec, Cambridge,

UK) The haemolymph pH was measured using a pHdigital meter (WP-80;TPS) equipped with a micro-pHelectrode (MI-710, Microelectrodes, Bedford, NH,USA) after a two-point calibration with precision buf-fers Following the measurement of pH, the haemo-lymph was divided into two portions to be used foreither determining the haemolymph THC or ammo-nia-N levels To measure the haemolymph THC, antic-oagulant (citrate-concentrated solution, 4% w/v) wasadded at a ratio of 1:9 (anticoagulant:haemolymph)and the haemocytes were counted on a haemocyt-ometer under a light microscope (magni¢cation

 10) After measuring for pH and a portion moved for THC measurements, the haemolymphsamples for measuring the ammonia-N were then di-luted with distilled water, frozen at 20 1C and ana-lysed within 2 days at the ACTFR using theNesslerization method 4500 NH3^G according toAPHA (1989) To determine the actual ammonia-Nconcentrations during the experiment, three sam-ples from each sub-lethal ammonia-N treatmentwere taken on the ¢rst and last day of the experimentand similarly analysed for ammonia-N at the ACTFR.The mean values of ammonia-N did not deviate fromthe stated concentrations by45% Between the 2days, the means of the actual concentrations in eachtreatment did not deviate by43% and the average ofthese actual values, of 0.706 and 2.798 mmol L 1ammonia-N will be used throughout to indicatethese two sub-lethal ammonia-N treatments.After each haemolymph sampling, four crabs wereimmersion ¢xed in a 10% (v/v) FAACC formalin solu-tion (4% formaldehyde, 5% acetic acid and 1.3% cal-cium chloride) for 3 days and transferred to 70% (v/v)ethanol until further processing For the histologicalexamination of the gills, the crabs were progressivelydehydrated at increasing concentrations of alcoholand embedded in para⁄n wax Sections (5mm) werecut using a rotary Leitz microtome (model 1512,Netley, NJ, USA), stained with haematoxylin and eo-sin The anterior and posterior gill structures wereexamined and digitally photographed under a lightmicroscope

re-All crabs subjected to haemolymph sampling wereremoved from the experiment and not used a secondtime Following the 48-h exposure to either 0.714 or2.857 mmol L 1 ammonia-N, all remaining crabswere immediately transferred to pristine seawaterwith no added ammonia-N to initiate a 96-h recoveryperiod To determine the duration required to stabi-lize haemolymph osmolality, Na1, K1and Ca21, pHand ammonia-N levels, as well as normalization of

the gill structure post sub-lethal ammonia-N

expo-sure, eight crabs were sampled at 1, 3, 6, 12, 24, 36, 48,

72, 84 and 96 h, respectively, adopting the same

pro-cedure mentioned above However, during this

recov-ery experiment, once the haemolymph ammonia-N

levels had returned to those of the pre-exposure level,

no further analysis was performed

Experiment 2

To determine if osmo-ionoregulation and

haemo-lymph pH regulation in early P pelagicus juveniles is

interrupted by ammonia-N-induced morbidity, 16

crabs were exposed to lethal concentrations of 7.142

and 10.714 mmol L 1 ammonia-N (or 100 and

150 mg L 1ammonia-N) (Romano & Zeng 2007a),

and another 16 crabs served as a control (no

ammo-nia-N added) Similar experimental protocols and the

same crab stage/batch were used as those of

experi-ment 1 (see Experiexperi-ment 1) However, hourly

observa-tions for moribund crabs were made til 24 h only

(48 h in experiment 1) Morbidity was diagnosed as

the crabs exhibiting severe disorientation (e.g an

in-ability to remain upright) and/or little movement

when gently stimulated with a glass rod.When a

mor-ibund crab was observed, it was sampled immediately

for haemolymph osmolality, Na1, K1, Ca21and pH

levels measurements using the methods described in

Experiment 1 For purposes of comparison, a control

crab was simultaneously sampled for analysis To

de-termine the ammonia-N levels of the water, three

samples from each treatment were measured using

the methods described in Experiment 1 The actual

values in the 7.142 and 10.714 mmol L 1

ammonia-N treatments were 7.036 and 10.518 mmol L 1

am-monia-N, respectively, and these values will be used

throughout the text

Data analysisThe haemolymph ammonia-N, Na1, K1and Ca21levels are expressed as mmol L 1 To convertmmol L 1to mg L 1, multiply mmol L 1by the re-spective molecular weights In the case of ammonia-

N conversions to mg L 1, the molecular weight ofnitrogen is used

Because the damage to the anterior and posteriorgills following elevated ammonia-N exposure ap-peared similar, only the anterior gill structuralchanges (pairs 2^4) were quanti¢ed To quantifythese histopathological changes, from each ammo-nia-N treatment and exposure/recovery duration, 30lamellae were randomly chosen from each of the fourreplicate crabs for measurements of lamellae width(based on the distance from each epithelia, see line

on Fig 1a, which de¢nes lamellae width, to quantifyswelling or collapse) and number of haemocytes pre-sent within the lamellae The lamellae width wasmeasured using a micrometer (1mm) and the number

of haemocytes was counted within each lamellae.The data from the 30 lamellae, of each replicate crab,were then pooled for statistical analysis

To determine any signi¢cant e¡ects (Po0.05) of theammonia-N levels and exposure/recovery duration onthe haemolymph composition or quanti¢ed gill histo-logical measurements either a one-way or a two-wayanalysis of variance (ANOVA)was used One-wayANOVAswere used to determine signi¢cant di¡erences overtime within each ammonia-N treatment as well as sig-ni¢cant di¡erences between ammonia-N treatments

at the same time interval A two-wayANOVAwas used

to determine any signi¢cant time, ammonia-N centration or interaction with the haemolymph com-position or gill histopathology To determine anysigni¢cant e¡ects (Po0.05) of ammonia-N-induced

con-Figure 1 The anterior gills of early Portunus pelagicus juveniles showing the lamallae of control crabs Note the presence

of intact pillar cells (PC), low incidences of haemocytes (HAE) within the lamellae and undamaged epithelium The labelled line refers to how the lamellae with was measured (a) Magni¢cation  20 and (b) magni¢cation  10; (a) scalebar 5 30mm and (b) 60 mm

un-morbidity on the haemolymph osmolality, Na1, K1,

Ca21and pH levels a one-wayANOVAwas used

Di¡er-ences between treatments were determined using

Duncan’s multiple range test (Duncan 1955) using the

SPSSstatistical software version 16.0

Results

Haemolymph osmolality, Na1, K1, Ca2, pH,THC

and ammonia-N

In both experiments 1 and 2, the haemolymph

osmol-ality and Na1, K1, Ca21levels of the crabs remained

above that of the external medium at a salinity of

30% (composition of seawater given in the ¢rst line

of Table 1) indicating hyperosmo-ionoregulation(Fig 2a^d; Table 1) The results of experiment 1showed that exposure to sub-lethal ammonia-Nlevels of 0.706 and 2.789 mmol L 1up to 48 h had

no signi¢cant e¡ect (P40.05) on the haemolymph molality, Na1, K1, Ca21or pH levels The haemo-lymph osmolality and pH of the crabs from theexperiment 1 ranged from 896 to 915 mOsm kg 1(Fig 2a) and 7.4 to 7.9 (Fig 3), respectively, whilemean haemolymph Na1, K1and Ca21 levels £uc-tuacted from 508 to 516, 10 to 15 and 10.8 to

os-14 mmol L 1, respectively, throughout the 48-h sure and subsequent 96-h recovery period (Fig 2b^d).Similarily in experiment 2, moribund crabs induced

expo-by exposed to subtantially higher lethal ammonia-N

Table 1 Osmolality, Na 1 , K 1 , Ca 21

and pH levels of the seawater (30%) and in the haemolymph ( SE) of the control and moribund early Portunus pelagicus crabs induced by exposure to lethal concentrations of ammonia-N (7.142 and 10.714 mmol L 1 ) within 24 h

Osmolality (mOsm kg 1 ), ionic composition (mmol L 1 ) and pH of seawater at a salinity of 30%

Ammonia-N

treatment

Osmolality (mOsm kg 1 ), ionic composition (mmol L 1 ) and pH of the crab haemolymph

No signi¢cant ammonia-N e¡ect on the haemolymph were detected (P40.05).

Figure 2 The mean haemolymph (a) osmolality (mOsm kg 1), (b) Na1, (c) K1and (d) Ca21levels (mmol L 1) ( SE) ofearly Portunus pelagicus juveniles exposed to two sub-lethal ammonia-N levels of 0.706 mmol L 1(dashed line) and2.798 mmol L 1(solid line) over 48 h and post-exposure recovery period No signi¢cant time or ammonia-N e¡ect weredeteced (P40.05)

levels of 7.036 and 10.518 mmol L 1also showed no

signi¢cant change (Po0.05) on their haemolymph

Na1, K1, Ca21and pH when compared with the

con-trol crabs (Table1) A two-wayANOVAdetected no

signif-icant time (of exposure and recovery duration),

ammonia-N level or interaction e¡ect (P40.05) on

the haemolyph osmolality, Na1, K1, Ca21or pH levels

In contrast to the haemolymph osmolality and

ions, both the haemolymph ammonia-N levels and

THC of the crabs exposed to the sub-lethal

ammo-nia-N levels showed an increasing trend with both

the ammonia-N concentration and duration of

expo-sure The mean haemolymph ammonia-N levels of

the control crabs (no added ammonia-N) was low at

0.69 10 3

mmol L 1 However, within1h of

expo-sure to 0.706 and 2.798 mmol L 1ammonia-N the

haemolymph ammonia-N levels signi¢cantly

in-creased (Po0.01) (Fig 4) The highest haemolymph

ammonia-N levels of 0.361 and 1.692 mmol L 1

oc-curred at 48 h, which was the longest exposure

dura-tion, at 0.706 and 2.798 mmol L 1 ammonia-N

respectively (Fig 4) However, within1h of

post-expo-sure recovery from 2.798 mmol L 1ammonia-N, the

haemolymph ammonia-N signi¢cantly decreased

(Po0.01) from the 48-h exposure peak By 12 h post

exposure to both sub-lethal ammonia-N levels the

haemolymph ammonia-N of the crabs was not

signif-icantly di¡erent (P40.05) from those in the control

(Fig 4) A two-wayANOVAdetected a signi¢cant time

(of exposure and recovery duration) and ammonia-N

concentration e¡ect (Po0.01) on the haemolymph

ammonia-N levels, however, no signi¢cant

interac-tion was detected (P40.05)

The mean haemolymph THC of the control crabs

was 29.51 ( 104

cells mL 1), however, by 12 and

6 h to f exposure to 0.706 and 2.798 mmol L 1

am-monia-N, respectively, the haemolymph THC

signi¢-cantly increased (Po0.01) (Fig 5) Once thissigni¢cant THC increase occurred, it continuedthroughout the exposure to both 0.706 and2.798 mmol L 1ammonia-N At the 84-h recoveryfrom post exposure to both sub-lethal ammonia-N le-vels, the haemolymph THC of the crabs signi¢cantlydecreased (Po0.05) to levels that were not signi¢-cantly di¡erent (P40.05) from those in the control(Fig 5) A two-wayANOVAdetected a signi¢cant timee¡ect (of exposure and recovery duration) (Po0.05)

on the haemolymph THC, however, no signi¢cantammonia-N level or interaction e¡ect was detected(P40.05)

Gill histopathological changesThe anterior lamellae of the control crabs showed in-tact PC, occasional presence of haemocytes (HAE)within the lamellae and a thin epithelium (Fig 2aand b) Although no mortalities occurred during the48-h exposure to 0.706 and 2.798 mmol L 1ammo-nia-N, at 3 and 1h, respectively, the gill lamellaewidth signi¢cantly decreased (Po0.01) and the num-ber of haemocytes (HAE) signi¢cantly increased(Po0.05) within the gill lamellae of these crabs Suchgill histopathological changes lasted throughout therest of the 48-h exposure period at both sub-lethalammonia-N levels although the degree of suchchanges was greater at the higher ammonia-N level

of 2.798 mmol L 1 (Table 2) Other histologicalchanges, which were di⁄cult to quantify, also oc-curred throughout the exposure duration, which in-cluded epithelial damage (e.g sloughing, thickeningand detachment), disrupted/necrotic pillar cells(DPC) leading to a complete breakdown in the intra-lamellar septum and lamellae distortion (Fig 6a^f)

Figure 3 The mean haemolymph pH ( SE) of early Portunus pelagicus juveniles exposed to two sub-lethal ammonia-Nlevels of 0.706 mmol L 1(dashed line) and 2.798 mmol L 1(solid line) over 48 h and post-exposure recovery period Nosigni¢cant time or ammonia-N e¡ect were deteced (P40.05)

When the recovery period was initiated, an

in-crease to the gill lamellae width and dein-crease in the

haemocyte number within the gill lamellae occurred

and these were more rapid for the crabs exposed to

0.706 mmol L 1ammonia-N than those exposed to

2.798 mmol L 1ammonia-N (Table 2) At 24 and

36 h of recovery, the lamellae width of the crabs

pre-viously exposed to 0.706 mmol L 1ammonia-N

sig-ni¢cantly increased (Po0.01) while the haemocyte

number decreased (Po0.05) when compared with

the 48-h exposed crabs However, for the crabs

pre-viously exposed to 2.798 mmol L 1ammonia-N, a

substantially longer recovery period of 48 and 72 hwas required for signi¢cantly increased lamellaewidth (Po0.01) and decreased haemocyte number(HAE) (Po0.05) to occur respectively (Table 2) In ad-dition, coinciding with signi¢cant increases to the la-mellae width and decrease in number of haemocyteswithin the lamellae, other signs of gill normalizationalso occurred, which included PC restructuring,epithelial healing, and a clearing of the intralamel-lae septum (Fig 7a^f) A two-wayANOVAdetected asigni¢cant treatment and time e¡ect (of exposureand recovery duration) on the lamellae width and

Figure 5 The mean haemolymph total haemocyte count (THC) ( SE) of early Portunus pelagicus juveniles exposed tosub-lethal ammonia-N levels of 0.706 mmol L 1(dashed line) and 2.798 mmol L 1(solid line) over 48 h and post-expo-sure recovery period Di¡erent lower and upper case letters indicate signi¢cant di¡erences (Po0.05) within 0.706 and2.798 mmol L 1ammonia-N respectively

Figure 4 The mean haemolymph ammonia-N levels (mmol L 1

) ( SE) of early Portunus pelagicus juveniles exposed totwo sub-lethal ammonia-N levels of 0.706 mmol L 1(dashed line) and 2.798 mmol L 1(solid line) over 48 h and post-exposure recovery period Di¡erent lower and upper case letters indicate signi¢cant di¡erences (Po0.05) within 0.706and 2.798 mmol L 1ammonia-N respectively

Table 2 The mean lamellae width (mm) (  SE) and haemocyte number (  SE) within the anterior lamellae of early nus pelagicus juveniles exposed to two sub-lethal ammonia-N levels over 48- and the 96-h post-exposure recovery period

Portu-Sub-lethal ammonia-N concentration 0.706 mmol L 1 ammonia-N 2.798 mmol L 1 ammonia-N Lamellae width (lm) Haemocyte number Lamellae width (lm) Haemocyte number

Figure 6 Histopathological changes to the anterior gills of early Portunus pelagicus juveniles exposed to 2.798 mmol L 1ammonia-N for (a) 1h, (b) 6 h, (c) 12 h, (d) 24 h, (e) 36 h and (f) 48 h Note the drastic structural changes that occurred at1h of exposure, which includes lamellae constriction/collapse leading to a reduction in the intralamellar septum, abun-dance of disrupted pillar cells (DPC), epithelial thickening and elevated presence of haemocytes (HAE) Magni¢cation

 20; scale bar 5 30 mm

haemocyte number with the lamellae of early P

pela-gicus juveniles, however, no signi¢cant interaction

(Po0.05) was detected

Discussion

Fluctuating ammonia-N levels in an aquaculture

set-ting can be routinely experienced by crustaceans,

and may a¡ect productivity and their survival (Kir

& Kumlu 2006; Romano & Zeng 2007a) Because

the gills of aquatic animals are particularly

suscepti-ble to potential pollutants, the conditions of the

cur-rent experiment were created to examine/quantify

the e¡ects of sub-lethal ammonia-N exposure and

subsequent recovery on the gill structure of early

P pelagicus juveniles in sequence with a series of

hae-molymph parameters closely associated with their

function This may provide useful information

be-cause P pelagicus is currently being cultured on

recir-culating and pond systems, which may periodically

experience elevated ammonia-N levels

The results clearly show that despite exposure to

ammonia-N levels substantially below the threshold

of lethality (Romano & Zeng 2007a), the gills of early

P pelagicus crabs showed drastic histopathological

changes within a short period of exposure These

ob-served changes included epithelial ing, DPC due to necrosis, increased presence ofhaemocytes within the lamellae space, lamellae dis-tortion and lamellae constriction/collapse These

damage/thicken-¢ndings are in agreement with our previous study

on similar-sized crabs of the same species during anacute ammonia-N toxicity experiment (Romano &Zeng 2007a) One of the most drastic histopathologi-cal changes in the current study was lamellae col-lapse, which occurred within 3 and 1h of 0.706 and2.798 mmol L 1ammonia-N exposure, respectively,and once this occurred, it persisted to many of thelamellae throughout the remaining duration of ex-posure (i.e 48 h) This phenomenon within a rela-tively short time frame and at sub-lethal ammonia-Nlevels was unexpected because in our previous acuteammonia-N toxicity experiment utilizing a longerexposure duration (of 96 h) and higher ammonia-Nlevels (up to 7.137 mmol L 1) caused lamellae col-lapse only at the highest ammonia-N concentrationsclose to or at the 96-h LC50values (Romano & Zeng2007a) Although it may be possible that the longerammonia-N exposure of our previous experiment of

96 h allowed time for the crabs to adapt/acclimate

at the lower ammonia-N concentrations, because alevelling of LC50values at 60 h onwards due to a re-duction in mortality frequency is typical of early

Figure 7 Histopathological changes to the anterior gill of early Portunus pelagicus juveniles during the recovery period(no added ammonia-N) at (a) 1h, (b) 6 h, (c) 12 h, (d) 24 h, (e) 36 h and (f) 48 h from post-48 h exposure to 2.798 mmol L 1ammonia-N Note lamellae collapse was still evident at1and 6 h to recovery although some localized expansions () began

to occur at 12 h of recovery Greater expansions to the entire gill lamellae is observed at increasing recovery durations of

24 and 36 h and a nearly complete recovery has occurred by 48 h including the prevalence of intact pillar cells (PC).Magni¢cation  20; scale bar 5 30 mm

P pelagicus (Romano & Zeng 2007a) and other

crus-taceans (Lin & Chen 2001), further experimentation

on a longer time frame is needed to con¢rm this

In spite of lamellae collapse there was, indeed,

evi-dence suggesting a healing response of the crabs

be-cause the THC in the haemolymph, which are

responsible for the phagocytosis of foreign material/

necrotic tissue (Battistella et al 1996; Johansson et al

2000), signi¢cantly increased at 12 and 6 h of

expo-sure to 0.706 and 2.798 mmol L 1ammonia-N

re-spectively This rapid immunological response

coinciding with a signi¢cant increase to the

haemo-cytes within the gill lamellae (at 6 and1h of exposure

to 0.706 and 2.798 mmol L 1ammonia-N

respec-tively) may indicate this was initiated to cope with gill

structure damage The haemocyte increase within

the haemolymph and/or gill lamellae appears to be a

common response for early P pelagicus juveniles at

increased nutrient concentrations based on our

pre-vious investigations following their exposure to

ele-vated NH4^N, NO2^N, NO3^N, KNO2, KNO3 and

KCl^K levels (Romano & Zeng 2007a, b, 2009) This

may therefore have some potential as a non-speci¢c

biomarker for this species to elevated nutrient

expo-sure Interestingly, these results are in direct contrast

with previous reports on other crustaceans including

the giant freshwater prawn Macrobrachium

rosenber-gii (Cheng & Chen 2002), the Kuruma shrimp

Penaeus japonicus (Jiang, Yu & Zhou 2004), the white

shrimp Litopenaeus vannamei (Liu & Chen 2004), the

Chinese mitten-handed crab Eriocheir sinensis (Hong,

Chen, Sun, Gu, Zhang & Chen 2007), the Indian spiny

lobster Panulirus homarus (Verghese, Radhakrishnan

& Padhi 2007), the blue shrimp Litopenaeus

styliros-tris (Mugnier, Zipper, Goarant & Lemonnier 2008)

and the southern white shrimp Litopenaeus schmitti

(Rodr|¤guez-Ramos 2008) In all these species an

in-crease in external ammonia-N levels reportedly

caused a signi¢cant decrease in the haemolymph

THC It is unclear if such a haemolymph THC

reduc-tion in these species would diminish their healing

ca-pacity of the gills and further investigations may be

warranted due to the increasing threat of

anthropo-genic discharges to various ecosystems

During the initiated recovery period following

their exposure to both sub-lethal ammonia-N

expo-sure after 48 h, obvious signs of gill healing occurred

and to the best of our knowledge, our experiment

ap-pears to be the ¢rst to describe and quantify such an

occurrence Based on observations and

quanti¢ca-tion of the gill lamellae, the minimum duraquanti¢ca-tion to

al-low for a complete recovery from a 48-h exposure to

0.706 and 2.798 mmol L 1ammonia-N was 24 and

48 h respectively Although this recovery was clearlydependent on both the previous ammonia-N expo-sure concentration and duration of recovery, indivi-dual variability was also noticed Nevertheless, the

¢rst signs of recovery appeared similar, which cluded localized expansions along the gill lamellaeallowing for increases in the intralamellae septum(i.e space within the lamellae) This presumably per-mitted a further in¢ltration of haemocytes to healand phagocytize dead/damaged tissue (Harrison &Humes 1992) This was generally followed by a great-

in-er expansion to the intralamellar septum, epithelialhealing/thinning, reduced presence of haemocytesand PC normalization until the lamellae completelyexpanded and the structure was restored to the pre-exposed state.While this demonstrates gill healing ispossible following gill damage induced by the tem-porary exposure to sub-lethal ammonia-N concen-trations, it is unclear if exposure to other pollutantsbeing discharged into various ecosystems would eli-cit a similar healing response and warrants furtherinvestigations

Interestingly, despite the severe gill damage ring soon after the crabs were transferred to sub-lethal ammonia-N levels, which persisted to the end

occur-of 48-h exposure, no signi¢cant changes to the molymph osmolality, Na1, K1, Ca21or pH levels ofthe P pelagicus juveniles were detected while the hae-molymph ammonia-N levels remained substantiallylower than the external environment Furthermore,

hae-at ammonia-N levels of 10-fold higher, which werechosen to induce the morbidity of these crabs, a simi-lar result of undisrupted haemolymph osmolality,ions and pH was obtained Although early P pelagicusjuveniles are known to be weak osmoregulators (Ro-mano & Zeng 2006), the haemolymph ions were stilldi¡erent from the environment indicating ion regula-tion at a salinity of 30% Therefore these results sug-gest that, even for moribund crabs, the gill enzymes(i.e carbonic anhydrase and Na1/K1-ATPase activ-ity) responsible for transporting ions, regulating pHand excreting ammonia-N across a gradient were stillfunctioning This undisrupted haemolymph osmolal-ity and/or Na1ions are in contrast to other crusta-ceans such as the American clawed lobster Homarusamericanus (Young-Lai et al 1991), P japonicus (Chen

& Chen 1996), the freshwater cray¢sh Pacifastacusleniusculus (Harris et al 2001) and the mud crab Scyl-

la serrata (Romano & Zeng 2007d) It is believed that

a reduction in haemolymph Na1 ions, caused byammonia-N exposure, was the result of either a

breakdown of the NH41/Na1 excretion process

(Young-Lai et al 1991; Chen & Chen 1996) or a

depo-laristaion of the cell membrane (Harris et al 2001)

However, for the estuarine crab, Neohelice granulata

(also known as Chasmagnathus granulata), no

signi¢-cant reduction in haemolymph Na1 ions were

de-tected in spite of ammonia-N levels reaching lethal

levels and causing severe gill damage including

la-mellae collapse (Rebelo et al 2000) Rebelo et al

(2000) suggested that this result may be linked with

the relatively strong osmoregulatory abilities of N

granulata However, it is unlikely this suggestion can

be generalized for all crustaceans because in the

cur-rent study P pelagicus experienced no signi¢cant

change to haemolymph Na1 levels in spite of this

crab species being a relatively weak osmoregulator

(Romano & Zeng 2006) In contrast, it was previously

shown that elevated ammonia-N levels signi¢cantly

reduced the haemolymph Na1ions in S serrata

(Ro-mano & Zeng 2007d) despite this crab species being

considered a relatively strong osmoregulator (Chen

& Chia 1997) These results suggest highly

species-speci¢c responses/adaptations of crustaceans when

challenged with elevated ammonia-N levels and,

furthermore, that gill damage may not necessarily

be linked with disrupted osmoregulation

It seems possible that the undisrupted

osmo-ionregulation may contribute to the relatively high

tolerance of P pelagicus juveniles to ammonia-N

(Romano & Zeng 2007a), although another likely

contributing factor was the substantially lower

hae-molymph ammonia-N levels of the crabs compared

with the external environment throughout the

expo-sure duration This general ¢nding has been similarly

reported for other crustaceans (e.g Schmitt & Uglow

1997; Rebelo, Santos & Monserrat 1999; Weihrauch,

Becker, Postel, Luck-kopp & Siebers 1999; Harris et al

2001), which has been attributed to various factors

including low gill permeability, active ammonia-N

excretion and/or ammonia-N detoxi¢cation

depend-ing on the species (Schmitt & Uglow 1997; Weihrauch

et al 1999) However, direct comparisons between

species is di⁄cult because previous studies utilize

dif-ferent experimental conditions, which a¡ect the

ra-tios of the more permeable NH3levels Because NH3

crosses the gill bilayers, the proposed mechanism for

active ammonia-N excretion is the haemolymph

pro-tonation of NH3to NH41, which is subsequently

ex-creted to the environment (Weihrauch, Morris &

Towle 2004) and such a process was suggested to

contribute to signi¢cantly increasing the

haemo-lymph pH of early S serrata juveniles (Romano

& Zeng 2007d) However, in the current study thehaemolymph pH of P pelagicus juveniles remainedunaltered throughout the sub-lethal ammonia-N ex-posure or lethal ammonia-N levels causing morbidity.Rebelo et al (2000) detected a similar result with N.granulata juveniles and suggested a compensatory ef-fect occurred as a result of increasing haemolymph

CO2(leading to a haemolymph pH decrease) ing with haemolymph NH3protonation (leading to ahaemolymph pH increase) Obviously, further experi-mentation is warranted to determine the extent ofammonia-N exposure on the haemolymph CO2/O2levels in this species to help explain the exact cause(s)for the unaltered haemolymph pH and potential me-chanisms to ammonia-N-induced morbidity How-ever, importantly, elevated ammonia-N levels doesnot disrupt osmoregulation or acid/base balanceand, furthermore, causes no permanent damage tothe gill structure of these animals when su⁄cienttime is provided in pristine seawater

match-AcknowledgmentThis project was funded by the Graduate ResearchScheme, James Cook University, 2007 and it formspart of the PhD thesis for Nicholas Romano

ReferencesAPHA (1989) Standard Methods for the Examination of Water and Wastewater, 17th edn American Public Health Asso- ciation,Washington, DC, USA.

Battistella S., Bonivento P & Amirante G.A (1996) cytes and immunological reactions in crustaceans Ita- lian Journal of Zoology 63, 337^343.

Hemo-Biao X., Zhuhong D & Xiaorong W (2004) Impact of sive shrimp farming on the water quality of the adjacent coastal creeks from Eastern China Marine Pollution Bulle- tin 48, 543^533.

inten-Chen J.C & inten-Chen C.T (1996) Changes of osmotic and trolyte concentrations in the haemolymph of Penaeus japonicus exposed to ambient ammonia Comparative Bio- chemistry and Physiology 114C, 35^38.

elec-Chen J.C & Chia P.G (1997) Osmotic and ionic concentration

of Scylla serrata (Forskl) subjected to di¡erent salinity levels Comparative Biochemistry and Physiology 177A, 239^244.

Cheng W & Chen J.C (2002) The virulence of Enterococcus

to freshwater prawn Macrobrachium rosenbergii and its immune resistance under ammonia stress Fish Shell¢sh Immunology 12, 97^109.

Dave G & Nilsson E (2005) Increased reproductive toxicity

of land¢ll leachate after degradation was caused by nitrite AquaticToxicology 73, 11^30.

Duncan D.B (1955) Multiple-range and multiple F test

Bio-metrics 11, 1^42.

Freire C.A., Onken H & McNamara J.C 2008 A structure^

function analysis of ion transport in crustacean gills and

excretory organs Comparative Biochemistry and

Physiol-ogy 151A, 272^304.

Harris R.R., Coley S., Collins S & McCabe R (2001)

Ammo-nia uptake and its e¡ects on ionoregulation in the

fresh-water cray¢sh Pacifastacus leniusculus (Dana) Journal of

Comparative Physiology, B Metabolism Transport Function

171, 681^693.

Harrison F.W & Humes A.G (1992) Microscopic Anatomy of

Invertebrates.Volume 10: Decapod Crustaceans John Wiley

& Sons, New York, NY, USA.

Hong M., Chen L., Sun X., Gu S., Zhang L & Chen Y (2007)

Metabolic and immune responses in Chinese

mitten-handed crab (Eriocheir sinensis) juveniles exposed to

ele-vated ammonia Comparative Biochemistry and Physiology

145C, 363^369.

Jiang G., Yu R & Zhou M (2004) Modulatory e¡ects of

am-monia-N on the immune system of Penaeus japonicus to

virulence of white spot syndrome virus Aquaculture

241, 61^75.

Johansson M.W., Keyser P., Sritunyalucksana K & S˛derhll

K (2000) Crustacean haemocytes and haematopoiesis.

Aquaculture 191, 45^52.

Kir M & Kumlu M (2006) Acute toxicity of ammonia to

Penaeus semisulcatus postlarvae in relation to salinity.

Journal of World Aquatic Society 37, 231^235.

Lin Y.C & Chen J.C (2001) Acute toxicity of ammonia on

Li-topenaeus vannamei Boone juveniles at di¡erent salinity

levels Journal of Experimental Marine Biology and Ecology

259, 109^119.

Liu C.H & Chen J.C (2004) E¡ect of ammonia on the

im-mune response of white shrimp Litopenaeus vannamei

and its susceptibility to Vibrio alginolyticus Fish Shell¢sh

Immunology 16, 321^334.

Meade M.E & Watts S.A (1995) Toxicity of ammonia, nitrite

and nitrate to juvenile Australia cray¢sh, Cherax

quadri-carinatus Journal of Shell¢sh Research 14, 341^346.

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)

Aquacul-ture 277, 192^196.

Mugnier C., Zipper E., Goarant C & Lemonnier H (2008)

Combine e¡ect of exposure to ammonia and hypoxia on

the blue shrimp Litopenaeus stylirostris survival and

phy-siological response in relation to molt stage Aquaculture

274, 398^407.

Pe¤queux A (1995) Osmotic regulation in crustaceans

Jour-nal of Crustacean Biology 15, 1^60.

Rebelo M.F., Santos E.A & Monserrat J.M (1999) Ammonia

exposure of Chasmagnathus granulata (Crustacea,

Deca-poda) Dana, 1851: accumulation in haemolymph and

e¡ects on osmoregulation Comparative Biochemistry and Physiology 122A, 429^435.

Rebelo M.F., Rodriguez E.M., Santos E.A & Ansaldo M (2000) Histopathological changes in gills of the estuarine crab Chasmagnathus granulata (Crustacea-Decapoda) following actue exposure to ammonia Comparative Bio- chemistry and Physiology 125C, 157^164.

Rodr|¤guez-Ramos T (2008) E¡ects of Echerichia coli saccharides and dissolved ammonia on immune re- sponses in southern white shrimp Litopenaeus schmitti Aquaculture 274, 118^125.

lipopoly-Romano N & Zeng C (2006) The e¡ects of salinity on the survival, growth and haemolymph osmolality of early ju- venile blue swimmer crabs, Portunus pelagicus Aquacul- ture 260, 151^162.

Romano N & Zeng C (2007a) Ontogenetic changes in ance to acute ammonia exposure and associated histolo- gical gill alterations during early juvenile development of the blue swimmer crab, Portunus pelagicus Aquaculture

toler-266, 246^254.

Romano N & Zeng C (2007b) Acute toxicity of sodium nitrate, potassium nitrate and potassium chloride and their e¡ects on the haemolymph composition and gill structure

of early juvenile blue swimmer crabs (Portunus pelagicus Linnaeus, 1758) (Decapoda, Brachyura, Portunidae) Envir- onmental Toxicology and Chemistry 26, 1955^1962 Romano N & Zeng C (2007c) E¡ects of potassium on nitrate mediated changes to osmoregulation in marine crabs AquaticToxicology 85, 202^208.

Romano N & Zeng C (2007d) Acute toxicity of ammonia and its e¡ects on the haemolymph osmolality, ammonia-

N, pH and ionic composition of early juvenile mud crabs, Scylla serrata (Forskl) Comparative Biochemistry and Physiology 148A, 278^285.

Romano N & Zeng C (2009) Subchronic exposure to nitrite, potassium and their combination on survival, growth and total haemocyte count and gill structure of juvnile blue swimmer crabs, Portunus pelagicus Ecotoxicology and Environmental Safety 72, 1287^1295.

Schmitt A.S.C & Uglow R.F (1997) E¡ects of ambient nia levels on blood ammonia, ammonia excretion and heart and scaphognathite rates of Nephrops norvegicus Marine Biology 127, 411^418.

ammo-Timmons M.B., Ebeling J.M., Wheaton F.W., Summerfelt S.T.

& Vinci B.J (2002) Recirculating Aquaculture Systems, 2nd edn Northeastern Regional Aquaculture Center, Ithaca,

NJ, USA.

Verghese B., Radhakrishnan E.V & Padhi A (2007) E¡ect of environmental parameters on immune response of the Indian spiny lobster, Panulirus homarus (Linnaeus, 1758) Fish Shell¢sh Immunology 23, 928^936.

Walker T (2006) Coral Coast Mariculture claws: a market for softshell Austasia Aquaculture 3^5.

Weihrauch D., Becker W., Postel U., Luck-Kopp S & Siebers D (1999) Potential of active excretion of ammonia in three

di¡erent haline species of crabs Journal of Comparative

Physiology B 169, 25^37.

Weihrauch D., Morris S & Towle D.W (2004) Ammonia

ex-cretion in aquatic and terrestrial crabs: a review Journal of

The impacts of aquaculture development on food

security: lessons from Bangladesh

Khondker Murshed^e-Jahan1, Mahfuzuddin Ahmed2& Ben Belton3

1

The WorldFish Center, Bangladesh and South Asia, Banani, Dhaka, Bangladesh

2 Asian Development Bank, Manila, Philippines

3 University of Stirling, Stirling, UK

Correspondence: K M Jahan, The WorldFish Center, Bangladesh and South Asia, House No 22B, Road No 7, Block F, Banani, Dhaka 1213, Bangladesh E-mails: k.jahan@cgiar.org, kmjahan@gmail.com

Abstract

Fish contribute a signi¢cant amount of animal

pro-tein to the diets of people in Bangladesh, about 63%

of which comes from aquatic animals In

Bangla-desh, ¢sh is mainly derived from two sources: capture

and culture Aquaculture has shown tremendous

growth in the last two decades, exhibiting by about

10% average annual growth in production Capture

¢sheries, although still the major source of supply of

¢sh, have become static or are in decline due to

over-¢shing and environmental degradation, and it is now

believed that aquaculture has the greatest potential

to meet the growing demand for ¢sh from the

in-creasing population At present, aquaculture

produc-tion accounts for about one-third of the total ¢sh

production in Bangladesh This paper examines the

impact of an aquaculture development project in

Bangladesh on food security, with particular

empha-sis on the poor The analyempha-sis shows a positive impact

of aquaculture development on employment, income

and consumption A number of implications for

pol-icy in areas that might strengthen these outcomes

are discussed and recommendations are presented

Keywords: aquaculture, food security, poor, poverty,

Bangladesh

Introduction

Bangladesh is one of the most densely populated

countries in the world, with 143 million people living

in a land mass of only 1475 00 km2 Approximately

76% of the people live in rural areas, of which about

one half lives below the national poverty line (theminimum daily food intake is  2122 kilo caloriecapita 1day 1) (Government of Bangladesh andUnited Nation 2007) According to FAO statistics,about 43.1 million su¡er from under-nutrition,equivalent to 5% of the total undernourished people

in the world Estimates also show that around 36% ofthe population of Bangladesh spendsoUS$1day 1and are unable to access food as required, even if it isavailable (FAO 2006a) Therefore, to achieve the tar-get of the United Nations Millennium developmentgoals, food security has emerged as the main chal-lenge for the Bangladesh government

Attaining self-su⁄ciency in the production of eals and other staple food commodities has domi-nated the debate over food security in developingcountries like Bangladesh over recent decades Therecognition that food insecurity can result from pov-erty and lack of access to su⁄cient nutritious food isrelatively recent (Ahmed & Lorica 2002; Maxwell &Slater 2003; Hishamunda & Ridler 2006; Thorpe,Reid, Anrooy, Brugere & Becker 2006; Bose & Dey2007) This is because food self-su⁄ciency is neithernecessary nor su⁄cient to guarantee food security.Self-su⁄ciency is not necessary because there isscope in an open economy for food imports to meetdemand On the other hand, poverty can prevent thepoor from accessing nutritious food due to their lim-ited purchasing power Like many other developingcountries, Bangladesh therefore places emphasis

cer-on diversi¢ed producticer-on, employment and incomegeneration on farms as a means of achieving foodsecurity in its Poverty Reduction Strategy Paper(PRSP) (Bangladesh Planning Commission 2005)

Aquaculture is the fastest-growing food-producing

sector in Bangladesh and has demonstrated

continu-ous increases in total production throughout the

re-cent decades By contrast, cereals and similar

commodities have shown a relatively slow growth

Estimates show that from 1983/1984 to 2005/2006,

aquaculture production increased by an average

an-nual growth rate of 10% In contrast, cereal

produc-tion exhibited an average growth rate of only 3% over

this period (BBS 2006; FSYB 2006).Widespread

adop-tion of new aquaculture technologies and

improve-ments in farming techniques have helped aquaculture

to maintain a high growth rate In Bangladesh,

aqua-culture is mainly a rural activity, where about 73% of

rural households engage in some form of freshwater

aquaculture (Mazid 1999)

Because of its rapid expansion (both in Bangladesh

and in many other Asian countries), aquaculture is

often considered to have the potential to enhance

food security among adopters and the population at

large Aquaculture is claimed to enhance food

secur-ity directly by the production of ¢sh for household

consumption and by improving the supply and

redu-cing the price of ¢sh in the market, and indirectly by

contributing to farm diversi¢cation and the creation

of new employment opportunities and income

streams As a result of this logic, aquaculture has

been promoted as a mechanism for rural development

with a poverty alleviation focus for several decades

(Edwards 1999; Dey, Rab, Paraguas, Piumsombun,

Bhatta, Alam & Ahmed 2005; World Bank 2006)

These interlocking claims possess an intuitive

ap-peal and are repeatedly stated in a voluminous grey

and academic literature (Halwart, Funge-smith &

Moehl 2003; FAO 2006b; for example) However, with

a handful of exceptions (Thompson, Firoz Khan &

Sultana 2006; Irz, Stevenson,Tanoy,Villarante &

Mor-issens 2007; To¢que & Gregory 2008), peer-reviewed

research based on primary data and dealing with the

impacts of speci¢c projects designed to promote

aquaculture has been limited This paper aims to

augment this body of knowledge by examining the

potential of aquaculture to impact food security

using data returned from the monitoring and

evalua-tion of a development project implemented by the

WorldFish Center in Bangladesh, namely the

Devel-opment of Sustainable Aquaculture Project (DSAP)

The intention of doing so is to evaluate the validity of

assumptions about aquaculture (which play a

signi¢-cant role in the policy formulation of numerous

do-nor, state and third sector institutions), and to

identify mechanisms by which aquaculture’s

theore-tical contribution to food security may be enhanced

in practice

The paper comprises six sections The next sectionpresents an overview of the ¢sheries sector of Ban-gladesh Section three outlines the analytical frame-work for the study, including a discussion ofhypotheses concerning the role of aquaculture vis-a-vis food security Section four presents the ¢ndings

of the study and explains the farm- and level impacts of aquaculture adoption with respect

household-to income, employment and ¢sh consumption tors that mediate the impacts of aquaculture adop-tion with respect to food security, and theirassociated policy implications are discussed in the

Fac-¢fth section The conclusions are discussed in the

¢nal section

Fisheries sector of BangladeshThe ¢sheries sector has played an increasingly im-portant role in e¡orts to uplift the economy of Ban-gladesh The ¢sheries sector accounted for about 3%

of the country’s GDP, 28% of the value added in cultural production and 6% of total export earnings

agri-in 2003/2004 (Bangladesh Economic Review 2005)

It is a labour-intensive and quick-yielding sector inwhich an estimated 1.3 million of the rural popula-tion are directly employed (Karim, Ahmed, Talukder,Taslim & Rahman 2006) Fish have highly desirablenutrient pro¢les, provide an excellent source ofhigh-quality animal protein that is easily digestibleand of high biological value and are rich in vitaminsand minerals (especially calcium, phosphorus, iron,selenium and iodine in marine products) Fish is themain source of animal protein in Bangladesh,around 63% of animal protein intake being derivedfrom ¢sh and ¢sheries products (Bangladesh Eco-nomic Review 2005)

In Bangladesh, ¢sh comes from two sources: land and marine The inland ¢shery is classi¢ed intotwo broad categories: capture (also called open) andculture (also called closed) Aquaculture plays a sig-ni¢cant and increasingly promising role in the coun-try’s total ¢sh production According to the statistics,the total output of aquaculture in 2005/2006amounted to 892 049 t, comprising 38.3% of the total

in-¢sh production, as compared with 117 025 t (15.5%)

in 1983/1984 Table 1 shows that the production ofaquaculture has grown at a higher rate than of in-land and marine capture ¢sheries The slow growth

of capture ¢sheries is mainly due to environmental

degradation by progressive physical degradation,

shrinkage and pollution of natural water bodies and

overexploitation of resources Aquaculture therefore

has a potentially major role to play in the coming

years in ful¢lling the growing demand for ¢sh

Although aquaculture has grown signi¢cantly

over the years, its full potential is yet to be realized

Sub-optimal productivity in ponds already under

cul-ture, the existence of huge areas of derelict ponds and

the dominance of extensive culture practices over

in-tensi¢ed ones all point to the existence of unutilized

potential in ¢sh farming (Alam, Jahan, Kamal,

Rah-man & Janssen 2004) Estimates show that about

24% of cultivable ponds in Bangladesh remain

un-used for this purpose (Karim et al 2006) A number

of factors including a lack of appropriate extension

approaches and technological know-how, high prices

of feed and other inputs, lack of quality ¢ngerlings

(ju-venile ¢sh) and the limited economic capital of many

farmers are thought to limit the sector’s expansion

(Alam et al 2004; Sarker, Chowdhury & Itohara

2006) The expansion of low-cost aquaculture

tech-nologies that are feasible and a¡ordable for poor

farm-ers might, therefore, have an important role to play in

satisfying subsistence needs and/or providing sources

of income for farming households in Bangladesh

Methodology and data

Analytical framework

It is believed that aquaculture can contribute to

im-proved food security and nutrition of the poor

house-holds in several ways; ¢rstly, by generating income

for the purchase of food (¢sh as a source of income);

secondly, by creating alternative

employment-gener-ating activities and increasing labour productivity;

and thirdly, by increasing available food supply and

consumption (¢sh as food) (Edwards 1999; Ahmed &Lorica 2002) The issues are examined in the studyusing the following framework (Fig 1)

One of the driving forces behind household fooddemand is income and purchasing power (Dey et al.2005) Dawson and Ti⁄n (1998) show that signi¢cantimprovements in calorie intake occur as economiesgrow Evidence suggests that as incomes rise, consu-mers reallocate their food budget away from starchystaple food items such as rice and maize towards non-staple food items such fruits, vegetables and animalproducts (¢sh and meat) (Bouis, De La Briere, Gutier-rez, Hallman, Hassan, Hels, Quabili, Quisumbing,Tilsted, Zihad & Zohir 1998; Miah 2000) The adop-tion^income hypothesis therefore assumes thatadoption of aquaculture will increase the income ofhouseholds and will thereby generate increases in

¢sh consumption

Intensi¢cation of any agricultural sector will exert

an in£uence on the overall employment in a nity Hence, adoption of aquaculture is expected toincrease the demand for family as well as hired la-bour Family labour is the most important productionfactor in developing country agriculture, and main-tenance and enhancement of labour productivity iscentral to securing and increasing income (WorldBank 1986; Zeller & Sharma 1998) A household’s abil-ity to earn income depends mainly on the nutritionalhealth of the household labour force Adoption^em-ployment linkages to food security are based on thehypothesis that adoption of aquaculture will in-crease the productivity of agricultural labour andhence engender higher earnings for both family andhired labour, which in turn will improve the nutri-tional health of the labour force

commu-The existence of adoption^consumption linkages

is hypothesized as follows: (i) adopting householdsconsume a disproportionately higher amount of ¢sh

Table 1 Aquaculture growth rate and percentage contribution to total ¢sh production from 1983/1984 to 2005/2006

Years

Fish production (metric tonnes) 

Inland capture Aquaculture Marine capture Total

Aquaculture production before 1983/1984 is not available in the statistics.

Source: Mazid (2002) and FSYB (2006).

that are rich in both protein and micro-nutrients, and

hence improvements in the nutritional status can be

achieved through adoption^home consumption

lin-kages; and (ii) adoption of aquaculture increases

con-sumption from own produce and increases market

supply, thus holding ¢sh prices down and increasing

the consumer intake of nutrient-rich food-¢sh (Bouis

et al 1998)

Study sites and data

Data that informs this paper were collected from the

research component of the DSAP, conducted by the

project monitoring and evaluation unit Funded by

the United States Agency for International

Develop-ment (USAID), DSAP was impleDevelop-mented by the

World-Fish Center in Bangladesh between 2001 and 2005

The research programme of DSAP was extended to

July 2006 with the ¢nancial support of the WorldFish

Center The project aimed at improving resource use

e⁄ciency and increasing productivity at the farm

level in a sustainable manner by the di¡usion of

low-cost aquaculture technologies Development of

Sustainable Aquaculture Project dealt with

fresh-water ¢sh culture in pond and paddy ¢elds The

spe-cies most commonly cultured in these systems are

Indian major carps and Chinese carps The DSAP

was implemented in 42 out of 64 districts in

Bangla-desh where agro-ecological and socio-economic

characteristics were suitable for the adoption of

in-land aquaculture (north, south, east and west

re-gions of the country) Altogether, 48 partner NGOs

participated in the programme and disseminated the

low-cost aquaculture technologies (DSAP 2005)

The DSAP provided 3 years of continuous trainingsupport to the farmers to enhance their knowledge ofaquaculture The farmers received 3-day-long train-ing sessions during the ¢rst year, and two and a sin-gle period of follow-up training, respectively, in theconsecutive years Formal training was complemen-ted by regular informal sessions, such as group meet-ings at the pond/plot side using the ParticipatoryAdaptive Learning (PAL) approach and annual parti-cipatory evaluation sessions The project adopted awhole family approach, whereby both husband andwife participate in training sessions

The training focused on extension of simple, cost proven pond management practices with ascienti¢c basis These included: control of predatory

low-¢shes, staggered harvesting and stocking regimes,organic and inorganic pond fertilization to increaseprimary production, aquatic weed control, betterwater management, increased regularity and consis-tency of supplementary feeding and regular observa-tion of water colour, turbidity, ¢sh behaviour and thepresence of natural feed items

For the purposes of the study, a total of 225 farmerswere selected from four DSAP working areas(Mymensingh, Comilla, Magura and Bogra) in 2002/

2003 before the project commenced These farmerswere absorbed into the project in 2003/2004 Thedata collected from these farmers during 2002/2003was benchmark information against which compari-sons in subsequent years were made to evaluate thebefore and after impact of the project The impact ofthe aquaculture intervention across project and con-trol farmers was another focus of the research pro-gramme Therefore, another set of non-projectfarmers (control farmers) was monitored from 2003/

Employment Link

ConsumptionLinkAbility to create

Higher return to capital fromaquaculture

High consumption Price effect andincreased demand

home-Adoption-IncomeLink

High income effect onnutrient rich foodconsumption

Figure 1 Framework for analysing aquaculture Linkages to food and nutritional security (adapted from Ahmed & Lorica2002)

2004 for this purpose One hundred and twenty

three such farmers, to whom no technical support

was given, were selected from the same

four working areas as a control group These control

farmers were selected from locations where no

pro-ject interventions promoting aquaculture had taken

place in the past

The farmers were selected in such a way that

farm-ers of di¡erent wealth ranks were represented A

par-ticipatory wealth-ranking exercise was undertaken

in the study sites to select the farmers for the study

The participants took land holdings of the

holds as the criteria for wealth ranking The

house-holds of the village were short-listed based on the

availability of natural resource types and their

oppor-tunities for integration with aquaculture Farmers

with an existing high-level knowledge of

aquacul-ture techniques were excluded Farmers having

 0.20 hectare (ha) of land were considered

func-tionally landless; farmers having 0.21^0.60 ha of

land were considered to be poor; farmers having

0.61^1.21ha of land were considered to be medium

farmers; and farmers having41.21ha of land were

considered to be rich

A number of surveys were conducted in the area to

provide a detailed picture of the impacts of project

tervention A whole farm monitoring survey that

in-cluded input and output information for the farm

enterprize was started in 2002/2003 In order to

cap-ture other important socio-economic changes as a

result of project intervention, livelihood and

house-hold consumption surveys were started from 2003

to 2004 In addition to these surveys, a census of

ponds was conducted in the study area in 2006 to

get an idea of the wealth status of the pond owners,

ownership patterns and technologies practiced

Dur-ing the study period, £oodDur-ing occurred in 2004/2005

and a¡ected 60% of project farmers (135 out of 225

farmers) and control farmers (73 out 126 farmers)

The ponds of a¡ected farmers were inundated by

£ood water, which led to damage of the pond dykes

and stock losses Estimates shows that on average

the a¡ected farmers received 30% less production in

2004/2005 than non-a¡ected farmers

The study was based on a ‘before-and-after,

with-and-without’experimental design Data collected from

respondents were tabulated and analysed in

accor-dance with the objective of the study Mostly

descrip-tive analyses were performed to present the ¢ndings of

the survey The one-way analysis of variance (ANOVA)

procedure was used to observe the signi¢cant change

over the years among the project and control farmers

On the other hand, t-tests were also conducted to pare the di¡erences between control and project farm-ers in a speci¢c year The Consumer Price Index (CPI)was used to adjust the income ¢gures of 2003/2004,2004/2005 and 2005/2006 to 2002/2003 prices (1US$ 5 58 Bangladesh Taka in 2002/2003)

com-Results and discussionsSample characteristicsThe average age of farmers in the study was around

40 On average, project farmers were younger (44.1years) than control farmers (47.0 years), with a highpercentage of farmers in the former category being

in the 30^45-year age brackets and the latter in the45^60-year age brackets The average number ofcompleted years of education of project farmers (6.9years) was signi¢cantly higher than among controlfarmers (5.8 years) About 6% of project farmers and4% of control farmers had completed a college degree.The average household size of six persons was similarfor both groups and was higher than that of the na-tion as a whole (4.5; source BBS 2006) The resultsshow that members of both groups had practiced ¢shculture for around 8 years; however, on average pro-ject farmers received 3 years of technical supportfrom DSAP during this 8-year period The averagefarm size of the project and control farmers wasaround 0.97 and 0.85 ha, respectively, with no signi¢-cant variation across groups The average sizes of thepond area of project and control farmers were 0.13and 0.15 ha respectively There was no signi¢cant var-iation with respect to pond size between groups.During the survey, it was observed that the major-ity of farmers cultured Indian major carps such asrohu, Labeo rohita (project farmer: 98% and controlfarmer: 97%); catla, Catla catla (project farmer: 75%and control farmer: 62%); and mrigal, Cirrhinus mri-gala (project farmer: 87% and control farmer: 85%) Asigni¢cant number of the sample farmers also cul-tured Chinese carps such as silver carp, Hypophthal-michthys molitrix (project farmer: 92% and controlfarmer: 85%); common carp, Cyprinus carpio (projectfarmer: 54% and control farmer: 59%); and grasscarp, Ctenopharyngodon idella (project farmer: 31%and control farmer: 20%) In addition, rajpunti, Pun-tius gonionotus (project farmer: 72% and controlfarmer: 59%), and tilapia, Oreochromis niloticus (pro-ject farmer: 2% and control farmer: 2%) were alsocultured by a number of sample farmers

Impacts of aquaculture at the farm and

household level

The extent to which adoption of improved

aquacul-ture techniques can be considered to have impacted

on food security at the farm and the household level

among project and non-project (control) participants

is examined in the following sections in terms of

in-come, employment and consumption, using data

re-turned from the project’s monitoring activities

Income e¡ects

Figures 2 and 3 show that after the project

interven-tion the gross income of the project households grew

from Tk 96 423 in 2002/2003 to Tk 127 276 in 2005/

2006 with an average growth rate of 8.1% year 1 On

the other hand, control farmers experienced

moder-ate increases in income with an average growth rmoder-ate

of 0.9% year 1from Tk 89 478 in 2003/2004 to Tk

90736 in 2005/2006 The di¡erence is mainly

ac-counted for by increases in farm and ¢sh income as

no signi¢cant income increases were observed

be-tween the project and the control farms from o¡-farm

and non-farm sources and remittances The results

show that the contribution to incomes from ¢sh

cul-ture increased substantially among project farmers

after the project intervention The gross income from

¢sh production among participating households

in-creased from Tk 6644 in 2002/2003 (before project

intervention) toTk.12809 in 2005/2006 (after project

intervention) This equates to an increase from 14% to

17% of the total farm income and 7^10% of the totalhousehold income respectively The sudden decline inincome in 2004/2005 was mainly due to £ooding,which led to major crop losses that severely a¡ectedlarge numbers of both project and control farmers.The adoption of improved ¢sh culture practiceshas o¡ered the possibility of generating higher re-turns than those derived from traditional practices.Table 2 shows that the net annual income (gross in-come less total operating and input costs required forcultivation) from ¢sh culture among project farmerswas Tk 8047 per household after the project inter-vention in 2005/2006, which is signi¢cantly higherthan the net income of ¢sh culture of control farmers,which is Tk 3586 per household Improved aquacul-ture technology with a more systematic use of feedand protein supplements (regularity and timing offeeding), recommended stocking and harvestingpractices gained a greater return from aquacultureoperations than non-project farmers who continued

to follow traditional techniques based on using ple stocking and harvesting practices The study alsorevealed that net income from ¢sh culture amongproject and control farmers, although lower thanthat from cereals, was higher than for all other farm-ing activities bearing comparable risks such as vege-tables poultry and livestock

sim-Employment e¡ectsTable 3 shows that, with the exception of the year2003/2004, the amount of total labour (family and

income Income from

fish

Non -farm income

Remittances

Income sources

2002/032005/06

Figure 2 Household income by sources of the project farmers over the years

hired) employed is signi¢cantly higher on project

farms than on control farms The average growth of

labour use on project farms increased at a higher rate

over the years than that of control farms This can

be explained by the improved technologies used by

project farms that require that more time is spent on

implementing improved management practices than

on control farms This evidence tends to support

stu-dies that ¢nd that although the labour e¡ort required

for aquaculture is lower than that of other farmingactivities (the present study showed that only 10% offarm labour is utilized for aquaculture), adoption ofintensi¢ed culture practices has the potential to in-crease the total labour requirement to some extent(Ahmed, Abdur Rab & Bimbao 1995; Dey 2000;Ahmed & Lorica 2002)

Another direct way in which the poor stand to e¢t from adoption of aquaculture is by improving re-

fish

Off - farm income

Non - farm income

Remittances

Income sources

2003/042005/06

Figure 3 Household income by sources of the control farmers over the years

Table 2 Income from di¡erent farming enterprises in 2005/2006

Enterprises Items (Tk per household) Project farmer (n 5 225) Control farmer (n 5 123)

Mean difference test between project and control farmers

turns to labour in terms of physical and monetary

units The study shows that the performance of the

project farmers in terms of labour productivity and

returns to family labour improved signi¢cantly after

the project intervention and it is signi¢cantly higher

than for control farmers, with the exception of labour

productivity in the year 2004/2005 During the

pro-ject period, the return on family labour grew at an

average rate of 15.6% year 1 from Tk 314

man-day 1in 2002/2003 to Tk 476 man-day 1in 2005/

2006 Control farmers experienced moderate

in-creases in income with an average growth rate of

8.3% year 1from Tk 193 to Tk 230 man-day 1 The

di¡erence was mainly due to improved management

practices and project technical support, which

helped the farmers use inputs and labour in a more

e⁄cient manner than control farmers The ¢ndings

also suggest that although the contribution of hired

labour is small for most aquaculture operations, poor

people with no access to land may potentially gain

some bene¢ts from aquaculture through increased

wage rates

Another important aspect of employment e¡ects is

the total occupational shift in households following a

rapid development of aquaculture As many as 4% of

project farmers reported aquaculture to be their

pri-mary activity before the project intervention,

in-creasing to 10% after the project intervention Other

studies such as MAEP (1995) and Miah (2000) have

also documented a similar occupational shift and

ar-gued that modern aquaculture technologies tend tofoster more commercial attitudes towards ¢sh cul-ture and generate higher earnings than traditionalculture technologies Aquaculture also produces aseries of backward linkages (such as hatcheries,nurseries and seed, feed and input deliveries) andforward linkages (such as post-harvest handling,processing and marketing), all of which are signi¢-cant absorbers of labour (Lewis, Gregory & Wood1996; Faruque 2007) However, the net e¡ects ofsuch transformation on local employment may notalways be readily discernible, and require moredetailed analysis

Consumption e¡ectsThe results for consumption data show that theannual per capita ¢sh consumption of project house-holds increased at a rate of 6.6% from 1.50 kgcapita 1month 1in 2003/2004 to 1.79 kg capita 1month 1in 2005/2006, the exceeding national aver-age per capita ¢sh consumption of 0.95 kg capita 1month 1(Bangladesh Economic Review 2005) Dur-ing the same period, the per capita monthly consump-tion of ¢sh of control households increased at a rate of2.3% per annum from 1.29 kg capita 1month 1in2003/2004 to 1.38 kg capita 1 month 1 in 2005/

2006, which was also higher than the average percapita national consumption The results support the

Table 3 Employment e¡ect for adoption of aquaculture farms over the years

Items

Project farmer (n 5 225) Control farmer (n 5 123)

Mean difference test between project and control farmer 2002/

2003 2003/

2004 2004/

2005 2005/

2006 Average growth (%)

2003/

2004 2004/

2005 2005/

2006 Average growth (%)

2003/

2004 2004/

2005 2005/ 2006

Family labour

wLabour productivity is calculated as the total production divided by the total family labour in person days.

zReturn to family labour is calculated as the gross return less ¢xed cost (opportunity cost of capita and depreciation) divided by the total family labour in person days.

argument of Dey (2000) and Ahmed et al (1995)

that per capita consumption of ¢sh in rural areas is

noticeably higher for producing households than for

non-producers

The study also reveals that consumption of staple

foods such as cereals increased at a rate of 0.6%

per annum for project households (15.52 kg capita 1

month 1 in 2003/2004 to 15.78 kg capita 1

month 1in 2005/2006) and1.5% for the non-project

households (15.05 kg capita 1month 1 in 2003/

2004 to 15.13 kg capita 1month 1in 2005/2006)

A higher price elasticity and income elasticity for ¢sh

compared with rice documented by other studies

(Ahmed & Shams 1994; Alam 1999; Dey 2000; Dey

et al 2005) in rural Bangladesh may account for the

higher ¢sh consumption among the project farmers

after the intervention The ¢ndings of the study

re-vealed that during the study period, ¢sh prices

in-creased at a rate of 4%, while at the same time rice

prices increased at a rate of 17%

Table 4 demonstrates that increases in

productiv-ity per hectare result in a larger proportion of ¢sh

being sold, with the quantity disposed in the market

increasing at a rate of 28.8% and 4.5% for project and

control farmers respectively However, although ¢sh

consumption decreased as a proportion of total

household ¢sh production, larger absolute volumes

of ¢sh were consumed, with consumption increasing

at a rate of 9.9% and 1.3% for the project and control

households respectively The results showed that the

total consumption of self-produced ¢sh in project

households was signi¢cantly higher than that of

con-trol households The distribution of ¢sh among theneighbours and relatives, particularly during festi-vals, is a cultural tradition of the Bengali community.Evidence showed that the gifting of ¢sh to neigh-bours and relatives increased signi¢cantly at a rate

of 22.8% among the project farmers over the years.This indicates that increases in ¢sh productivitymay also help to strengthen social relations to somedegree This type of non-income- and consumption-related e¡ect may make the adoption of small-scaleaquaculture attractive even where ¢nancial or nutri-tional returns are modest (Haque 2007) but is fre-quently overlooked in impact evaluations

An analysis of the ¢sh species consumed by thetwo groups of farmers (including self-produced, pur-chased and wild-caught ¢sh) shows that the majorportion of the species consumed are cultured speciessuch as Indian major carps, Chinese carps and exotic

¢shes, particularly silver barb and tilapia (Table 5).This indicates the critical contribution of ¢shfarming to household food security in areas amongadopting households, further demonstrating the im-portance of on-farm availability of ¢sh Approxi-mately 65% and 70% of the total ¢sh consumptionrequirements of the project and non-project house-holds’ were ful¢lled by their own production It maythus be concluded that ¢sh supplies from familyfarms will increasingly ful¢l the consumption needs

as on-farm ¢sh production increases, but that creasing incomes from sales of ¢sh will facilitate thepurchase of increased quantities of ¢sh from othersources

in-Table 4 Production and disposal Patterns for harvested ¢sh among project and control farmers

Items

Project farmer (n 5 225) Control farmer (n 5 123)

Mean difference test between project and control farmer

2002/

2003

2003/

2004 2004/

2005 2005/

2006

Average growth (%) 2003/

2004 2004/

2005 2005/

2006

Average growth (%) 2003/

2004 2004/

2005 2005/ 2006