The toxigenic marine dinoflagellate Alexandrium tamarense as the probable cause of mortality of caged salmon in Nova Scotia

Harmful Algae 1 (2002) 313–325

The toxigenic marine dinoflagellateAlexandrium tamarenseas the probable cause of mortality of caged

salmon in Nova Scotia

A.D. Cembellaa,∗, M.A. Quilliama, N.I. Lewisa, A.G. Baudera,C. Dell’Aversanoa,1, K. Thomasa, J. Jellettb, R.R. Cusackc

a Institute for Marine Biosciences, National Research Council (Canada), 1411 Oxford Street, Halifax, Nova Scotia, Canada B3H 3Z1b Jellett Biotek Ltd., 47 Wake Up Hill Road, RR1, Chester Basin, Nova Scotia, Canada B0J 1K0

c Department of Agriculture and Fisheries, Veterinary Pathology, P.O. Box 550, Truro, Nova Scotia, Canada B2N 5E3

Received 1 August 2002; received in revised form 28 August 2002; accepted 1 September 2002

Abstract

The toxins associated with paralytic shellfish poisoning (PSP) are potent neurotoxins produced by natural populations ofthe marine dinoflagellateAlexandrium tamarense. In early June 2000, a massive bloom (>7× 105 cells l−1) of this dinoflag-ellate coincided with an unusually high mortality of farmed salmon in sea cages in southeastern Nova Scotia. Conditions inthe water column in the harbour were characterised by the establishment of a sharp pycnocline after salinity stratificationdue to abundant freshwater runoff. In situ fluorescence revealed a high sub-surface (2–4 m depth) chlorophyll peak related tothe plankton bloom. The intense bloom was virtually monospecific and toxicity was clearly related to the concentration ofAlexandriumcells in plankton size fractions. Cultured clonal isolates ofA. tamarensefrom the aquaculture sites were verytoxic on a per cell basis and yielded a diversity of PSP toxin profiles, some of which were similar to those from planktonconcentrates from the natural bloom population. The toxin profile of plankton concentrates from the 21–56�m size fractionwas complex, dominated by theN-sulfocarbamoyl derivative C2, with levels of other PSP toxins GTX4, NEO, GTX5 (=B1),GTX3, GTX1, STX, C1, and GTX2, in decreasing order of relative abundance. Although no PSP toxin was found systemicallyin the fish tissues (liver, digestive tract) from this salmon kill event, the detection ofAlexandriumcells and low levels of PSPtoxins in salmon gills provide evidence that the enhanced mortalities were caused by direct exposure to toxicAlexandriumcells and/or to soluble toxins released during the bloom.© 2002 Elsevier Science B.V. All rights reserved.

Keywords: Alexandrium; Fish kill; Paralytic shellfish toxins; PSP

∗ Corresponding author. Tel.:+1-902-426-4735;fax: +1-902-426-9413.

E-mail address:[email protected] (A.D. Cembella).1 Present address: Dipartimento di Chimica delle Sostanze

Naturali, Universit̀a degli Studi di Napoli “Federico II”, Via D.Montesano 49, 80131 Napoli, Italy.

1. Introduction

Harmful algal blooms are often linked to finfishmortalities, either via direct toxicity effects or becauseof gill clogging and mucus production caused by ex-posure to high biomass of the causative organisms(Bruslé, 1995). The economic and ecological conse-quences can be severe. For example, a recent algal

1568-9883/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved.PII: S1568-9883(02)00048-3

314 A.D. Cembella et al. / Harmful Algae 1 (2002) 313–325

Fig. 1. Structures of the major naturally occurring PSP toxins found in dinoflagellates and bivalve shellfish.

bloom in Chile, off the island of Chiloe, resulted inthe loss of 1800 metric tonnes of salmon production(reported by the Association of Chilean Salmon andTrout Farmers, 2002).

The accumulation of potent neurotoxins associatedwith paralytic shellfish poisoning (PSP) in shellfish(Fig. 1), and the resulting regulatory and human healthconsequences, are also well documented (Shumway,1990). Nevertheless, reports of finfish mortalities as-sociated with PSP toxin-producing dinoflagellatesare comparatively rare. In the North Atlantic coastof North America, species of the toxigenic dinoflag-ellate genusAlexandrium(Halim) Balech are mostoften the cause of PSP toxicity events (Andersonet al., 1994). Prior to the event reported herein fromthe eastern coast of Nova Scotia, Canada, the onlywell-documented case of mass death of aquaculturedsalmonids associated with anAlexandriumbloom (asGonyaulax excavata= Alexandrium tamarense) oc-curred in 1984 in Trongisvagur Fjord, Faroe Islands,Denmark (Mortensen, 1985).

Wild fish can be exposed to lethal levels of PSPtoxins in several ways, depending upon the species,nutritional mode and life cycle stage (White, 1984).

Larval stages are particularly susceptible to PSP tox-ins (Robineau et al., 1991), but even adult fish maybe exposed to a lethal dose of such toxins by oralingestion of prey (White, 1981a). Phytoplanktivo-rous fish, such as menhaden, can accrue high levelsof PSP toxins by directly grazing on toxic plankton(White, 1984). Furthermore, these toxins can be ac-cumulated (up to a lethal dose) in fish through thefood chain, via ingestion of toxic phytoplankton byherbivorous copepods, pteropods, or gelatinous zoo-plankton, which are then eaten by zooplanktivorousfish (Montoya et al., 1996; White, 1981b). These areunlikely routes of exposure to PSP toxins by caged,adult salmon whose primary diet consists of foodpellets.

In early June 2000, sudden unusually high fishmortalities were reported from among caged salmon(Salmo salarL.) at sites on the southeast shore ofNova Scotia. Affected fish were observed to be lethar-gic and disoriented, swimming near the surface andthrowing their head from side to side. Dead fish hadsevere edema and irritation of the gills and an autopsyrevealed the presence of blood clots in the heart tissue(Cusack et al., 2002).

A.D. Cembella et al. / Harmful Algae 1 (2002) 313–325 315

In this study, we present further evidence thatthe toxigenic properties and high concentration ofthe bloom of A. tamarensewas responsible forthe enhanced salmon mortalities and provide newinformation on the toxin content and composi-tion of this dinoflagellate from the coast of NovaScotia.

2. Materials and methods

2.1. Collection, enumeration and identificationof plankton

In response to an apparent rapid increase in mor-tality rate of caged salmon in southeastern, NovaScotia (latitude 43◦45′15′′ longitude 65◦19′00′′),plankton samples were first collected on 5 June 2000by bucket from surface waters at the Salmon Wharfand at 0, 4.5 and 9 m depth by divers at Site D (seeFig. 2). Unconcentrated plankton samples were pre-served in 2% formalin–acetic acid and identified andcounted upon 25 mm 1�m, pore size Poretics poly-carbonate filters according to the filter-transfer-freeze(FTF) method (Hewes and Holm-Hansen, 1983) un-der phase-contrast microscopy. Species identificationwas confirmed at 400× and cells were counted at100–200× magnification.

On 9 June 2000, plankton samples were collectedby Niskin entrapment bottles at 0, 3 and 7 m depth

Fig. 2. Map of the harbour indicating location of salmon aquaculture and sampling sites.

and using a 2.5 cm diameter Tygon tube to collect anintegrated water sample from 0 to 7 m depth. Bulkplankton samples were also collected by pumpingfrom 3 m depth at the Salmon Wharf with a highvolume (ca. 200 l min−1) centrifugal pump into aconical Nitex plankton net (0.5 m × 2.5 m) of 20�mmesh. The plankton concentrated from ca. 12,000 l ofseawater was harvested for PSP toxin analysis andoptical microscopy with this method. The contentsof the cod-end of the net were divided into two sizefractions (21–56 and >56�m) using Nitex screens.Live cells from the smaller-size fraction were re-tained for culturing and a sample of each size fractionwas diluted (1:400, 21–56�m; 1:10, >56�m) withseawater and preserved in 1% paraformaldehyde (fi-nal concentration) for plankton identification andenumeration.

Plankton present in net concentrates were identi-fied and counted in duplicate under phase-contrastmicroscopy (magnification: 312.5×). The identifi-cation of A. tamarensewas confirmed by Nomarskicontrast-interference microscopy and by epifluores-cence microscopy of thecal plates stained by cal-cofluor (magnification: 400×; Fritz and Triemer,1985).

Several rRNA-targetted oligonucleotide probes toregions of the large subunit (ls) and small subunit(ss) ribosomal DNA ofAlexandriumspp. were ap-plied to field samples to further confirm identity ofthese taxa. The application of this method to fixed


field samples of plankton is provided in detail inJohn et al. (2002).

2.2. Physical and chemical oceanographic data

Physical data (temperature, salinity, pressure) re-lated to stratification of the water column were ob-tained during rising- to high-tide by profiling with aconductivity–temperature–depth sensor (Applied Mi-crosystems STD-12 Plus). An attached fluorometer(Seatech) was used to log in situ fluorescence as-sociated with chlorophyll. At sites A and C, and atthe Salmon Wharf, a vertical cast was made to 8 mdepth.

Inorganic nutrient samples were collected by Niskinbottles at the Salmon Wharf (0, 3, and 7 m) and at SiteA (0 m), to determine whether high nutrient loadingmight have contributed to the bloom. Approximately20 ml of seawater for each sample was filtered througha syringe mounted 25 mm Whatman GFF glass-fibrefilter into an acid-washed plastic vial. Samples werestored frozen (−20◦C) prior to analysis. Nutrientanalysis was performed by standard procedures, withcurrent modifications for automated analysis as de-scribed byStrain and Clement (1996). Inorganic nu-trients, including silicate, phosphate, nitrate+ nitrite,and ammonium were analysed by the colourimetrictechnique on a Technicon AutoAnalyzer II (AAII)segmented-flow system.

2.3. Culture of Alexandrium clones

The concentrate from the 21–56�m size fractionwas diluted in filtered seawater and held overnight at16◦C. Individual cells were isolated from this assem-blage by micropipette and transferred to small glasstubes containing 1 ml of L1 growth medium diluted1:10 with sterile seawater. Plankton isolates wereincubated at 16◦C under a photon flux density of120�mol m−2 s−1 on a 14:10 h light:dark photocycle.When cell concentrations reached >100 cells ml−1

cultures were transferred to 125 ml Erlenmeyer flasks.Exponentially growing cells were transferred to Fern-bach flasks containing 700–1700 ml of L1 mediumand incubated for 16–25 days under the same con-ditions as for the establishment of the clones. Thecultures were harvested in late exponential growthphase for PSP toxin analysis.

2.4. Preparation of toxin extracts

Field plankton collected by pumping at 3 m depthand cultured cells were concentrated by gravity filtra-tion on a 20�m Nitex sieve. From theAlexandriumclones, between 1.7 and 14.4 × 106 cells were har-vested in late exponential growth phase. The cell con-centrate from these cultures as well as the concentratedsurface sample collected by bucket, were pelleted bycentrifugation at 4000× g for 20 min at 5◦C in 15 mlFalcon centrifuge tubes. For liquid chromatographywith fluorescence detection (LC–FD) analyses, thefield plankton samples were extracted by sonication(2 min at 25 W; 50% pulse duty cycle) in an ice-bathusing a tapered microtip (VibraCell, Sonics & Mate-rials, Danbury, CN, USA) in 5:1 (v/w) of 0.1 M aceticacid. Cultures were extracted in 1 ml 0.1 M acetic acidand sonicated 4 min as described earlier. Microscopicobservations (125× magnification) were performedby phase-contrast microscopy to confirm completecell breakage. Plankton extracts were centrifuged for15 min at 4000× g at 5◦C. The supernatant wasfiltered through a Luer-mount 0.45�m Millex GV4syringe filter (Millipore, Bedford, MA, USA).

For analyses by hydrophilic-interaction liq-uid chromatography with detection by tandemmass spectrometry (HILIC–MS/MS), a planktoncell pellet (1 g) was mixed with 1.5 ml acetoni-trile:water:formic acid (80:19.9:0.1) and allowed torest undisturbed for 10 min. The suspension wascentrifuge-filtered through a 0.45�m membrane (Mil-lipore Ultrafree-MC, Bedford, MA), then the filterwas washed twice with 1.5 ml of the same solventmixture. The filtrates were combined and the vol-ume was adjusted to 5 ml with extraction solvent. Aclean-up step was performed by passing 1 ml of thisextract through a polyhydroxyethylAspartamideTM

cartridge (Poly-LC, Columbia, MD, USA) previouslyconditioned with water, acetonitrile:water:formic acid(10:89.9:0.1) and then extraction solvent. The car-tridge was washed first with 1 ml extraction solvent,then the toxins were eluted with 2.5 ml acetoni-trile:water:formic acid (10:89.9:0.1).

Blue mussels (Mytilus edulis) were collected fromthe surface of the Salmon Wharf and from 0 to 8 mon a salmon cage at Site C on 9 June 2000. Wholemussel soft tissues, removed from a large number ofspecimens, were pooled, homogenised and extracted


in boiling 0.1N HCl according to theAOAC (2000)method for PSP mouse bioassay. Prior to LC analy-sis, the extract supernatant was passed through a solidphase extraction (SPE) cartridge (Waters Oasis HLB)preconditioned with methanol, deionised water, and0.1 M acetic acid. The cartridge was dried by vacuum,then 1 ml of the extract was loaded and collected drop-wise.

Samples of liver, viscera and gill tissues harvestedfrom dead salmon from the toxic event and controlsamples excised from healthy adult salmon raisedin tanks at the NRC Aquaculture Research Stationat Sandy Cove, NS were extracted according to theAOAC (2000)method.

2.5. Determination of toxins and toxicity

2.5.1. PSP toxin assaysMouse bioassays (n = 3) were performed on ex-

tracts of whole mussels according to theAOAC (2000)protocol and the results were converted to toxicityunits: �g saxitoxin equivalents (STXeq) kg−1 wetweight of edible mussel tissue.

Fish tissues were assayed for the presence ofPSP toxins with an immunodiagnostic test, anenzyme-linked immunosorbent assay (ELISA) forsaxitoxin analogues developed by the Institute forMarine Biosciences, NRC and modified by JellettBiotek, Ltd. (Cembella et al., 2002). Fish samples ofunknown toxin content and controls were incorpo-rated into this microtitre plate assay by pre-incubatinga serial dilution of the extracts with PSP toxin anti-bodies for 30 min at room temperature. These incu-bates were transferred to a conjugate-coated plate forfurther incubation (1 h at 37◦C). After exposure tosecondary antibodies coupled to horseradish peroxi-dase, colour development at 490 nm was determinedspectrophotometrically on a microplate reader.

2.6. Instrumental analysis of PSP toxin composition

The presence of PSP toxins in the field plankton andmussel samples was confirmed by HILIC–MS/MS(Quilliam et al., 2001) on an Agilent (Palo Alto, CA)HP1090 LC system with detection by a Perkin-ElmerSCIEX (Concord, ON, Canada) API-III+ triplequadrupole mass spectrometer equipped with anion-spray source. Qualitative analyses were carried

out in positive ion mode by both full scan and se-lected reaction monitoring (SRM). Argon was usedas the collision gas in the second rf only quadrupole.A collision energy of 20 eV was used. Separationswere carried out on a 250 mm× 2 mm i.d. columnpacked with 5�m TosoHaas TSK-GEL Amide-80material with 0.2 ml min−1 of mobile phase com-posed of CH3CN:H2O (62:38, v/v) containing 2 mMammonium formate and 3.5 mM formic acid. A 5�linjection volume was used and the column effluentwas split, with 10% going to the mass spectrometer.

Toxin composition of the field plankton,Alexan-drium cultures and whole mussel tissue extractswas determined by high performance LC–FD af-ter post-column oxidation of the toxin derivatives(Oshima, 1995). Qualitative microbore LC–FD anal-yses were performed on an Agilent (Palo Alto, CA)HP1090 LC system equipped with an HP1040A flu-orescence detector operated with a gain setting of18 and excitation and emission wavelengths of 330and 390 nm, respectively. Toxins were oxidised tofluorescent derivatives in a post-column reaction coil(0.2 ml woven Teflon tubing) held at 70◦C using0.08 ml min−1 0.05 M periodic acid (pH 7.8). The elu-ent from the reactor was acidified with 0.08 ml min−1

0.75 M nitric acid. Separation of GTX, NEO and STXtoxins was achieved by gradient elution on a Phe-nomenex (Torrance, CA) column (2 mm× 250 mm)packed with 5�m AQUA stationary phase. The col-umn temperature was room temperature and the mo-bile phase flow rate was 0.2 ml min−1. Eluent A waswater with 6 mM heptane sulphonate, 15 mM ammo-nium phosphate (pH 7.1) and 0.25% (v/v) THF. EluentB was water:acetonitrile (70:30, v/v) with 6 mM hep-tane sulphonate, 15 mM ammonium phosphate (pH7.1) and 3% (v/v) THF. Separation of the C toxinswas performed isocratically on the same Agilent LCsystem using a Phenomenex column (2 mm× 50 mm)packed with 5�m LUNA stationary phase. The col-umn temperature was maintained at 36◦C and themobile phase flow rate was 0.2 ml min−1. The mo-bile phase consisted of 2 mM tert-butyl ammoniumphosphate at pH 5.8.

Quantitative measurements of toxin compositionof various samples were determined by LC–FD usingthe post-column oxidation method ofOshima (1995)as modified byParkhill and Cembella (1999). Threeseparate isocratic elutions were employed to separate


the complete set of PSP toxins at a flow rate of0.8 ml min−1. Toxins were separated on the followingcolumns: (a) Betabasic-C8 (4.6 mm× 250 mm; Key-stone Scientific, Bellefonte, PA) for the STX, dcSTXand NEO group, and the C1–C4 toxins group; (b)Zorbax SB-C8 (4.6 mm×250 mm; Agilent, Palo Alto,CA) for the gonyautoxins group, GTX1-GTX6. Forquantitation, 10�l injections of extract were comparedwith external toxin standards (PSP-1C) provided bythe Certified Reference Material Program (CRMP) ofthe Institute for Marine Biosciences, NRC, Halifax,Canada. Toxin concentrations (�mol l−1) were con-verted by the formula given inParkhill and Cembella(1999) to toxicity units (pg STX equivalents cell−1)using specific toxicity conversion factors (in mouseunits �mol−1) provided inOshima (1995).

3. Results

In the first week of June 2000 a massive bloom(>7× 105 cells l−1) of the dinoflagellateAlexandrium(Halim) Balech coincided with an enhanced mortal-ity of farmed salmon in aquaculture cages at sites inNova Scotia. There seemed to be some localisation ofthe phenomenon, with the highest number of deathsoccurring at Site A, some at Site C, and none at Site B

Fig. 3. Stratification of the water column and variation ofA. tamarensecell concentrations with depth at the Salmon Wharf.

(seeFig. 2) indicating a “patchiness” of the causalagent. Just prior to the salmon deaths there had beena heavy rainfall, resulting in brackish water at the sur-face with a dark brown colour due to runoff of organicmaterial.

By June 9, after the peak of salmon mortalities,the water column was characterised by the establish-ment of a sharp pycnocline following salinity stratifi-cation due to abundant freshwater runoff—conditionsstill apparently favourable to the dinoflagellate bloom.A weak shallow thermocline (12–16◦C) and a verypronounced halocline (21–30 psu) had developed atthe Salmon Wharf (Fig. 3). The strong pycnoclineindicated by the sharp near-surface density gradient(sigmat [σ t ]) was due primarily to the effect of abun-dant freshwater runoff. At this site there was relativelyhigh fluorescence associated with phytoplankton andother chlorophyllous particulates from 0 to 7 m, peak-ing at 2–3 m. The results of the CTD casts at Sites Aand C are not shown as there was little difference inthe temperature or salinity from 0 to 8 m. There was,however, a large peak in fluorescence between 4 and6 m at these stations.

There was no evidence of nutrient loading in theharbour during and subsequent to the peak of salmonmortalities. At the Salmon Wharf and Site A, nutrientlevels ranged from 0.3 to 0.5�M for nitrate+ nitrite,


0.7 to 0.8�M for soluble reactive phosphate, 1.0 to2.0�M for ammonium, and 3.7 to 9.4�M for dis-solved silicate. Furthermore, there was no apparenttrend with depth in the water column, except for sili-cate levels, which were highest at the surface.

The intense dinoflagellate bloom in the harbour wasvirtually monospecific and toxin in the water columnwas clearly related to the concentration ofAlexan-drium cells in the plankton size fractions. In the firstweek of June, during the maximal period of salmonmortalities,Alexandriumcell concentrations were con-sistently higher at the surface than at depth at all sta-tions. By June 9, in situ fluorescence revealed a highsub-surface (2–4 m depth) chlorophyll peak clearly re-lated to this plankton bloom, although at the SalmonWharf concentrations ofAlexandriumcells were actu-ally the highest towards the bottom of the water col-umn (Fig. 3).

On June 9, the 21–56�m size fraction was domi-nated byAlexandriumcells (>87% of the total count),in various life-history phases, including the vegetativestage, hypnozygotes (cysts) and putative gametes andplanozygotes. Vegetative cells were well-pigmentedand apparently healthy. The morphological character-istics of vegetative cells (size, shape) and the thecalplate features (Kofoid notation), including presence ofa ventral pore on the first apical (1′) plate, form of theapical pore complex (APC), and the shape of the 6′ andthe posterior sulcal (S.p.) plates, of field and culturedspecimens were in accord with the description ofA.tamarense(Lebour) BalechsensuBalech (1995). Thethecal plate tabulation (Kofoid notation) was: Po, 4′,6′′, 5′′′, 2′′′′, 6 c and 9–10 s. Most field specimens con-formed to the description of “var.tamarensis” as op-posed to “var.excavata”, as distinguished byBraarud(1945), i.e. the absence of the deeply “excavated” sul-cal region.

The use of species-specific rRNA probes forA.tamarenseandA. ostenfeldiirevealed that theAlexan-drium cells present in the field samples were associ-ated with the North American clade ofA. tamarense.Both oligonucleotide probes forA. ostenfeldiigavea negative response, indicating that this species wasessentially absent for theAlexandriumbloom. Thesedata are consistent with the morphological observa-tions performed by microscopy.

Niskin bottle samples collected at 0, 3 and 7 m weredominated byA. tamarense(Fig. 4A) with the lowest

concentration, 2.8 × 103 cells l−1, at the surface, in-creasing to 2.3 × 104 cells l−1 at 3 m depth, with themaximum, 3.0 × 104 cells l−1, at 7 m depth (Fig. 3).The dinoflagellatesDinophysis norvegicaand Cer-atium longipesoccurred in very low numbers (0.1 ×103 to 0.2×103 cells l−1) at all depths sampled. In thesurface sample, other thecate dinoflagellates, such asHeterocapsa triquetra, were also observed in very lownumbers (<0.2 × 103 cells l−1). At 3 and 7 m depth,Scrippsiella trochoideaandGyrodinium spiralewerepresent at a concentration of 0.3 × 103 to 0.8 × 103

cells l−1.In the 21–56�m size fraction,S. trochoidea, which

could easily be confused with smaller cells ofAlexan-drium upon cursory observation, andProtoperidiniumbrevipes, occurred in low abundance (about 5% each).Other dinoflagellates, includingGonyaulax spinifera,Amylax triacantha, D. norvegicaand D. acuminatamade up<2% of the sample in this size fraction, asdid the diatomsLicmophorasp., Fragilaria sp. andStriatella unipunctata(prox.) (Fig. 4B).

The >56�m size fraction was dominated by thediatom S. unipunctata(37%), the dinoflagellateC.longipes(26%) and larger cells ofA. tamarense(22%),including putative zygotes. There were low occur-rences of copepods, tintinnids, pollen grains, other di-noflagellates (3%), includingCeratium lineatum, D.norvegica, Protoperidinium depressum, P. pellucidum,S. trochoidea, Gyrosigma spirale, and other diatoms(8%), includingAsterionella formosa, Fragilaria sp.,Leptocylindricus danicus, Licmophora sp., Chaeto-ceros laciniosusandNitzschia longissima, in this frac-tion (Fig. 4B).

Thirty-four clonal isolates ofAlexandrium weresuccessfully established from the harbour in the weekfollowing the peak of the salmon mortalities. Theseclones are all believed to beA. tamarense, althoughthe critical identification of only a few isolates (e.g.SB30, SB31 and SB32) has been confirmed by No-marski contrast-interference microscopy. Cells inculture are approximately 36�m in diameter, andhave thecal features characteristic ofA. tamarense,including a small ventral pore on the first apical (1′)plate and the distinctive shape of the 6′′ plate.

The results of HILIC–MS/MS analysis, as shown inthe chromatogram inFig. 5 for the 21–56�m plank-ton size fraction collected at the Salmon Wharf, con-firmed unequivocally that PSP toxins were present in


Fig. 4. Plankton composition (% total abundance) at the Salmon Wharf subsequent to the peak in salmon mortalities. (A) Niskin bottlesamples at 0, 3 and 7 m depth; (B) size fractionated net-concentrate samples pumped from 3 m depth.

Fig. 5. HILIC–MS/MS analysis of PSP toxins in an extract of the 21–56�m plankton size fraction, collected from the Salmon Wharf.The SRM mode was used with the following precursor-product ion combinations: (a)m/z 412> 332 and 412> 314 for GTX1/4; (b)m/z396> 316, 396> 298, 316> 298, and 316> 220 for C1/C2 and GTX2/3, with the last two combinations being also indicative of NEO;(c) m/z 380> 300, 380> 282, 300> 282 and 300> 204 for GTX5, with the last two combinations being also indicative of STX. Thechromatograms were generated by summing the signals within each group. Other conditions as inSection 2.


the plankton bloom. The PSP toxins were identifiedbased on: (1) retention time; (2) molecular mass (i.e.[M + H]+ ions at m/z 476 for C1/C2,m/z 412 forGTX1/4,m/z396 for GTX2/3,m/z380 for GTX5,m/z316 for NEO, andm/z 300 for STX); (3) structural in-formation, such as the presence of one to four diagnos-tic fragments for each analyte; and (4) the ion ratios.The high selectivity of SRM, due to elimination ofsignals from other co-extractives, made interpretationof the results easy and unambiguous. Precursor andproduct ions were selected according to fragmentationpatterns observed for each standard PSP toxin. The

Fig. 6. LC–FD chromatograms of PSP toxins in field plankton collected by pumping at 3 m depth and net concentration at the SalmonWharf. (A) >56�m size fraction; (B) 21–56�m and (C) AOAC extract of whole mussels from Salmon Wharf.

samples exhibited complex PSP toxin profiles dom-inated by theN-sulfocarbamoyl derivatives C2 andGTX5 (=B1), the 11-hydroxysulfate carbamate toxinsGTX4 and GTX3, and the carbamate toxins NEO andSTX. Low levels of the toxins C1, GTX1 and GTX2,the �-epimers of C2, GTX4 and GTX3, respectively,were also observed, but these could have been formedas a result of epimerization during extraction.

Quantitative analysis of the same sample by LC–FD(Fig. 6) provided support for the HILIC–MS/MSresults and showed, in decreasing order of relativeabundance, the following toxins: C2, GTX4, NEO,


Fig. 7. Relative toxin composition (% molar) determined by LC–FD of (A) plankton sample collected by bucket from the surface at SiteD; (B) 21–56�m size fraction from pump concentrate at the Salmon Wharf; (C)Alexandrium tamarenseclone SB31; (D)Alexandriumtamarenseclone SB32.

GTX5 (=B1), GTX3, GTX1, STX, C1, and GTX2.A trace amount of decarbamoyl GTX3 was also de-tected by LC–FD, but this was not confirmed byHILIC–MS/MS. The toxin profile of the unfraction-ated field sample collected 2 days earlier at Site Dwas virtually identical (<20% coefficient of variationamong all the toxins) and the calculated cell toxicity(10.7 pg STXeq cell−1) was close to that determinedfrom the size fractionated material in the 21–56�m(7.5 pg STXeq cell−1) and the >56�m (6.2 pg STXeqcell−1) fractions collected on 9 June 2000.

Cultured clonal isolates ofA. tamarensewere verytoxic on a per cell basis. Calculated cellular toxic-ity varied from 3 to 60 pg STXeq cell−1 among theAlexandriumisolates (n = 34), compared to 6–11 pgSTXeq per cells in the field samples collected duringthe bloom (based only upon totalAlexandriumcells).

The cultures yielded a diversity of PSP toxin profiles,only some of which were similar to the toxin profilein plankton concentrates from the natural bloom pop-ulation (Fig. 7).

No PSP toxins were detected by the ELISA methodin the salmon liver, viscera or muscle samples col-lected from dead specimens from the harbour. Never-theless, low levels were detected in the gill tissues ofthese same specimens, varying from approximately 5to 10�g STXeq kg−1 of tissue.

By the time salmon mortalities had declined, PSPtoxin levels in wild mussels attached to the SalmonWharf and the salmon cages were extremely high. Atthe Salmon Wharf on 9 June 2000, the PSP toxicity,as determined byAOAC (2000)mouse bioassay, was2.2×104 �g STXeq kg−1 total soft tissues, with evenhigher levels of 6.7×104 and 3.7×104 �g STXeq kg−1


measured at the surface and 8 m depth, respectively,at Site C. The toxin profile determined by LC–FDwas virtually identical among the sites and was verysimilar to that of the natural plankton sample of the21–56�m size fraction (Fig. 6). These results werealso confirmed by HILIC–MS analyses of the musseltissue extract.

4. Discussion

The localized bloom ofA. tamarenseremains themost likely explanation for the enhanced mortality ofsalmon in the cages. Salmon farmers did report a rustybrown discharge from a ship in the harbour, prior tothe enhanced salmon mortality event. The proximity ofthe reported discharge (>1 km from the nearest salmoncage) and pattern of mortalities was not consistentwith this point source as a plausible cause. Interviewswith Coast Guard personnel and the Harbourmasterregarding ship traffic and maintenance at the dockyardprovided further evidence that the mortalities were notrelated to such “pollution”.

Dead salmon from the cages that exhibited en-hanced mortalities were examined for the presence offish parasites and pathogens—none were found, in-cluding the absence of the organisms associated withbacterial kidney disease (BKD) and infectious salmonanaemia (ISA;Cusack et al., 2002). Histopathologicalexamination of fish tissue, particularly of damage tothe gill epithelium, was consistent with that recordedin a previous case of salmon mortality attributed to anAlexandriumbloom in the Faroe Islands (Mortensen,1985).

No PSP toxins were found systemically in thefish tissues (liver, spleen, heart, muscle, or viscera)from the mortality incident by mouse bioassay, im-munoassay or LC–FD analysis (Cusack et al., 2002).A few intact Alexandrium cells were observed inthe alimentary canal of some of the dead fish (N.Lewis and A. Cembella, personal observation), butthis inadvertent ingestion is unlikely to yield enoughtoxin to result in mortality. Low levels of PSP tox-ins have been previously reported in the viscera ofwild adult chum salmon,Oncorhynchus ketare-turning to Japanese rivers to spawn (Sato et al.,1998), although this toxin was not necessarily re-lated to the presence of dinoflagellate cells. In the

Nova Scotian event, a few cells ofAlexandriumwerefound embedded in the gill interstices, trapped in theabundant mucus that was produced (Cusack et al.,2002). Saturating oxygen levels measured throughoutthe water column at the salmon cages effectively rulesout oxygen depletion caused by high biomass bloomsas the cause of enhanced mortalities. It is, therefore,most probable that the fish mortalities were causedby direct exposure to toxicAlexandriumcells and/orto soluble toxins released during the decline of themassive bloom.

The origin of the bloom in the harbour remainsunknown. It was not possible to determine if theAlexandrium-linked fish kill originated from endemicdinoflagellate populations within the harbour or byexogenous introduction from surrounding coastal wa-ters. Although the harbour is not routinely sampledfor PSP toxicity, low level toxicity has been occasion-ally detected since the 1970s in bivalve shellfish fromadjacent areas outside the harbour (Canadian FoodInspection Agency Toxicity Reports). Yet it is likelythat the combination of an extremely high magnitudeAlexandriumbloom, high levels of toxicity in mus-sels, and increased fish mortalities associated withAlexandriumwere unprecedented occurrences priorto the year 2000.

Seasonal blooms ofAlexandriumspp. in the NorthAtlantic are known to be initiated by the triggering ofexcystment from benthic resting cysts (hypnozygotes)(Anderson, 1998), but no cyst surveys are available forembayments along the eastern coast of Nova Scotia.Furthermore, it is unclear whether or not such cystsor vegetative cells may have been introduced to theharbour either just prior to the bloom reported hereor from historical events (e.g. via ship deballasting).For many years, the practice (now illegal) was to pro-cess scallops from Georges Bank and other offshorelocations at Shelburne by excising only the adductormuscle and discarding the rest of the tissues as wasteinto the harbour. Since sea scallops (Placopecten mag-ellanicus) from this offshore region may contain ex-tremely high PSP toxin levels (Cembella et al., 1993,1994) and may also contain viableAlexandriumcystsin the viscera (Bourne, 1965), it is conceivable thatthis may have contributed to “seeding” the harbour.

The nutrient levels measured in the harbour dur-ing the mortality event were typical of those foundin coastal embayments of Atlantic Canada during the


summer (Strain and Clement, 1996; Strain, 2002),even in salmon farming areas, although the ratio of am-monium:nitrate was somewhat elevated, perhaps dueto biological activity associated with the fish farm-ing. Nevertheless, there was no apparent eutrophica-tion that may have contributed to the bloom.

Alexandriumblooms have occurred in this area inthe years subsequent to the salmon mortality event re-ported here, but they have not been associated withsuch high levels of PSP toxicity in molluscan shellfishfrom southeastern Nova Scotia. Mussels are known tobe susceptible to rapid toxification when exposed toPSP toxin-producing blooms and thus they are oftenused as “sentinel species” for toxin monitoring pro-grams. The high levels of PSP toxins accumulated inmussels on the salmon cages in the harbour within10 days of the onset of enhanced salmon mortalitiesin 2000 provides excellent evidence of a persistenthigh magnitude toxicAlexandriumbloom. This corre-sponds well with the extremely high measurements ofPSP toxicity (9×105 �g STXeq kg−1) in mussel bedsadjacent to the site of salmonid mortalities within 2weeks of the onset of a similar incident in the FaroeIslands (Mortensen, 1985).

The PSP toxin profiles of cloned cultures isolatedfrom theAlexandriumbloom involved in the salmonkill were very diverse, and the field population cannotbe considered to represent a common genotype. Thepresence of small amounts of�-epimers (C1, GTX2,GTX1) in the field plankton is likely due to facileepimerization of the dominant�-epimers (C2, GTX3,GTX4) synthesised by the dinoflagellate during theacid extraction and processing of the samples (Hallet al., 1990; Cembella, 1998). Epimerization mayaccount for small apparent differences in the toxinprofile of samples prepared for analysis by differ-ent methods, while other slight differences in toxincomposition between mussels and the field plank-ton may be related to the extraction of the formersamples in hot 0.1N HCl, which can cause somedecomposition ofN-sulfocarbamoyl toxins to theirrespective carbamate analogues (Cembella et al.,1993). Traces of decarbamoyl GTX3 identified inthe field plankton (very unusual for dinoflagel-lates) by LC–FD, but not confirmed by LC–MS,may have been due to slight decomposition dueto differential sample processing or it may havebeen present as a trace metabolite in the plank-

ton via bacterial activity or in zooplankton fecalpellets.

In any case, the relatively high cellular toxin content(>50 fmol cell−1) and calculated toxicity in the naturalAlexandrium-dominated population are in agreementwith the evidence of high potency of this bloom againstfarmed salmon. The fact that field samples collectedon different days by different methods were generallyconsistent in toxicity and toxin profile indicate thatthe data are quantitatively reliable. In the absence ofother possible causative agents, the aetiology of thefish deaths, and the coincident timing of the salmonmortalities with the bloom dynamics, the high biomasstoxigenic bloom ofAlexandriumis the most plausibleculprit for this event.

Acknowledgements

We thank Jeff Nickerson for assistance with sam-ple collection. Laura Brown, Institute for Marine Bio-sciences (IMB) performed preliminary analysis of fishtissues for signs of pathology, and Cheryl Craft andSheila Crain, IMB assisted with the preparation oftissues for assay and analysis. Don Richard, Cana-dian Food Inspection Agency, Moncton, performed theAOAC mouse bioassays on the mussel extracts. Nutri-ent analyses were conducted courtesy of Peter Strain atthe Bedford Institute of Oceanography, Bedford, NS.Uwe John and Linda Medlin of the Alfred-WegenerInstitute, Bremerhaven, Germany provided access tothe oligonucleotide probes and assistance with theirapplication to field samples. This publication is NRCCNo. 42366.

References

Anderson, D.M., 1998. Physiology and bloom dynamics oftoxic Alexandrium species, with emphasis on life cycletransitions. In: Anderson, D.M., Cembella, A.D., Hallegraeff,G.M. (Eds.), Physiological Ecology of Harmful Algal Blooms,NATO-Advanced Study Institute Series, vol. 41. Springer,Heidelberg, pp. 29–48.

Anderson, D.M., Kulis, D.M., Doucette, G.J., Gallagher, J.C.,Balech, E., 1994. Biogeography of toxic dinoflagellates in thegenusAlexandrium from the northeastern United States andCanada. Mar. Biol. 120, 467–468.

Association of Official Analytical Chemists (AOAC), 2000.Paralytic shellfish poison. Biological method 959.08. In:


Horowitz, W. (Ed.), Official Methods of Analysis. Associationof Official Analytical Chemists Int., Gaithersburg, MD.

Balech, E., 1995. The GenusAlexandriumHalim (Dinoflagellata),Sherkin Island Marine Station Publication. Sherkin Island,County Cork, Ireland.

Bourne, N., 1965. Paralytic shellfish poison in sea scallops(Placopecten magellanicus, Gmelin). J. Fish. Res. Bd. Can. 22,1137–1149.

Braarud, T., 1945. Morphological observations on marine dino-flagellate cultures (Porella perforata, Gonyaulax tamarensis,Protoceratium reticulatum), Avh. Norske Vidensk. Akad. Oslo,Mat. Naturv. Kl. 11, 1–18.

Bruslé, J., 1995. The Impact of Harmful Algal Blooms on Finfish.Editions IFREMER, Centre de Brest, Instaprint S.A., Tours,France. 75 pp.

Cembella, A.D., 1998. Ecophysiology and metabolism of paralyticshellfish toxins in marine microalgae In: Anderson, D.M.,Cembella, A.D., Hallegraeff, G.M. (Eds.), PhysiologicalEcology of Harmful Algal Blooms, NATO-Advanced StudyInstitute Series, vol. 41. Springer, Heidelberg, pp. 381-404.

Cembella, A.D., Shumway, S.E., Lewis, N., 1993. A comparisonof anatomical distribution and spatio-temporal variation ofparalytic shellfish toxin composition in two bivalve speciesfrom the Gulf of Maine. J. Shellfish Res. 12, 389–403.

Cembella, A.D., Shumway, S.E., Larocque, R., 1994. Sequesteringand putative biotransformation of paralytic shellfish toxins bythe giant sea scallopPlacopecten magellanicus: seasonal andspatial scales in natural populations. J. Exp. Mar. Biol. Ecol.180, 1–22.

Cembella, A.D., Doucette, G.J., Garthwaite, I., 2002. In vitroassays for phycotoxins. In: Hallegraeff, G.M., Anderson,D.M., Cembella, A.D. (Eds.), Manual on Harmful MarineMicroalgae, Monographs on Oceanographic Methodology, vol.11. Intergovernmental Oceanographic Commission (UNESCO),Paris, in press.

Cusack, R.R., Carver, C., Peacock, N., Quilliam, M.A., Cembella,A.D., Craft, C., Nickerson, J., Ferguson, H.W., 2002. Mortalityof farmed Atlantic salmon (Salmo salarL.) associated withAlexandrium tamarense, a marine microalga responsible forparalytic shellfish poisoning. Dis. Aquat. Org., in preparation.

Fritz, L., Triemer, R.E., 1985. A rapid simple technique utilisingCalcofluor white M2R for the visualisation of dinoflagellatethecal plates. J. Phycol. 21, 662–664.

Hall, S., Strichartz, G., Moczydlowski, E., Ravindran, A.,Reichardt, P.B., 1990. The saxitoxins. Sources, chemistryand pharmacology. In: Hall, S., Strichartz, G. (Eds.), MarineToxins: Origin, Structure and Molecular Pharmacology. ACSSymposium Series 418. Am. Chem. Soc., Washington, DC,pp. 29–65.

Hewes, C.D., Holm-Hansen, O., 1983. A method for recoveringnanoplankton from filters for identification with the microscope:the filter-transfer-freeze (FTF) technique. Limnol. Oceanogr.28, 389–394.

John, U., Cembella, A., Hummert, C., Elbrächter, M., Groben,R., Medlin, L.K., 2002. Discrimination of the toxigenic

dinoflagellate speciesAlexandrium tamarenseandAlexandriumostenfeldii in co-occurring natural populations from Scottishcoastal waters using species-specific rRNA targeted probes. Eur.J. Phycol., in press.

Montoya, N.G., Akselman, R., Franco, J., Carreto, J.I., 1996.Paralytic shellfish toxins and mackerel (Scomber japonicus)mortality in the Argentine Sea. In: Yasumoto, T., Oshima,Y., Fukuyo, Y. (Eds.), Harmful and Toxic Algal Blooms.Intergovernmental Oceanographic Commission (UNESCO),Paris, pp. 417–420.

Mortensen, A.M., l985. Massive fish mortalities in the FaroeIslands caused by aGonyaulax excavatared tide. In: Anderson,D.M., White, A.W., Baden, D.G. (Eds.), Toxic Dinoflagellates.Elsevier, New York, pp. 163–170.

Oshima, Y., 1995. Post-column derivatisation HPLC methodsfor paralytic shellfish poisons. In: Hallegraeff, G.M.,Anderson, D.M., Cembella, A.D. (Eds.), Manual on HarmfulMarine Microalgae, IOC Manuals and Guides No. 33.Intergovernmental Oceanographic Commission (UNESCO),Paris, pp. 81–94.

Parkhill, J.P., Cembella, A.D., 1999. Effects of salinity, light,and inorganic nitrogen on growth and toxigenicity of themarine dinoflagellateAlexandrium tamarensefrom northeasternCanada. J. Plank. Res. 21, 939–955.

Quilliam, M.A., Hess, P., Dell’ Aversano, C., 2001. Recentdevelopments in the analysis of phycotoxins by liquidchromatography mass spectrometry. In: deKoe, W.J., Samson,R.A., Van Egmond, H.P., Gilbert, J., Sabino, M. (Eds.),Mycotoxins and Phycotoxins in Perspective at the Turn of theCentury. deKoe, W.J., Wageningen, Netherlands, pp. 383–391.

Robineau, B., Gagné, J.A., Fortier, L., Cembella, A.D., 1991.Comparison of the response of five larval fish species to thetoxic dinoflagellateAlexandrium excavatum(Braarud) Balech.J. Exp. Mar. Biol. Ecol. 152, 225–242.

Sato, S., Ogata, T., Kodama, M., 1998. Trace amounts of saxitoxinsin the viscera of chum salmonOncorhynchus keta. Mar. Ecol.Prog. Ser. 175, 295–298.

Shumway, S.E., 1990. A review of the effects of algal bloomson shellfish and aquaculture. J. World Aquacult. Soc. 21,65–104.

Strain, P.M., 2002. Nutrient dynamics in Ship Harbour, NovaScotia. Atmosphere Ocean 40, 45–58.

Strain, P.M., Clement, P.M., 1996. Nutrient and dissolved oxygenconcentrations in the Letang Inlet New Brunswick in thesummer of 1994. Can. Data Rep. Fish. Aquat. Sci. 1004,33.

White, A.W., 1981a. Sensitivity of marine fishes to toxins fromthe red-tide dinoflagellateGonyaulax excavataand implicationsfor fish kills. Mar. Biol. 65, 255–260.

White, A.W., 1981b. Marine zooplankton can accumulate andretain dinoflagellate toxins and cause fish kills. Limnol.Oceanogr. 26, 103–109.

White, A.W., l984. Paralytic shellfish toxins and finfish. In: Ragelis,E.P. (Ed.), Seafood Toxins, ACS Symposium Series 262, Am.Chem. Soc., Washington, DC, pp. 171–180.

The toxigenic marine dinoflagellate Alexandrium tamarense as the probable cause of mortality of caged salmon in Nova Scotia

Documents