Harmful Algae 2 (2003) 273–281

Ichthyotoxicity ofChattonella marina (Raphidophyceae) todamselfish (Acanthochromis polycanthus): the synergistic

role of reactive oxygen species and free fatty acids

Judith-Anne Marshalla,∗, Peter D. Nicholsb, Brett Hamiltonc,Richard J. Lewisc, Gustaaf M. Hallegraeffa

a School of Plant Science, University of Tasmania, G.P.O. Box 252-55, Hobart, Tasmania 7001, Australiab CSIRO Marine Research, G.P.O. Box 1538, Hobart, Tasmania 7001, Australia

c Institute for Molecular Bioscience, The University of Queensland, St. Lucia, Brisbane, Qld 4072, Australia

Received 18 December 2002; received in revised form 18 March 2003; accepted 28 March 2003

Abstract

This investigation aimed to elucidate the relative roles of putative brevetoxins, reactive oxygen species and free fatty acids asthe toxic principle of the raphidophyteChattonella marina, using damselfish as the bioassay. Our investigations on AustralianC. marina demonstrated an absence or only very low concentrations of brevetoxin-like compounds by radio-receptor bindingassay and liquid chromatography–mass spectroscopy techniques.Chattonella is unique in its ability to produce levels ofreactive oxygen species 100 times higher than most other algal species. However, high levels of superoxide on their ownwere found not to cause fish mortalities. Lipid analysis revealed this raphidophyte to contain high concentrations of thepolyunsaturated fatty acid eicosapentaenoic acid (EPA; 18–23% of fatty acids), which has demonstrated toxic propertiesto marine organisms. Using damselfish as a model organism, we demonstrated that the free fatty acid (FFA) form of EPAproduced a mortality and fish behavioural response similar to fish exposed toC. marina cells. This effect was not apparentwhen fish were exposed to other lipid fractions including a triglyceride containing fish oil, docosahexaenoate-enriched ethylester, or pure brevetoxin standards. The presence of superoxide together with low concentrations of EPA accelerated fishmortality rate threefold. We conclude that the enhancement of ichthyotoxicity of EPA in the presence of superoxide canaccount for the highC. marina fish killing potential.© 2003 Elsevier B.V. All rights reserved.

Keywords: Ichthyotoxicity; Reactive oxygen species; Free fatty acids;Chattonella marina; Superoxide; Eicosapentaenoic acid

1. Introduction

Several hypotheses have been proposed for thetoxic mechanism of the fish killing raphidophyteChattonella marina. The original theory put for-

∗ Corresponding author. Tel.:+61-3-62261750;fax: +61-3-62262698.

E-mail address: judi.marshall@utas.edu.au (J.-A. Marshall).

ward by Okaichi (1983)was the production of freefatty acids (FFAs) byChattonella entering the bloodstream via the gills, resulting in fish mortalities.Subsequent studies have investigated reactive oxy-gen species (ROS) production and its role in gilldamage (Shimada et al., 1983; Oda et al., 1992a,b).Later investigations have centred on anoxia, mucusproduction, respiratory, ionoregulatory and cardio-vascular physiology (Ishimatsu et al., 1990, 1991,

1568-9883/$ – see front matter © 2003 Elsevier B.V. All rights reserved.doi:10.1016/S1568-9883(03)00046-5

274 J.-A. Marshall et al. / Harmful Algae 2 (2003) 273–281

1997). C. marina has also been claimed to producea fat-soluble toxin similar in structure to brevetoxin,first characterised from the dinoflagellateKareniabrevis (Onoue et al., 1990; Khan et al., 1995), but theraphidophyte toxin has yet to be fully characterisedchemically. The ROS superoxide (•O2

−), which isproduced byC. marina at levels 100-fold higherthan that of most other alga (Marshall et al., 2002a),has also been implicated in fish mortalities throughchanges in gill pathology (Ishimatsu et al., 1997).Ecophenotypic differences in•O2

− production be-tween Australian and Japanese strains have shown thathigher•O2

− production inC. marina correlates withfaster mortality in juvenile rainbow trout (Onchor-rynchus mykiss) (Marshall et al., 2001). Furthermore,•O2

− is primarily controlled through photosynthesis(Marshall et al., 2002a). However, when superoxidelevels were suppressed in dark treatments, fish mor-talities did not occur. This led to the hypothesis thata synergy between ROS and a toxic fraction couldbe responsible for fish mortalities (Marshall et al.,2001).

C. marina has been found to produce high lev-els (10%) of FFAs, which have also been impli-cated in Chattonella fish toxicity (Okaichi et al.,1989). Both Australian and Japanese strains con-tain high levels of potentially toxic polyunsaturatedfatty acids (PUFAs) including eicosapentaenoic acid(EPA, 20:5�3, 18–23%; Marshall et al., 2002b).Arzul et al. (1995,1998)demonstrated that FFA in-cluding steraradonic acid (STA, 18:4�3) and EPAfrom the dinoflagellateKarenia mikimotoi havean allelopathic effect on the diatomChaetocerosgracile, and can result in haemolysis of horse redblood cells and mortality ofPecten maximus shell-fish larvae. The present investigation aimed to elu-cidate the relative roles of putative brevetoxins,ROS and FFAs as the ichthyotoxic principle ofC. marina.

2. Methods

2.1. Algal culturing

A non-axenic culture ofC. marina (UTCMPL03)isolated from Port Lincoln, South Australia, in 2002was maintained in full strength seawater containing

GSe nutrients (Blackburn et al., 1989) at 25◦C and150�mol m−2 s−1 light. A culture of Dunaliella ter-tiolecta (CS-175) was used as an algal control inthe fish exposure experiments. Experimental cultureswere kept in exponential phase through the use ofsemi-continuous culture techniques to ensure nutrientlimitation did not occur.

2.2. Fish bioassay

Damselfish (Acanthochromis polycanthus Bleeker)weighing 600–1200 mg were obtained from the Uni-versity of Tasmania, School of Aquaculture, as threeheterogeneous batches. Fish were exposed to early ex-ponential stage cultures ofC. marina at concentrationsof 250–35,000 cells ml−1, which were adjusted to cellconcentrations to be tested 48 h prior to fish challengeto allow equilibrium of ROS levels. The experimen-tal cultures were not aerated and oxygen levels weremeasured using a WTW Oxyguard probe. Pure FFA(EPA and STA) were obtained from Sigma ChemicalCo. (St. Louis, MO). Other lipids used were an ethylester-enriched 22:6�3 (DHA) (66% of fatty acid frac-tion with 20% EPA) and tuna oil containing >99.5%tryglyceride (with 25% DHA and 8% EPA). Lipidswere dissolved in 2 ml of methanol and introduced to400 ml of seawater containing four damselfish at con-centrations of 2–25 mg l−1.

2.3. Reactive oxygen species

The oxygen radical superoxide (•O2−) was meas-

ured using the luciferin analogue 2-methyl-6-(p-meth-oxyphenyl)-3,7-dihydroimidazo[1,2-a]pyrazin-3-one(MCLA) as previously described byOda et al.(1992a,b) at a concentration of 5× 10−6 M andstandardised against 5× 10−6 M superoxide dismu-tase (SOD). All results were corrected for controlsconsisting of GSe media. Superoxide was generatedthough xanthine oxidase acting aerobically on a xan-thine substrate resulting in the production of uricacid (Halliwell and Gutteridge, 1999) as shown inFig. 1. Superoxide levels were maintained by period-ically adding 5× 10−6 M xanthine to 10–30 units l−1

of xanthine oxidase in filtered seawater, determinedto be equivalent to the production of superox-ide by C. marina and checked by MCLA lumine-scence.

J.-A. Marshall et al. / Harmful Algae 2 (2003) 273–281 275

Xanthine (substrate)

2H2O + 2O2

H2O2 + 2O2-

Uric Acid (Keto form)

Xanthine Oxidase (enzyme)

Fig. 1. Xanthine catalyses oxidation of xanthine to uric acid whilstreducing O2 to both superoxide (O2

−) and hydrogen peroxide(H2O2).

2.4. Toxin analysis

Chattonella cells were centrifuged, extracted indichloromethane, evaporated and partitioned betweenmethanol and hexane to remove fats and selected pig-ments, e.g. carotenoids. The residue for the methanolfraction was fractionated using silica flash chro-matography as per Lewis et al. (1991) with chloro-form/methanol (c:m) solutions of increasing polarity(1:0, 9:1, 7:3, 1:1, 2:8, 0:1). Fractions were assessedusing a tritium labelled brevetoxin radio-labelled lig-and binding (RLB) assay (Poli et al., 1986; Hamiltonet al., 2002) at 0.5, 0.05, 0.005, 0.0005, 0.00005 and0.000005%. Samples were further analysed using agradient of 2% B per min (A: 0.1% formic acid (aq);B: 90% acetonitrile (aq) + 0.1% formic acid) estab-lished on an Agilent 1100 series high-performanceliquid chromatography system coupled to a SCIEXQstar (Quadrupole/time-of-flight) mass spectrometer(LC/MS). The Institute of Medical and VeterinaryScience, South Australia, conducted mouse bioassaysusing IVMS method FH38. Duplicate mice were in-jected intraperitoneally with the equivalent of 3 × 106

cells of C. marina from Australia and Japan (strainNIES-118), extracted in methanol and suspended in10% Tween solution.

2.5. Statistical analysis

Results were analysed using a simple regressionwith a significance level of 0.05.

Toxicity of brevetoxins (PbTx-2 and PbTx-9) andFFA was calculated from the survival time of the fishand expressed as LT50, the time it takes for 50% offish to die at the prescribed concentration of toxin oralgal cells.

3. Results

3.1. C. marina exposure and superoxide production

Damselfish exposed to non-aerated cultures of C.marina less than 1000 cells ml−1 rarely exhibitedmortality. Fish exposed to algal concentrations of1000–8000 cells ml−1 suffered more rapid mortalitytimes (89 ± 6 min) than fish exposed to cell concen-trations above 8000 cells ml−1 (143 ± 8 min; Fig. 2).A regression (ANOVA) model for Chattonella densi-ties above 1000 cells ml−1 suggests that fish mortalitytime is inversely correlated with cell density (P =0.44) and closely correlated with superoxide levels(P = 0.67). Total superoxide levels were maximal inalgal cultures with higher cell densities of C. marina,but the fastest fish mortality time correlated with lowcell densities of 1000–8000 cells ml−1 and was asso-ciated with levels of •O2

− production per cell greaterthan 40 chemiluminescence units (Fig. 3).

No fish mortalities occurred during the 6 h experi-ments in aerated cultures. Dissolved oxygen (DO) lev-els in non-aerated experiments were maintained abovethat of the non-aerated control by oxygen releasedthrough photosynthesis of C. marina cells. The higher

R2 = 0.67

0

50

100

150

200

0 10000 20000 30000Chattonella cell concentration

(per ml)

LT

50 (

min

s)

Fig. 2. Time to death of 50% of experimental fish (LT50) as afunction of cell density of C. marina, which was maintained undersemi-continuous culture conditions in two separate experiments(n = 8 per treatment). Only one mortality was observed at celldensities <1000 cells ml−1 (n = 20) over a 3 h period.

276 J.-A. Marshall et al. / Harmful Algae 2 (2003) 273–281

020406080

100120140160180200

50 100 150 200 250 300Mean mortality time (mins)

Che

milu

min

esen

cse

per

cell <1000 cells

1000-8000 cells

>8000 cells

Fig. 3. Levels of superoxide per cell (determined by chemilumi-nescence) compared to mortality time of damselfish exposed toC. marina cultures. There is a clear distinction between low celldensity (1000–8000 cells ml−1) and rapid mortality time comparedto high cell density (>8000 cells ml−1) and longer mortality time.Only one mortality occurred at <1000 cells ml−1 (n = 20).

C. marina density resulted in higher DO (r2 = 0.83;data not shown). C. marina in the presence of fishproduced significantly higher •O2

− than C. marinanot exposed to fish, regardless of whether the cultureswere aerated (Fig. 4). Fish exposed to low concen-trations of C. marina (1000–8000 cells ml−1) rapidly

20

30

40

50

60

0 50 100 150 200

Che

milu

min

esce

nce

Uni

ts

(x 1

04 )

Aerated C. marina + Fish

Non-aerated C. marina + Fish

Aerated

Non-aerated C. marina

C. marina

C. marina

C. marina

C. marina

Time (mins)

Fig. 4. The effect of aeration on total superoxide production ofC. marina cultures as determined by chemiluminescence, in thepresence and absence of damselfish. Open symbols denote aeratedcultures and closed symbols denote non-aerated cultures. Diamondsdenote the presence of damselfish and triangles denote no fishpresent.

developed symptoms of hyperventilation, mucus ex-cretion, vasodilation of the gills (as seen through theopercular cover), followed by an inability to maintainposition in the water column. Pre-mortality responseincluded gulping at the water surface before sinkingand lying laterally prior to respiratory cessation.

3.2. Xanthine oxidase generatedsuperoxide exposure

Superoxide was produced and maintained throughperiodic additions of the substrate xanthine to the en-zyme xanthine oxidase for a period of 3 h. Levels of•O2

− produced, observed every 15 min did not deviatesignificantly from those assessed in the Chattonellacontrol of 6000 cells ml−1 (Fig. 5). Fish exposed to thexanthine oxidase generated •O2

− did not display stressbehaviour for the first 150 min when compared to thefiltered seawater, the non-toxic Dunalliella and Chat-tonella controls. After this time, the fish appeared slug-gish until termination of the experiment at 180 min.No fish mortalities resulted in the xanthine oxidasecontrol trial, nor were any behavioural responses cor-responding to Chattonella toxicity noted.

3.3. Brevetoxin analysis

The RLB analysis did not indicate the presence ofPbTx-like material in any of the fractions. The LC/MSexperiments did not detect any of the major PbTxs(PbTx-1, -2, -3, -7, -9, -10). In the 7CHCl3:1MeOHsilica fraction, we observed: (1) [M + Na]+ at m/z901.50, [M + NH4]+ at m/z 896.54, [M + H]+ at m/z897.50, and [M+H–H2O]+ at m/z 861.50 at 29.1 min;(2) [M + Na]+ at m/z 843.47, [M + NH4]+ at m/z838.47, [M +H]+ at m/z 821.45, and [M +H–H2O]+at m/z 803.45 at 27.8 min; and (3) [M + Na]+ at m/z899.48, [M + NH4]+ at m/z 894.55, and [M + H]+ atm/z 877.51 at 29.6 min. These chemical species werepresent at very low levels and the masses do not cor-respond to any known PbTxs. The residual masses(0.45–0.55 Da), a function of the relative number ofmass deficient elements such as oxygen present ineach of the molecules, indicate that they share a verysimilar elemental composition to all known polyethermarine toxins and perhaps are structurally similar.

Mouse bioassays on lipophilic extracts of bothAustralian and Japanese strains of C. marina did not

J.-A. Marshall et al. / Harmful Algae 2 (2003) 273–281 277

0

10

20

30

40

50

60

70

80

90

0 20 40 60 80

Time (mins)

Che

milu

min

esen

ce U

nits

(x10

4 )

Chattonella

Xanthine Oxidase + Xanthine

Xanthine Oxidase + Xanthine every 15 mins

Chattonella

Fig. 5. Production of superoxide generated from the xanthine oxidase system compared to superoxide production by Chattonella marinaat 6000 cells ml−1. A single application of xanthine shows a gradual reduction in superoxide levels, however regular additions of xanthinemaintained high levels of superoxide.

result in mouse mortality (T = 24 h). There was novisible evidence of mouse liver abnormality or dam-age on post mortem examination. Fish exposed topure PbTx-2 and -9 standards at 25 �g l−1 showedno signs of respiratory distress within the first 6 h,but did display uncoordinated swimming behaviour(swimming laterally or up-side down) as describedby Lewis (1992).

3.4. Free fatty acid exposure

Exposure of damselfish to pure EPA resulted in aLC50 of 2.7 mg l−1 (Fig. 6), with mortality of 50%of fish occurring at 155 min. Only 25% of fish diedwhen exposed to 25 mg l−1 of STA, but all fish showedsymptoms of being adversely impacted by a toxicsubstance. Lethal and sub-lethal doses of EPA andSTA resulted in the fish showing pronounced opercu-lar movement (suggesting respiratory distress), pro-duction of mucus, inactivity, an inability to maintainposition in the water column and a loss of rightingreflex. The gill region appeared red, suggestive of va-sodilation. No symptomatic fish behaviour or mortal-ities occurred when the fish were challenged with theesterified fatty acids in TAG or the ethyl ester form,the latter containing 66% DHA (22:6�3).

Fish exposed to superoxide from xanthine oxidasessystem in conjunction with low levels of EPA (2 and4 mg l−1) displayed similar behavioural toxic symp-toms to those fish exposed to Chattonella, but ex-hibited significantly accelerated mortality rates whencompared to fish exposed to EPA on its own (Table 1;Fig. 7a and b).

y = -1.77x2 + 28.90x - 10.45

R2 = 0.94

0

20

40

60

80

100

0 2 4 6 8 10

EPA concentration (mg per L)

% M

orta

lity

Fig. 6. Toxicity of eicosapentaenoic acid (EPA) to damselfishcalculated as percentage mortality. The fitted polynomial curve ofy = −1.77x2 + 28.90x − 10.45 (r2 = 0.94) was used to calculatean LD50 of 2.7 mg l−1 EPA.

278 J.-A. Marshall et al. / Harmful Algae 2 (2003) 273–281

Table 1The effect on damselfish exposed to sub-lethal (2 mg l−1) and lethal (4 mg l−1) concentrations of the free fatty acid form of eicosapentaenoicacid (EPA) in the presence or absence of superoxide, expressed as time to death (min)

EPA treatment (mg l−1) Time to death (min)(EPA only)

Time to death (min)(EPA + superoxide)

EPA equivalent(mg l−1)

2 131 ± 10 (n = 8) 51 ± 7 (n = 3) 4.94 82 ± 13 (n = 8) 44 ± 5 (n = 3) 5.8

Xanthine oxidase 0 (n = 3) – –

The EPA equivalent is calculated from the equation fitted to EPA exposure data shown in Fig. 7a.

y = 213.64e-0.18x

R2 = 0.87

0

20

40

60

80

100

120

140

160

180

200

0 2 4 6 8 10

EPA (mg L-1)

Tim

e to

dea

th (

min

s)

y = 58.79e-0.07x

R2 = 1.00

0

2040

6080

100

120140

160180

200

0 2 4 6 8 10

EPA (mg l-1)

Tim

e to

dea

th (

min

s)

EPA

EPA + SO

(a)

(b)

Fig. 7. Toxicity of eicosapentaenoic acid (EPA) to damselfish.(a) The fitted exponential curve produced the response equationy = 213.6e−0.18x (r2 = 0.87). (b) Combined data from exposureof sub-lethal (2 mg l−1) and lethal (4 mg l−1) concentrations ofthe free fatty acid form of EPA, in the presence of xanthinegenerated superoxide, was fitted with a hypothetical exponentialcurve (y = 58.8e−0.07x) to predict the effect of low concentrationsof EPA in the presence of superoxide.

4. Discussion

4.1. Brevetoxin-like compounds

Two independent investigations have failed todetect the presence of significant quantities of fat-soluble brevetoxins in Australian strains of C. ma-rina. Hallegraeff et al. (1998) detected only tracelevels of brevetoxin-like compounds in an Australianstrain (0.006–0.03 fg per cell), using brevetoxin radio-receptor assays. In the present work, none of the ma-jor PbTxs were identified in an extract by either RLBor LC/MS. The RLB did not suggest the presenceof any site 5 sodium channel toxin activator toxins,although it is possible that small amounts of lowerpotency toxins were present. Similarly, an intraperi-toneal mouse bioassay produced negative results forneurotoxicity or liver damage for both the Australianor Japanese (NIES-118) strains. We therefore con-clude that no PbTx-like compounds are detectable atsignificant levels in the Australian strain of C. marina.This is supported by the observation that exposure ofdamselfish to pure PTx-2 resulted in fish behaviourcomparable to that described by Lewis (1992). Bycontrast, fish behaviour when exposed to either C.marina cells or the FFA—EPA and STA—was char-acterised by markedly different symptomology.

4.2. Superoxide

If C. marina toxicity would be solely a function ofa neurotoxin, one would expect a linear increase infish toxicity with increasing Chattonella cell density,rather than the inverse relationship shown by this study(Fig. 2). These results have been previously repli-cated for the salmonid Onchorynchus mykiss (Mar-shall, unpublished data) and support the hypothesisof a synergistic effect of ROS with some other toxic

J.-A. Marshall et al. / Harmful Algae 2 (2003) 273–281 279

principle. Superoxide has been claimed to result indamage to the gill epithelium (Ishimatsu et al., 1997;Shimada et al., 1983), causing subsequent respiratoryproblems. Such toxic insult is thought to be localised atthe gill membrane, where contact occurs between thegill epithelia and the Chattonella cells. If this were thecase, the greater production of •O2

− per cell shouldproduce more fish gill lamellae damage. In the presentwork, cell densities of less than 1000 cells ml−1 didnot appear to provide toxicity, possibly due to lowerlevels of algal cell to gill contact. Previous studieshave also reported no lethality of Chattonella culturesat cell densities lower than 103 cells ml−1 (Haque andOnoue, 2002, 2 × 103 cells ml−1; Khan et al., 1996).Temperature also affects lethal cell density for C. ma-

Chattonella cells

Gill Impact

Chattonella cell rupture

ROS FFA Lipid

peroxidation

Hyperventilation

Gill epithelium damage Reduction of PaO2 Osmoregulatory

dysfunction

Fish Death

Fig. 8. Diagram summarising possible mechanisms of fish mortality when exposed to Chattonella derived FFA in combination with theROS superoxide: Chattonella cells are brought into contact with the fish gill lamellae through ventilation. Cell contents released frombroken Chattonella cells hydrolyse in high ROS environment to produce higher levels of FFA and ROS. Lipid peroxidation occurs on gilland other membranes resulting in reduced respiratory and osmoregulatory capacity and allowing the transfer of FFA and O2

− into theblood stream. Damage to the chloride cells of the gills can also lead to reduced osmoregulatory capacity. Toxic mechanisms may occurin isolation or combination.

rina, with higher temperatures apparently decreasingthe toxic cell density threshold (Okaichi et al., 1989).Superoxide production is inversely related to cell den-sity, possibly to maintain a minimum ambient envi-ronmental level of ROS as an allelopathic protection.

4.3. Dissolved oxygen

DO self-evidently plays a role in the survivorshipof the damselfish. The alleviation of respiratory dis-tress may be instrumental in reducing the toxicity ofC. marina. Hyperventilation during periods of lowDO would increase ventilation volume and velocityof C. marina cells crossing the gill membranes, en-hancing the incidence of gill contact with C. marina

280 J.-A. Marshall et al. / Harmful Algae 2 (2003) 273–281

cells. Aeration also prevented fish mortalities fromthe dinoflagellate K. mikimotoi, but not for Gymno-dinium cf. maguelonnense (Jenkinson and Arzul,2001). Arzul et al. (1998) postulated that increasedoxygen availability may be offset by greater free rad-ical mediated cytotoxicity associated with haemolyticactivity. In the present study, no significant increasein •O2

− production resulted from aeration of algalcultures, but significantly higher •O2

− was detectedin the presence of fish, indicating that the presence offish, not aeration, stimulates the greater •O2

− produc-tion by the cells, which may lead to more cytosolicactivity. Lectins found in fish mucus have previouslybeen found to stimulate •O2

− production in C. marina(Oda et al., 1998).

4.4. Free fatty acids

We found that PUFAs in the FFA form play apredominant role in Chattonella ichthyotoxicity. TheLC50 for EPA in the FFA form was calculated fromthe fitted polynomial curve (Fig. 6) as 2.7 mg l−1 fordamselfish. Esterified PUFA, such as TAG or an ethylester enrichment of DHA (both also containing EPA),are not toxic to fish. Marshall et al. (2002b) foundthat levels of EPA in C. marina under optimal growthconditions are 1.5–2 mg l−1 and that around 10% ofthe algal lipid is in the FFA form. By extrapolatingthe data for the toxicity of FFAs in the presence of•O2

−, using an exponential model (Fig. 7b), we pos-tulate that small amounts of FFAs can become toxicto fish. Our model predicts that 0.2 mg l−1 EPA in thepresence of superoxide provides a LT50 of 83 min,which is equivalent to either 4 mg l−1 of EPA oraround 1000 cells ml−1 of Chattonella culture. Bothconcentrations are toxic to fish.

The rupture of fragile Chattonella cells may leadto a rapid increase in EPA concentration in the freeform in the fish buccal cavity, exacerbated by fish hy-perventilation. The liberation of the free form of EPAfrom cell lysis has been demonstrated with epilithicdiatom biofilms (Jüttner, 2001). Free EPA from lyseddiatom cells were shown to provide an allelopathicrole against predator grazers, with an LC50 for zoo-plankton at 34 �M (=10 mg l−1). EPA has also beenshown to provide an allelopathic protection of di-noflagellates against diatoms at 1.5 mg l−1 (Arzulet al., 1998). FFAs in combination with high levels of

ROS may lead to the peroxidation of the biologicalmembranes (Halliwell and Gutteridge, 1999). Thepresence of FFAs with ROS damaged gill membranesmay result in fish mortality through a number ofcauses: (a) breakdown of gill membranes resulting inreduced respiratory capacity; (b) absorption of FFA orsuperoxide into the blood stream resulting in reducedblood pH, leading to suppressed respiratory and/orosmoregulatory capacity; and/or (c) destruction ofthe chloride cells leading to reduced osmoregulatorycapacity (Fig. 8). Fish mortality from C. marina cellsis most likely due to a combination of all these fac-tors. The above fish killing mechanism is believed tobe a common factor in all C. marina strains, whichproduce high levels of ROS and contain abundantlevels of PUFA such as EPA, regardless of brevetoxinconcentration. Further research is in progress to moreprecisely determine the minimum toxic concentra-tions of FFA and ROS and elucidate the respiratoryand cardiovascular fish pathology associated withfatty acid toxicity.

Acknowledgements

We thank the following people from the Univer-sity of Tasmania; Quinn Fitzgibbon and Prof. NedPankhurst from the School of Aquaculture for theculture of the damsel fish, Dr. Barry Munday of theSchool of Biomedical Science for advice on fishpathology, Dr. Tom Ross from the School of Agri-cultural Science for help with the xanthine oxidasesystem and models, Andrew Pankowski from IASOSfor helpful discussion on oxygen radical chemistry,Helen Bond of the School of Plant Science for cul-turing assistance. We also thank the Institute of Med-ical and Veterinary Science, South Australia for themouse bioassays. This work was supported by theAustralian Research Council; Grant No. A00106134and by the ARC Special Research Centre for AppliedGenomics (University of Queensland). Experimentswere conducted under the auspices of the Universityof Tasmania ethics approval No. A0006359.

References

Arzul, G., Gentien, P., Bodennec, G., Toularastel, F., Youenou,A., Crassous, M.P., 1995. Comparison of toxic effects of

J.-A. Marshall et al. / Harmful Algae 2 (2003) 273–281 281

Gymnodinium cf. nagasakiense polyunsaturated fatty acids.In: Lassus, P., Arzul, G., Erard, E., Gentien, P., Marcaillou,C. (Eds.), Harmful Marine Algal Blooms. Lavoisier, Paris,pp. 395–400.

Arzul, G., Bodennec, G., Gentien, P., Bornens, P., Crassous, M.-P.,1998. The effect of dissolved oxygen on the haemolytic propertyof Gymnodinium ichthyotoxins. In: Reguera, B., Blanco, J.,Fernandez, M.L., Wyatt, T. (Eds.), Harmful Algae. Xunta deGalicia and Intergovernmental Oceanographic Commission ofUNESCO, Vigo, pp. 611–614.

Blackburn, S.I., Hallegraeff, G.M., Bolch, C.J., 1989. Vegetativereproduction and sexual life cycle of the toxic dinoflagellateGymnodinium catenatum from Tasmania, Australia. J. Phycol.25, 577–590.

Hallegraeff, G.M., Munday, B.L., Baden D.G., Whitney, P.L.,1998. Chattonella marina raphidophyte bloom associated withmortality of cultured bluefin tuna (Thunnus maccoyii) in SouthAustralia In: Reguera, B., Blanco, J., Fernandez, M.L., Wyatt, T.(Eds.), Harmful Algae. Xunta de Galicia and IOC of UNESCO,Vigo, pp. 93–96.

Halliwell, B., Gutteridge, J.M.C., 1999. Free Radicals in Biologyand Medicine. Oxford University Press, London.

Hamilton, B., Hurbungs, M., Vernoux, J.-P., Jones, A., Lewis, R.J.,2002. Isolation and characterization of Indian Ocean ciguatoxin.Toxicon 40, 685–693.

Haque, S.M., Onoue, Y., 2002. Variation in toxin composition oftwo harmful raphidophytes, Chattonella antiqua and Chatonellamarina, at different salinities. Environ. Toxicol. 17 (2), 113–118.

Ishimatsu, A., Maruta, H., Tsuchiyama, T., Ozaki, M., 1990.Respiratory, ionoregulatory and cardiovascular responses of theyellowtail Seriola quinqueradiata to exposure to the red tideplankton Chattonella. Nippon Suisan Gakk. 56 (2), 189–199.

Ishimatsu, A., Tsuchiyama, T., Yoshida, M., Sameshima, M.,Pawlukm, M., Oda, T., 1991. Effect of Chattonella exposure onacid-base status of the yellowtail. Nippon Suisan Gakk. 57 (11),2115–2120.

Ishimatsu, A., Maruta, H., Oda, T., Ozaki, M., 1997. A comparisonof physiological responses in yellowtail to fatal environmentalhypoxia and exposure to Chattonella marina. Fish. Sci. 63 (4),557–562.

Jenkinson, I.R., Arzul, G., 2001. Mitigation by cysteine compoundsof rheotoxicity, cytotoxicity and fish mortality caused bythe dinoflagellates, Gymnodinium mikimotoi and G. cf.maguelonnense. In: Hallegraeff, G.M., Bolch, C.J.S., Blackburn,S.I., Lewis, R. (Eds.), Harmful Algal Blooms 2000. UNESCO,Paris, pp. 461–464.

Jüttner, F., 2001. Liberation of 5,8,11,14,17-eicosapentaenoic acidand other polyunsaturated fatty acids from lipids as a grazerdefense reaction on epilithic diatom biofilms. J. Phycol. 37,744–755.

Khan, S., Ahmed, S., Arakawa, O., Onoue, Y., 1995. Propertiesof the neurotoxins separated from a harmful red tide organismChattonella marina. Isr. J. Aquacult. 47, 137–141.

Khan, S., Arakawa, O., Onoue, Y., 1996. A toxicological studyof the marine phytoflagellate, Chattonella antiqua (Raphido-phyceae). Phycologia 35, 239–244.

Lewis, R.J., Sellin, M., Poli, M.A., Norton, R.S., MacLeod,J.K., Sheil, M.M., 1991. Purification and characterisation ofciguatoxins from moray eel (Lycodontis javanicus, Mura-enidae). Toxicon 29, 1115–1127.

Lewis, R.J., 1992. Ciguatoxins are potent ichthyotoxins. Toxicon30, 207–211.

Marshall, J.A., Hovenden, M., Hallegraeff, G.M., 2002a. Photosyn-thesis does influence reactive oxygen species production in thered tide alga Chattonella marina (Raphidophyceae). J. PlanktonRes. 142, 1231–1236.

Marshall, J.A., Nichols, P.D., Hallegraeff, G.M., 2002b. Chemota-xonomic survey of sterols and fatty acids in six marine raphi-dophyte algae. J. Appl. Phycol. 14 (4), 255–265.

Marshall, J.A., Munday, B., Yoshizawa, Y., Hallegraeff, G.M.,2001. Effect of irradiance on superoxide production byChattonella marina (Raphidophyceae) from South Australia andJapan. In Hallegraeff, G.M., Blackburn, S.I., Bolch, C.J., Lewis,R.J. (Eds.), Harmful Algal Blooms 2001. IntergovernmentalOceanographic Commission of UNESCO, Paris, pp. 316–319.

Oda, T., Akaike, T., Sato, K., Ishimatsu, A., Takeshita, S.,Muramatsu, T., Maeda, H., 1992a. Hydroxyl radical generationby red tide algae. Arch. Biochem. Biophys. 294, 38–43.

Oda, T., Ishimatsu, A., Shimada, S., Takeshita, S., Muramatsu,T., 1992b. Oxygen-radical-mediated toxic effects of the redtide flagellate Chattonella marina on Vibrio alginolyticus. Mar.Biol. 112, 505–509.

Oda, T., Nakamura, A., Okamoto, T., Ishimatsu, A., Muramatsu,T., 1998. Lectin-induced enhancement of superoxide anionproduction y red tide phytoplankton. Mar. Biol. 131, 383–390.

Okaichi, T., 1983. Marine environmental studies on outbreaks ofred tides in neritic waters. J. Oceanogr. Soc. Jpn. 39, 267–278.

Okaichi, T., Ochi, T., Nishio, S., Takano, H., Marsuno, K.,Morimoto, T., Murakami, T., Shimada, M., 1989. The causeof fish kills associated with red tides of Chattonella antiqua(Hada) Ono. In: Miyachi, S., Karube, I., Ishida, Y. (Eds.),Curent Topics in Marine Biotechnology. Japanese Society forMarine Biotechnology, Tokyo, pp. 185–188

Onoue, Y., Haq, M.S., Nozawa, K., 1990. Separation of neurotoxinsfrom Chattonella marina. Nippon Suisan Gakk. 56, 695.

Poli, M.A., Mende, T.J., Baden, D.G., 1986. Brevetoxins, uniqueactivators of voltage sensitive sodium channels, bind to specificsites in rat rain synaptosomes. Mol. Pharmacol. 30, 129–135.

Shimada, M., Murakami, T.H., Imahayashi, T., Ozaki, H.S.,Toyoshima, T., Okaichi, T., 1983. Effects of sea bloom,Chattonella antiqua, on gill primary lamellae of the youngyellowtail, Seriola quinqueradiata. Acta Histochem. Cytochem.16, 232–244.