Toward
predicting
Dinophysis
blooms
off
NW
Iberia:
A
decade
of
events
M
aTeresa
Moita
a,d,*
,
Yolanda
Pazos
b,
Carlos
Rocha
c,
Rita
Nolasco
c,
Paulo
B.
Oliveira
a,*
aInstitutoPortugueˆsdoMaredaAtmosfera(IPMA),RuaAlfredoMagalha˜esRamalho,6,1495-165Lisboa,Portugal
bInstitutoTecnolo´xicoparaoControldoMedioMarin˜o,PeiraodeVilaxoa´n,s/n,36611Vilagarcı´adeArousa,Spain
c
CESAMeDepartamentodeFı´sicadaUniversidadedeAveiro,3810-193Aveiro,Portugal
d
CCMAR,UniversidadedoAlgarve,CampusdeGambelas,8005-339Faro,Portugal
1. Introduction
HarmfulAlgalBloom(HAB)speciesareregularcomponentsof thephytoplanktonin northwesternIberian upwelling waters,a coastannually affectedmainlyduring summerand autumn by blooms, sometimes in very high concentration of cells, of Dinophysisspecies. These dinoflagellatesare diarrhetic shellfish poisoning(DSP)producersandarethemaincauseforlongperiods of shellfish harvesting closures with serious socio-economic
impacts in the region. Dinophysis, though tolerating upwelling turbulence,arefavoredbythermal andhalinestratification and there is little evidence that blooms are induced by elevated nutrientconcentrations(Delmasetal.,1992;Regueraetal.,1995; GEOHAB,2008).Theyseemtobeexcelentsurvivorsandcanattain highgrowthrates(Regueraetal.,2012).
Dinophysis are obligate mixotrophic species that in cultures depend simultaneously on the availability of prey, e.g. the phototrophic ciliate Mesodinium rubrum, and on light intensity forphotosynthesis(Kimetal.,2008).Innature,populationsofthe predator andpreystill needtoaggregatetoenhanceencounter rates. Despitethesespecificities,thedevelopment ofDinophysis blooms follow the general phases of population development (initiation, growth, maintenanceand decay)that,togetherwith theirtransportbycurrents,needtobestudiedinordertobetter
A R T I C L E I N F O
Keywords:
Harmfulalgalblooms
Upwelling
Interannualvariability
Meteorologicalanomalies
Shelfcirculation
A B S T R A C T
DinophysisacuminataandDinophysisacutaarerecurrentspeciesoffNWIberiabuttheiroutbreaksoccur underdifferentconditions. Adecade(2004–2013)ofweekly dataforeachspeciesattwo sentinel stationslocatedattheentranceofRiasdeAveiro-AV(NWPortugal,40838.60N)andPontevedra-PO
(Galicia,Spain,42821.50N),wereusedtoinvestigatetheregionalsynchronismandmesoscaledifferences
relatedtospeciesdetection,bloom(>200cellsL1)initiationanddevelopment.Resultshighlightthe
high interannual variability of bloom events and summarize the associated meteorological/ oceanographicconditions.D.acutabloomswereobservedin2004–2008and2013,andthespecies highestmaximaatAVoccurredafterthehighestmaximaofitspreyMesodinium,withatime-lagof2–3 weeks.D.acuminatabloomswereobservedeveryyearatbothstations.Thecellconcentrationtimeseries showsthatthebloomsgenerallypresentasequencestartinginMarchwithD.acuminatainPOandthree weekslaterinAV,followedbyD.acutathatstartsatAVandthreemonthslaterinPO.Exceptionally,D. acuminatabloomsoccurredearlieratAVthanPO,namelyinhighspringupwelling(2007)orriverrunoff (2010)years.Afour-yeargap(2009–2012)ofD.acutabloomsoccurredafterananomalous2008autumn withintenseupwelling whichis interpreted astheresultof anequatorwarddisplacement ofthe populationcore.Numericalmodelsolutionsareusedtoanalyzemonthlyalongshorecurrentanomalies andtesttransporthypothesesforselectedevents.Theresultsshowastronginterannualvariabilityinthe poleward/equatorwardcurrentsassociatedwithchangesinupwellingforcingwinds,theadvectionofD. acutabloomsfromAVtoPOandthepossibilitythatD.acuminatabloomsatAVmightresultfrominocula advectedsouthwardfromPO.However,thesensitivityoftheresultstoverticalpositionofthelagrangian tracerscallformorestudiesonspeciesdistributionatthevariousbloomstages.
ß2015ElsevierB.V.Allrightsreserved.
* Correspondingauthorat:InstitutoPortugueˆsdoMaredaAtmosfera(IPMA),
RuaAlfredoMagalha˜esRamalho,6,1495-165Lisboa,Portugal.
E-mailaddresses:tmoita@ipma.pt(M.T.Moita),pboliveira@ipma.pt
(P.B.Oliveira).
ContentslistsavailableatScienceDirect
Harmful
Algae
j our na l ho me p a ge : w ww . e l se v i e r . com / l oc a te / h a l
http://dx.doi.org/10.1016/j.hal.2015.12.002
predicttheeventsandmitigatetheireffects(Regueraetal.,2012). Furthermore,Regueraetal.(2012)pointedoutthattheconditions fortheinitiation,orforthesupplyofinoculumpopulations,may holdthekeytoexplainDinophysisbloominterannualvariability. TwentyfiveyearsofHABmonitoringonNWIberiawatersshow thatbloomsoftheendemicdinoflagellatesDinophysisacutaand Dinophysisacuminata are recurrent although presenting a high interannualvariabilitywhosecausesarestillpoorlyunderstood. Bothspeciescoexistbuttheirmaximanevercoincideinspaceorin time(Palma et al.,1998), and their peakscan alternate in late summerandearlyautumnassociatedwithdifferentstratification patterns(Regueraetal.,1993).InthewestcoastofPortugal(Fig.1), theregion between FigueiradaFoz(F.Foz)and Aveiro(AV), i.e. between408100Nand418N,isthemostproblematicarea,where thespecies maximaarenormally reported (Palmaet al., 1998; GEOHAB,2005;Valeetal.,2008),andhavebeenobservedinthin subsurfacelayerswithinthepycnocline(Moitaetal.,2006).This areaisingeneralconsideredtheepicenterforD.acutabloomsin northwesternIberia wherethey oftenoccur first andare more intensethaninadjacentareas(Moitaetal.,2006;Escaleraetal., 2010).Forsimilarreasons,Rı´adePontevedra,inGalicia,isthought tobeahot-spotforD.acuminatabloomsinNWIberia(Pazosetal., 2005). Exceptionally insome years, intenseblooms of D.acuta wereobserved onlyin Galiciaand notfurther south(e.g.1989, Regueraetal.,1995)orthereversewithD.acuminatabloomsin Portugal(e.g.2007,thiswork).
TheverysuddenincreaseinDinophysiscellnumbersdetectedin somecoastal monitoringsites cannotbefully explainedby the speciesgrowthratesalone.Thereisevidencethatpopulationsof both species can be transported alongshore but relaxation/ downwellingeventsmayalsocause theiraccumulationinshore (Reguera et al., 1995; Sordo et al., 2001; Crespo et al., 2006; Escaleraetal.,2010).Upwelling,throughpycnoclineshallowing, canserveasamechanismofconcentratingpopulationsthatcanbe
advected by the inshore poleward currents when the winds weaken,aphysicalmechanismobservedbySordoetal.(2001) andOliveiraetal.(2009).Contrastingclimaticconditionswere also evoked to explain the selection in the Galician rias of DinophysisacuminataandDinophysisacutabloomsin2002and in 2003 respectively (Escaleraet al., 2010): D.acuminata was associated with thermohaline stratification and persistent northerlywindswhileD.acutawasrelatedtoanexceptionally hotsummerwiththelowestupwellingindexoftheprevious50 years.Dı´az etal. (2013)showed thattheearly initiationof D. acuminatabloomsintheGalicianRı´asBaixas,observedin2002 and2012,wererelatedtopositiveanomaliesinSSTandatypical predominance of upwelling winds during the immediately previouswinters.
Thisworkaimsatinvestigatingindetailtheconditionsrelated toDinophysisacutaandDinophysisacuminatabloominitiationand intensificationintwo sitesoffNWIberia(Galician riasand NW coastofPortugal)separatedby200km.Datainclude:(i)adecade (2004–2013) of Dinophysis weekly data samples from two monitoring stations where the species often present regional maxima; (ii) meteorological reanalysis data and (iii) daily sea surface maps to provide a synoptic view of the sea surface conditionspriortobloomdevelopment;(iv)hydrodynamicmodel solutions toinvestigatetheadvectionof Dinophysispopulations betweenthetworegions.
2. Studyarea,dataandmethods 2.1. Thestudyarea
TheareaofstudyislocatedattheNWcoastofIberia(Fig.1),in thenorthernlimit oftheupwelling systemassociatedwiththe north Atlantic anticyclonic gyre. In this region, the upwelling occurs seasonally, from late spring to early Autumn with a
Fig.1.MapofNWIberiashowingthepositionofthetwoHABlongtermmonitoringstations,POlocatedinfrontofBue´uinRı´adePontevedra(Galicia,NWSpain)andAVatthe
maximuminJuly,undertheinfluenceofnortherlywinds,whose eventsnormallylast for5–10days (Woosteretal., 1976;Fiu´za et al., 1982). As reported for other eastern boundary systems (Smith,1981),thevelocityandwidthofthecoastalupwellingjetis dependent on theduration and strength ofthe northerlywind events.Theupwellingfrontcanexpandovertheslope,orcontract towardthecoast,undernortherlyorsoutherlywindsrespectively. Severalstudieshaveshowntheexistenceofapolewardcounter currentcarryingwarmoceanicwaternorthwardsovertheslope (Frouinetal.,1990;HaynesandBarton,1990),andalsoan inner-shelfpolewardcurrentthatadvectswarmwaterafterthecessation ofupwelling(Sordoetal.,2001).Theseauthorsalsoshowedthis currentpromotedthealongshoretransportofHABs.
ThelocalupwellingpatternsofNWIberiaalsodependonthe coastal morphology, which is NW-SE oriented between cape Finisterre(Galicia)and418Nlatitude(Porto,Portugal)and NE-SW from that latitude until cape Carvoeiro. In Galicia, the coastlineisdeeplyindentedbyRı´asBaixasallowingadynamic exchangeofwaterbetweentheshelfandtheRı´as,leadingtoa local increase in phytoplankton productivity (Figueiras et al., 2002). The Rı´as also act as resonance boxes for HABs events (Escaleraetal.,2010).Tothesouth,thecoastlineisstraightand the shelf becomes gradually wider and flatter until F.Foz (40.28N),contributingtothethermalincreaseofwatercolumn stratification. This stratification is also enhanced by the dischargesof severalimportant rivers (e.g.Douro andMinho). The coast between 42 and 418N has been described as the sourceforarecurrentupwellingfilamentthat,dependingonthe wind conditions and its interaction with the offshelf eddies, developswithdifferentintensities(Relvaset al.,2007). 2.2. Dataandmethods
DinophysisacutaandDinophysisacuminatadatafromApril2004 untilDecember2013wereprovidedbytheGalician(INTECMAR) andthePortuguese(IPMA)HABsMonitoringProgramsandare fromtwoselectedsamplingstationswhere,overtheyears,these species maxima often occurred. Data on Mesodinium rubrum countsat Aveiro,fromMay 2004toDecember 2006andfrom April2008untilDecember2010,arealsopresented.Additional data of D. acuta from all the HAB monitoring stations to the northofLisbonwereusedtodetermineineachyeartheareaof the coast where blooms initiated and the maxima were observed.
StationPO-Bue´u(knownasP2inGalicianmonitoringprogram) is located at the southern side of ria de Pontevedra, Galicia (42821030.0000N; 8846020.0000W; 30m depth) and station AV-Mare´grafo is located at the entrance of ria de Aveiro (40838038.9500N; 8844055.1800W; 5m depth), a site always sampledonehourbeforehightidetomonitorshelfwatersthat entertheria(Fig.1).Watersampleswerecollectedwitha hose-samplerto15matPOandatthesurfacewithaNiskinbottleatAV. Inbothcases,thesamplesareconsideredtorepresenttheupper mixedlayerofthewatercolumn,resultingfromthetidalandwave mixing at the entrance of Ria de Aveiro, and the use of the integratedsamplerinPO.Allsampleswereimmediatelypreserved with Lugol solution. Subsamples of 25–50ml were allowed to settlebeforecounting.Thespecieswereidentifiedandcountedby usingtheUtermo¨hltechniqueandinvertedcontrastmicroscopy. Dailyseasurfacetemperature(SST)mapswereobtainedfrom theOperationalSea SurfaceTemperatureandSea IceAnalysis– OSTIA (spatial resolution 0.058, 5km) available from the MyOcean service. Meteorological parameters (surface wind components,photosyntheticactiveradiation–PAR)wereobtained from the ECMWF ERA Interim re-analyses (spatial resolution 0.2580.258,25km).
Bakun (1973)upwellingindices– UI(m3s1/100m)=
t
y/f – were calculated,based on the N-S wind component over the shelf (98W)and a latitudebetween AVandPO (41.58N). The wind stress was computed using the quadratic drag lawt
y=r
aCDjvjn
, where jvj is the wind speed (ms1),n
is the northward wind component (ms1),r
a is the air density (1.22kgm3)andCDisthedragcoefficient(0.0014).Usingthis formulation, the UI has the same signal as the NS wind component; therefore negative UI corresponds to offshore Ekmantransportdrivenbytheupwellingfavorableequatorward winds, while positive UI corresponds to convergence/down-welling drivenby polewardwinds.
ALagrangianmodel(RegionalOceanModelingSystem-ROMS) wasusedtotestseveralhypothesesofbloominitiation/detection afterpassivetransport.Particle-tracking experimentsuseda 3D hydrodynamic model configured toreproduce NW Iberia shelf circulation. Monthly maps of current component anomalies, relativetothedecadeaverage(2004–2013),werecomputedfrom NEMO model solutions distributed as the MyOcean ‘‘Atlantic-Iberian Biscay Irish-Ocean Physics’’ products. The decade-long datasetforNWIberiawasconstructedbymergingthe‘‘reanalysis’’ productfortheperiod2004–2011andthe‘‘analysisandforecast’’ productfor2012–2013.
AnEmpiricalOrthogonalFunction(EOF)analysisofOSTIASST data for theperiod 2004–2013 wasused to identify the main variabilitypatterns.TheEOFswerecomputedusingthesingular valuedecompositionroutineimplementedintheFerretsoftware (Hankinetal.,1996).Dependingontheauthor,slightlydifferent EOF terminologies areused.Here we usethetermEOF forthe eigenvectors Fi(x) and EOF coefficients ai(t) such that SST(x,t)=SUM[Fi(x)ai(t)]. To aidthevisualization ofthe relative magnitudeofthefirstthreeEOFs,that accountfor98.9%ofthe temporal variance, the eigenvectors and corresponding coeffi-cientswererespectivelyscaledbyfactorskand1/k,sothatallEOF coefficientsai(t)fallwithinsimilarranges.Theresultspresented below refer totheso-called ‘‘temporal EOFs’’in thesense that diagonalization of the covariance matrix was obtained by removingthetemporalmean.Toreducethecomputermemory requirements, the daily SST fields at 5km resolution were averagedto5-dayand1/148(7km)resolutionpriortotheEOF decomposition.TocomplementtheEOFanalysis,SSTanomalies over theinner-shelf (depthsup to 50m) between 39.58N and 438Nwerecomputedbysubtractingthedailymapsfromits10 yearmean.
Thefreshwatercontinentalrunoffwasestimatedbasedonthe sumof daily discharges fromdams of riversLima,Ca´vado and Douro madeavailablebyServic¸oNacionaldeRecursosHı´dricos (SNIRH,AgeˆnciaPortuguesadoAmbiente).
2.3. Timeseriesanalysis(2004–2013)
TostudytheconditionspriortoDinophysisbloominitiation,we defineabloomeventasthetimecorrespondingtoDinophysiscells concentrationlargerthan200cellsL1.Thisthreshold,currentlyin useastheearlywarninglevelbytheHABmonitoringprogramin Portugal,waschosenbecauseabovethislimittheconcentrationof Dinophysiscellsgenerallyincreasesandleadtoshellfishtoxicity levelsaboveregulatorylimits.Wealsousedthetermepicenterfor theareaofthecoastwherebloomsarefirstdetectedeachyear.In addition,asecondthresholdof103cellsL1wasusedtoanalyze theconditionsrelatedtothedevelopmentofeventswithahigh concentrationofcells.Thedifferentsubsetsweregeneratedand analyzed using the R statistical computing environment (R Development Core Team, 2011). Summary statistics of the Dinophysisoccurrencedates,SST,UIandPARwerecalculatedfor eachareaandfordifferentspeciesconcentrations.
2.4. Lagrangianofflinemodel
InordertosimulateDinophysispathways,anIndividualBased Model(IBM),similartothatusedintheNolascoetal.(2013a),was coupledtoaROMSconfiguration(Nolascoetal.,2013b)usingROFF (Carretal.,2007).ROFFisadrifter-trackingcodewhichsimulates particletrajectoriesfromstoredROMSvelocityandhydrological fields.
The drifter-tracking code simulates Dinophysis trajectories fromstoredROMSvelocityandhydrologicalfieldsusingahigh order predictor–corrector scheme to integrate the motion equationdX/dt=UROMS(X,t),with Xbeing theposition vector (x,y,z),andUROMSbeingthe modeled3Dvelocity vectorover time,givenaninitialconditionX(t0)=X0.Thisalgorithmisrun withatimestep,dt,of300s,whichisalsothetimestepusedto generatetheROMSfields,providedtoROFFasdailyaverages.In additiontotheadvectiongeneratedbythemodelvelocities,the particlemovementsincludedrandomvelocities in thevertical direction,whichwereusedtoparameterizeunresolvedturbulent processes.
3. Results
3.1. Mesoscaleandinterannualvariabilityoftheoceanographic conditions
The studiedperiod wascharacterized by a highinterannual variabilityofmeteorologicalandoceanographicconditions.2013 presentedtheweakest upwellingseasonof thestudieddecade, withupwelling almostabsentfromApriltoSeptember(Fig.2A, thick line, arrow 1). In contrast, 2007 showed an almost uninterrupted sequence of very strong upwelling events (Fig.2A,grayline,arrow3),fromMarchuntiltheendofAugust, withonlyatwoweekrelaxationperiodinJune(Fig.2A,thickline, arrow3).In2004,theupwellingwasalsoweakfrommidMayuntil theendofAugust,onlywithatwoweekeventofstrongnortherlies in early July. The year 2006 was characterized by very low upwellinginAprilandMay,followedbyweakupwellingepisodes untiltheendofAugust(Fig.2A,thickline,arrow2).Theremaining yearspresentedmoderateupwellingduringspringandsummer. Autumn2008deservesparticularattentionsince,contrarilytothe
Fig.2.(A)Dailyupwellingindex–UI–timeseries(grayline),5-dayrunningmean(thickline)anddailyregionalriverrunoff–RO(dashedline);(BandC)temporalEOF
coefficientsand(D–F)eigenvectorsofthefirstthreeEOFs,explaining98.9%oftheSSTvariance.Thetemporalcoefficientsarerepresentedaccordingtothecorresponding
eigenvector:(B)solidline(EOF1),dashedline(EOF2);(C)solidline(EOF3).Thedashedlinein(C)representstheSSTanomalyattheinner-shelf.Numbersinarrowsindicate
specialeventsdiscussedinthetext:1–noupwellingandpositiveSSTanomalyoffAV(EOF3);2–mildupwelling;3–strongspringandsummerupwelling;4–strongautumn
upwelling;5–highriverrunoff;6–dryyear;7–strongpositiveSSTanomaly(EOF2area);8–strongnegativeSSTanomaly(EOF3areas);9–thehighestpositiveSSTanomaly
downwellingepisodesnormallyobservedinthisseason,theperiod wascharacterized byverystrongupwelling,fromSeptemberto December(Fig.2A,thickline,arrow4).
Runoffdatashowthatthe2009–2010winterwasparticularly rainy(Fig.2A,dashedline,arrow5),withMarch2010beingthe wettest of the series. Autumn 2006, winter 2011 (January to March) and January to April 2013 also presented high river discharges.Thedriestyearswere2005and2012(Fig.2A,dashed line,arrow6).
The results of the EOF analyses, extracting the main SST variabilitypatterns,are presentedinFig.2B–F.Thefirst mode, which accounts for 96.9% of the variance, is related to the seasonal cycle of the cross[along]-shore SST gradient off west[north]Iberia.TheEOF1temporalcoefficient(Fig.2B,solid line)hasaclearannualperiodicityandEOF1hasanalongshore distributionoverWIberia,contrastingwiththenortherncoast (Fig.2D).Thelowervaluesofthismodeoverthewesternshelf showthattheamplitudeoftheseasonalcycleisloweralongthe coast, resulting from the cold coastal upwelling waters that occupytheshelfduringsummer.Theminimumvaluesarefound in the northermost tip of the region around cape Finisterre, defining the area of coldest summer temperatures. EOF2, accountingfor1.63%ofthevariance,showsmaximumpositive valuesovertheshelfcenteredat418N,southoftheGalicianRı´as (Fig.2E).EOF3,whichrepresents0.35%ofthevariance,followsa similarpatternoffWIberiabutwithlocalizedcoastalmaxima off Ria de Aveiro and the Galician Rı´as (Fig. 2F). Both EOFs presentconsiderableinterannualvariability(Fig.2Bdashedline andFig. 2C,solid line). The areaswith higher EOF values are thus interpreted as the locations with stronger interannual
fluctuations of the coastal processes driving the sea surface temperature, namely the summer coastal divergence/conver-gence.ThesimultaneousanalysesofEOF2andEOF3coefficients allows,for eachdate, tohighlight theperiods andeventsthat presentedmesoscaledifferencesorsimilaritiesbetweenthetwo samplingregions.
Someyearspresentperiodscharacterizedbyoppositesignalsof EOF2andEOF3coefficients,beingEOF3positive,meaningthatthe region off Aveiro was locally warmer than surrounding shelf watersingeneral,andconsequentlywarmerthantheGalicianRı´as further north.Examples ofthese events,focusing in theperiod fromtheendofspringuntiltheendofsummer,were:midJune– midSeptember 2005,May–midAugust 2008,midJune–August 2010andmidJuly–midSeptember2013(Fig.2A,e.g.arrow1).The oppositewasobservedduringseveralautumnperiodssuchasin 2008,whenwatersoffAveiroweremuchcooler(Fig.2C,arrow8) than October–November 2012, due to the abnormally intense autumnupwelling.
ThecomparisonbetweenEOFcoefficientsandtheSST anoma-liesovertheinner-shelfbetween39.58Nand438N(Fig.2C,dashed line) highlights the regional scope of the various events. The highestpositiveinner-shelfanomalieswererecordedinautumn 2006, corresponding mostly to the EOF2 pattern which also presents high coefficient values. Conversely, the lowest values (negative)inner-shelfanomalies inautumn 2008correspondto bothEOF2andEOF3,withastrongercontributionfromthelater revealingthattheveryintenseupwellinghasfurtheraffectedthe areasoffAveiro.Ontheotherhand,thesamecomparisonshows thatEOFdecompositioncanuncoverlocalizedfeaturesthatmight not be captured by the analysis of broader inner-shelf SST
D. acuminat
a
D. acuta
2004
2005 2006
200
7
2008
2009 2010
2011
2012 201
3
0
1000
2000
3000
4000
5000
300
00
600
00
900
00
Cells L
-10
1000
2000
3000
4000
5000
10000
20000
30000
40000
Cells L
-12004
200
5
2006
2007
2008 2009
2010 201
1
20
12
2013
AV- Aveiro
PO- Pontevedra
Bloo
m
thre
shol
d
Bloo
m
thre
shol
d
A
B
Fig.3.Interannualvariabilityof(A)Dinophysisacuminataand(B)DinophysisacutacellconcentrationatAveiro(AV)andPontevedra(PO)stations,from18May2004to31
anomalies. This is the case of the highest EOF3 summer2013 values,showingthattheeffectoftheveryweakupwellingseason mostlyaffectedthecoastalareasbetween408Nand418N. 3.2. InterannualvariabilityofDinophysisspecies
During thedecade,Dinophysis acuminatablooms weremore frequentalthoughlessintensethanthoseofDinophysisacutain bothstations (Fig.3 andAnnex1). At AV D.acuminata blooms reachedhigherintensitiesanddurationsin2007and2008thanin PO(e.g.9.4103cellsL1inAV,3103cellsL1inPO), decreas-inginintensityduringthefollowing years.In PO,D.acuminata blooms wereparticularly intense from 2010to 2012, reaching 27103cellsL1 in July 2010. During 2004, D. acuminata outbreaks appear to be the weakest of the decade (up to 103cellsL1),butthereisalackofdatabeforeMay.
AtAVstation,Dinophysisacutabloomedannuallyfrom2004to 2008, 2005 being characterized by exceptional blooms that reached 87103cellsL1 in September, when a historical maximum of the species in Portugal (14104cellsL1) was observed at a site located 10km to the south. A maximum (17.6103cellsL1)forthedecadewasalsoobservedinPObut detectedlaterinNovember.D.acutadidnotexceededthebloom thresholdlevelin2009andfrom2011to2012,whileinPOthe speciesdidnotbloomfrom2007until2012(Fig.3,Annex1).In 2010, only a short and weak peak (0.5103cellsL1) was observedatAV,inSeptember.Ingeneral,thebloomsweremore intense(slightly lowerin 2013) and lengthy in AV than inPO, confirmingAVastheepicenterofD.acutablooms.Evenin2006, whena peak of 3103cellsL1wasregistered inPO, a cruise coveringthePortugueseshelfbetweenF.FozandAveiroregistered 9103cellsL1ofD.acutaatmidshelf(Silvaetal.,inprep,data notshown).In2007and2008D.acutabloomswereonlyobserved atAV.In2007theylastduringtwoweeksinOctober,whilein2008 thebloomsinitiatedinMayandpersisteduntilmidAugust.Aftera decline in bloomevents from 2009 to 2012, in 2013 D. acuta reached4.6103cellsL1duringa shorttwo weeksepisodein midSeptember.InPOthebloomssuddenlyoccurredtwoweeks later,withaconcentrationof5103cellsL1andlastedforone month.
3.2.1. InterannualdistributionofDinophysisandMesodinium AllcountsofMesodiniumcellsusedinthepresentworkwere identifiedasMesodiniumcf.rubrumregardlesstheirsize.However, consideringtheseveralMesodiniumspeciespresentedrecentlyby Garcia-Cuetosetal.(2012),itispossiblethatourcountsincluded atleasttwoofthem,asmallerformthatmustbeM.rubrumanda
largerformthatprobablyis thenewspeciesMesodiniummajor, bothspeciesaremarine,pelagicandphototrophic,witha reddish-browncolor.
The available Mesodinium data atAV (Fig. 4)show that the ciliate was present from spring to autumn during the periods 2004–2006 and 2008–2010. Mesodinium peaks often coincided withDinophysisacuminata maximaandboth species werevery persistent, although Mesodinium normally remained after D. acuminata disappearance. The only exception was observed in 2005,coincidingwithDinophysisacutahighestpeak.ConcerningD. acuta, the maximum concentrations were not coincident but appearedrelatedtothepeaksofMesodinium.Theinitiationofthe intensebloomsofD.acutaonthe1stAugust2005,occurredthree weeksafterthehighestmaximumofMesodiniumregisteredinthe presenttime-series(44103cellsL1on 4thJuly).Atwoweek delaywasalsoobservedin2008,whentheinitiationofD.acuta outbreaksoccurred on 7 May (Annex1), aftera first moderate concentrationpeakofMesodinium(1.3103cellsL1)on21April. AsubsequentD.acutamaximum,reaching16103cellsL1on14 July,wasobservedthreeweeksafterthemaximumofMesodinium (12103cellsL1)observedon23June.In2004,despitethelack ofdatabeforeMay,theinitiationofthefirstD.acutabloomon27 JulyalsooccurredtwoweeksafteramaximumofMesodiniumthat reached8.2103cellsL1on12July.
3.2.2. Timeofspeciesfirstdetection,bloominitiationand maintenance
The time of species detection and bloom initiation in both stations,aswellastheperiodsofbloomoccurrence(>200cellsL1 and>103cellsL1)forthedecade,issummarizedinFig.5.Thedate of the first detections within the years, for each species and samplingstation,ispresentedinAnnex1.Dinophysisacuminata was normally detected earlier than Dinophysis acuta, between JanuaryandMay,insomeyearsfirstinAV,othersinPO,butinboth stationsmostdetectionsoccurredinanarrowwindowbetween thelastweekofFebruaryandthelastweekofMarch.Inturn,the onsetofD.acutaismarkedbyalargeinterannualvariability.The firstcellsintheyearcanbedetectedfromtheendofFebruaryand
Fig.5.JuliandayboxplotsofD.acuminataandD.acutaeventsatPOandAVduring
the2004–2013decade:(A)thefirstdetectionoftheyear,(B)firstbloomoftheyear,
(C) all blooms (>200cellsL1) and (D) all high concentrated blooms
(>103
cellsL1
).ThelabelsalongtheXaxisindicatethespecies,‘‘act’’refersto
D.acutaand‘‘acm’’toD.acuminata,observedinstationsAVandPO.Thenumbers
belowthelabelsindicatethenumberofsamplesineachdatasubset.
Fig.4.InterannualdistributionofD.acutaandD.acuminatacellconcentrationand
oftheirpotentialprey,theciliateMesodiniumcf.rubrum,attheAVstation,forthe
mid March, at AV and PO respectively, until the beginning of Octoberinbothstations(Annex1,Fig.5A).In2007and2010no cellsofD.acutawereobservedinPO.Withtheexceptionof2011,D. acutawasnormallyobservedearlierinAVthaninPOandtwice,in 2005and2006,thespecieswasdetectedalmostsimultaneouslyon bothstations.
Fig.5B–Dhighlightsthesequenceofbloomeventsthroughout theyear.Excluding2008,the1steventseachyear(Fig.5B)began with Dinophysis acuminata in PO (1st week of March) and afterwards in AV, followed by Dinophysis acuta blooms in AV (1stweekofMay)andafterwardsinPO.RegardingD.acuminata (Fig.5C),theinter-quartile(25–75%ile)ofalleventsinPOoccurred from the 1st week of May until the 3rd week of August (123<Jday<235,median: 173,where JdaymeansJulianday). In AV, the events were in general observed two weeks toone month later and lasted from mid June until mid September (168<Jday<254, median: 202). In PO, 50% of the high concentratedbloomsofD.acuminatawereobservedearlierthan the 25%ile, from the 3rd week of April until the end of July (108<Jday<203,median:147).InAV,theperiodisreduced,in 50%oftheevents,aboutonemonth(173<Jday<222,median: 193).
In AV, the inter-quartile of Dinophysis acuta events (>200cellsL1)showstheyoccurredfromthe3rdweekofJuly until the end of September (206<Jday<269, median: 206). Highlyconcentratedbloomsoccurredduringthesameperiodwith aslightreductioninlength(211<Jday<266)withthemedianin thefirstweekofSeptember(Jday:245).InPO,mosteventslasted from late September until the 1st week of November (271<Jday<307,median: 290)andtheperiodofbloomswith an elevated number of cells was shorter (278<Jday<297, median:287).
3.3. Dinophysisdistributionandphysicalconditions
InFig.6AandBalltheoccurrencesinAVandPOofeachspecies together with the averaged variations of SST, PAR and UI are represented.Dinophysisacuminatablooms,occurringearlierinthe
yearthanDinophysisacuta,appeartobemorerelatedtotheannual cycleofPARthanthebloomsofD.acutawhicharemoreassociated with the average annual SST cycle. Fig. 6C shows that most (25–75%ile) D. acuminata blooms occur in the PAR range of 6.0–9.4105Jm2inPOand7.8–10.8105Jm2inAV),although thedifferentmediansobservedforAVandPOindicatethatPARis closelyrelatedtothetimeoftheyearthebloomsoccurineachof thesites.Inturn,D.acutabloomsoccurwithindistinctPARranges atAV(9.6105Jm2)andPO(8.0105Jm2)andaretherefore notrelatedtoPAR.
Inrelationtotemperature(Fig.6D)Dinophysisacutaeventsin AV,andthefirsthighintensityeventsinPO(Fig.7C)werefound aroundthesameSSTmedian(178C),althoughsubsequentblooms in POoccurunderlowertemperatures (Fig.6D,medianaround 168C) because they occur later in autumn. Most Dinophysis
Fig.6.Cellconcentrationsof(A)D.acuminataand(B)D.acutaatAveiro-AV(4)andPontevedra-PO(&)superimposedonlow-pass10-year(2004–2013)dailymeanSST(8C,
solidline)/PAR(Jm2
,dashedline)andUI(m3
/s/100m,dash-dottedline)annualcycle(rightaxis).Largesymbolsindicatecellconcentrationsabove200cellsL1
.NegativeUI
valuesrepresentupwelling(horizontaldottedlineindicates UI=0).Boxplotsshowingthemedian,percentile25%and75%andoutliersofthebloom conditions
(>200cellsL1)foreachofthespecies:(C)PAR,(D)SSTand(E)UI.ThelabelsalongtheXaxisindicatethespecies,‘‘act’’referstoD.acutaand‘‘acm’’toD.acuminataobserved
instationsAVandPO.Thenumbersbelowthelabelsindicatethenumberofsamplesineachdatasubset.
Fig.7.D.acuminata(acm)andD.acuta(act)conditionsboxplotsatAVandPO
stationsshowing25and75%ile,medianandoutliersof:(AandB)SST(8C)and(C
andD)UI(m3
/s/100m),onthe15daysprecedingblooms.Leftcolumnrefersto
conditionsbeforethe1stbloom(>200cellsL1
)intheyearandtherightcolumnto
conditionsbeforethefirsteventwithahighconcentrationofcells(>103
cellsL1
).
acuminatabloomsin POoccurin adifferenttemperaturerange thanin AV. The numberof D.acuminata blooms withelevated concentrationsseemtodecreaseduringthemaximumintensityof upwelling(i.e.Fig.6A,aroundJday190–230),whilebloomsofD. acutatypicallyoccurattheendoftheupwellingseasonwhenthe frequency and intensity of the convergence episodes increase. Fig.6EhighlightsthatmostbloomsofD.acuminatainbothstations and D. acuta in AV develop under mild upwelling conditions althoughinPOtheyarefavoredbydownwellingconditions. 3.3.1. Conditionsforfirstbloominitiationandintensification
ThefirstannualbloomsofDinophysisacuminatainitiatewithin awideSSTrange(138–178),mostlybelow138CatPOand158Cat AV (see median in Fig. 7A). However an increase of 28C is apparentlyneededbeforethedevelopmentofeventswithahigh concentrationofcells,whichindicatesthetemperatureconditions favoringthespeciesgrowthoranaccumulationofnearshorecells duetocoastalconvergenceconditions.Onthecontrary,boththe initiationofbloomsorthedevelopmentofelevatedconcentration ofDinophysisacutacellsoccurbetween168Cand188Cwiththe medianofalleventsinAVorPOaround178C.
Regardingtheupwellingconditions(Fig.7BandD),eachyear outbreaksofbothspeciesinitiatedaftermildupwellingconditions, althoughthefirstintenseevents of Dinophysisacuminatain AV requirestrongerupwellingwindswhilethoseeventsofDinophysis acutain POoccurredafterdownwellingconditions (Fig.7D),as mentionedabove.
3.3.2. Oceanographicconditionsforthe1stbloomoftheyear Theresultspresentedintheprevioussectionsshowthatthe firstblooms(>200cellsL1)oftheyearoccurunderarelatively wide range of oceanographic and meteorologicalconditions. In addition,thedisappearanceofDinophysisacutabloomsforseveral yearssuggestthatabnormalconditionsleadtomajorchangesin the geographical location of the population cores. To further explorethesourcesofinterannualbloominitiationvariability,an elementary exercise to predict the 1st bloom event, for each species/station, wascarried outusing theabove results.Italso allowsidentificationofpossibleroutestoestablishcapacitiesfor forecasting Dinophysis events off NW Iberia. The prediction exercisewasperformed by identifyingthe simultaneous occur-rence of thefollowing conditions,whose dataare presentedin Table1:
i)Jdaybetween25%ileandmaximum; ii)SSTandUIbetween25and75%ile;
iii)existenceofDinophysiscellsinwatersamplesduringthetwo previousweeks.
Table2showsthatinatotalof37cases,thesimpleuseofthe overlapping conditionspredicted in 16 cases(43%) the appear-ance/lackofbloomsintheweekbeforeorthesameweekoftheir actualdetection/absence(i.e.14<n8days<5andpredictionof noblooms),atimelagconsideredeffectiveforearlywarningofan eventualshellfishban.Theanalysisalsopredicted3falsepositives ofDinophysisacutainAV.
OnehalfoftheDinophysisacuminatabloomsinPOcouldalsobe anticipatedwithadelayofonetotwoweeks.Theresultsofthe exercise also showedthat D.acuminata outbreaksare theless predictableusingasimpleparameterrangeintersection,especially inAVandinparticularyearssuchas2006,2010(bothatAVand PO) and in 2011. These results underpin the hypothesis that advectiveeffects,nottaken intoaccountinthis exercise,are at least equally important as the local conditions to predict Dinophysisoutbreaks.
Table2
PredictionofDinophysiseventsatPOandAV,withindicationofthen8ofdaysbefore(-)orafter(+)theobserveddateofbloominitiation.Bloomorpredictiondatesare
respectivelyprecededby‘‘B:’’or‘‘P:’’.Thecode‘‘nopred’’marksthattheexercisesuccessfullypredictstheabsenceofblooms,and‘‘predno’’marksthattheexercisepredicted
theabsenceofbloomsthatinrealityoccurredattheindicateddates.
Table1
Selectedinitial(ini)andfinal(fin)conditionsforthepredictionexerciseofthefirst
D.acuminataandD.acutabloomoftheyearinAVandPO.Conditions:JDay(julian
day),SST(8C)andupwellingindex–UI(m3/s/100m).
Sp/site JDay_ini JDay_fin SST_ini SST_fin UI_ini UI_fin
D.acum/PO 69 198 12.6 16.1 753 87
D.acum/AV 82 254 13.9 17.3 575 336
D.acut/AV 146 281 16.6 18.1 601 16
Fig.8.Ten-yearmonthlymeansurfacecurrents(0–30m)offNWIberia(leftcolumn,July-toptoOctober-bottom)andmonthlysurfacecurrentanomaliesforselectedyears
relativetothe10-yearmean.ShadedareascorrespondtoabsoluteNScurrents/anomalieshigherthan0.05ms1.PolewardandequatorwardNScurrents/anomaliesare
3.4. Modelingsimulationsforspeciestransportbetweenthetwo samplingsites
To explore the role of the oceanic circulation on bloom dynamicsandthespeciesadvectionbetweenthetwomonitoring sites, two approaches were developed based on different time scales.Firstly, monthly anomalies of surface (0–30m) currents relativetothe10-yearmeanwerecomputedtoinvestigateifthe circulationstructureswerefavorabletothealongshoreor cross-shoretransportandtohighlighttheroleofparticularfeatures,such ascounterflows,inshorecurrentsorupwellingfilaments.Secondly, event-scale numerical experiments were carried out using a
Lagrangiantransport modeltotest ifthetime lagbetween the bloomdetectionatthetwomonitoringsitescanbeexplainedby advectiveprocesses.
Themonthlyanomaliesforselectedyears(2005,2006,2008, 2013)andmonths(July–October)arepresentedinFig.8,together withthe10-yearmeanofmonthlycurrents,todrawattentionto years/monthswhenthegreatestDinophysisacutabloomsoccurred andwhenthecirculationfeaturesappearedtoexertasignificant impactonthoseblooms.Thedecadalmonthlymeans(Fig.8,left) clearlyillustratethatcirculationintheupper30mismoreintense overtheshelfinJulyandAugust,withastrongequatorwardjet overthenarrowershelfnorthof41.28N,becominglessintenseand
Fig.9.Lagrangianmodelresultsfor4bloomeventsshowing(AandB)twopoleward(deploymentatAV)and(C–F)twoequatorward(deploymentatPO)particletransport
experimentswithdifferentverticaladvectionconstraints:(A–D)theverticalmovementisunconstrainedandtheparticlesareadvectedaccordingtothe3Dcirculationfield;
(EandF)theverticalmovementisconstrainedsothattheparticlesaremaintainedwithintheinitial(7.5–12.5m)depthrange.Symbolsrepresentweeklyparticlepositions
morespatiallyvariableinSeptemberandOctoberwhenamean polewardcurrentestablishesovertheslopeandtheinner-shelf.It is also worth noting the flow branching at the Aveiro canyon (40.68N)wheretheshelfwidens,creatingarelativelysmallarea southofthecanyon(40.4–40.68N)wherethecurrentsarehigh overtheslopeandtheinnershelf,witharelativeminimumat mid-shelf.
Themonthlyanomaliesshowthestronginterannualvariability ofthesurfacecurrents.ThisisparticularlynoticeableinOctober whenthegenerallyweakinner-shelfpolewardcurrentwasmuch more intense and covered thewhole shelf in 2013, being also strongin2006butrestrictedtodepthsshallowerthan100m.This patternistotallyreversedin2008whentheout-of-seasonstrong upwelling winds drove intense equatorward currents over the entire shelf. The results also show the frequent occurrence of oppositecurrentanomaliesintheouter-shelfsouthofthePorto Canyonat41.68N,speciallyinJuly–August2005,September2006 andJuly2008.
LagrangianparticlemodelexperimentsarepresentedinFig.9. Theeventswereselectedbasedontheobservationofacleartime lagbetweenthebloomoccurrenceatAVandPOmonitoringsites. TwoeventscorrespondtothepolewardadvectionofDinophysis acuta blooms as inferred from the observation of high cell concentrations,firstatAVand3weekslateratPO(2005,2013), and two cases correspond to the equatorward advection of DinophysisacuminatabloomsthatwerefirstrecordedatPOand 3weekslater at AV(2006,2008). Several experimentswere carriedoutusingdifferentinitialparticlelocationsandvertical positions.Ingeneral,ahighersensitivityofmodelsolutionswas observedwithrespecttotheverticalparticledistributionduring southward/activeupwelling(northerlywind)periods.To illus-tratethiseffect,twosolutionsforthe2006and2008D.acuminata
bloomadvectionpatternsareshowninFig.9.Inthefirstcasethe particlesaredeployedatdepthsbetween7.5mand12.5mand areadvectedaccordingthe3Dcirculationfield(Fig.9CandE).In thesecondcasetheparticlesweredeployedatthesameinitial locationsanddepthsbuttheirverticalmovementisconstrained sothattheyaremaintainedwithinthesame(7.5–12.5m)depth range(Fig.9DandF).
Asexpectedinaregionwherethecirculationismainlydriven by coastal upwelling, the trajectory patterns are very distinct duringcontrastingwindforcingphases.Inthetwodownwelling/ polewardevents,theDinophysisacutacellsaretransportedmainly over the inner-shelf and the particles tend to have similar trajectories and converge to shallower depths (Fig. 9A and B). Onthecontrary,duringtheactiveupwellingphasethetrajectories aremoredisparateanddivergentwhiletheparticlesareadvected equatorward and offshore(Fig. 9Cand D). The resultswithno verticaldistributionconstraintshowthatitisveryunlikelythat Dinophysis acuminata blooms at AV might result from inocula advectedfromPO.However,whentheverticalparticlemovement is constrained so that theyare maintained within a 5mlayer centeredat10m, someparticlesfollowtrajectoriesclosetothe coastandarriveatAVwithatraveltimethatiscompatiblewiththe observed time lag between the bloomdetection at POand AV (Fig.9EandF).
3.5. InterannualchangeofDinophysisacutaepicenterandmaxima attheNWcoastofPortugal
Asdescribedabove,Dinophysisacutabloomswerenotobserved inPOfrom2007to2012,andalmostdisappearedinAVfrom2009 to2012,whereonlyasinglepeakofoneweekwasobservedin 2010. Despite a possible lack of the necessary conditions for
Table3
growth in AV, we hypothesized that the species could have changeditscenterofdistributiontoalocationfurthersouthinthe Portuguesecoast.InTable3itisconfirmedthat,withtheexception of2010whenonlyashorteventon the1stweek ofSeptember occurredinAVarea,D.acutabloominitiationandmaximamoved southwards until 2013, with substantial blooms reaching 6103cellsL1inLisbonBayinAugust2012.In2014,D.acuta bloominitiationandmaximaoccurredagaininAV.
4. Discussionandconclusions
4.1. Dinophysisdistributioninthewatercolumnandrelationship withMesodinium
AlthoughDinophysispopulationsoffIberiaandelsewhereare relatedtostratificationconditions(GEOHAB,2008),stratification itselfdependsondifferentfactors.Also,eachofthespeciesshows preferencesontheirpositioninthestratifiedwatercolumnthat haveconsequencesontheirtransportcross-shelfandalongshore asdiscussedbelowinSection4.4.
OntheNWIberianshelfDinophysisacuminataandDinophysis acuta are distributed on the wind driven layer and reflect the oscillations of its base (Moita and Silva, 2001). However, D. acuminataappearstodependmoreonlightavailabilitysincethe maximaaregenerallyobservedclosertothesurfacethanthoseof D.acutanormallyfoundinthetopofthepycnocline(Regueraetal., 1993;Palmaetal.,1998;Velo-Sua´rezetal.,2009). Bothspecies werefoundtoconcentrateinthinlayers,D.acuminatainsurface warmwaterpatchesat0–4mdepthinRı´adePontevedra( Velo-Sua´rezetal.,2008)andD.acutaat18–20minthecoastoffF.Foz (Moitaetal.,2006).ThereisalsoevidencethatinRı´adeVigoD. acuminata undertake diel migrations between the surface and 10mdepth(Villarin˜ oetal.,1995),activelyexploitinghighlevelsof irradianceduringthedayand havingaccesstohighernutrients concentrationduringthenight.CuriouslyMesodiniumrubrum,not knownatthetimetobepreyforDinophysis,waspartofthesame assemblage and characterized by a similar vertical migration behavior.
Velo-Sua´rezetal.(2014)highlighted,forthefirsttimeinIberian waters, the relationship between Dinophysis acuminata bloom initiation and maintenance with the presence of Mesodinium rubrumin Rı´a dePontevedra, although theyalso reported that thereisnosinglemechanismthataccountsforallthestagesofa bloom.Ourresultsshowthat,inAveirowaters,D.acuminataand Mesodiniumalmostalwaysco-occurredduringsixyearsofstudy but theinitiation of D. acuminata blooms are not immediately relatedtothepreyoccurrencesincetheciliategenerallyoccurs much earlier every year (e.g. 2010, Fig. 4). However, that dependencymayoccurfurthernorthinPO,asobservedin2007 byVelo-Sua´rezetal.(2014),whenthebloomsinitiateearlierthan inAVandbeforetheirpossibleadvectionsouthwards.Regarding Dinophysisacuta,thehigherpeaksoccurred2–3weeksafterthose ofMesodiniumaddingfurtherevidenceofarelationshipbetween thespeciesatAV,butalsothattheirgrowthisfavoredbysimilar environmental conditions. Nevertheless, the lack of D. acuta bloomsin2009and2010didnot correspondtotheabsenceof theciliate.
Previousfieldstudiesonthetime-laggedcorrelationsbetween MesodiniumandDinophysisshowquitevariableresults,7daysin thecaseofMesodiniumrubrumandDinophysisacuminatainKorean waters (Yih et al., 2013). Lags from 0 to 60 days between MesodiniumandDinophysisovumandevidencethatinsomeyears nopredator-preyrelationshipexistedbetweenthetwoorganisms werealsoreportedintheGulfofMexico,underconditionssimilar totemperatewatersand cultures(Harredand Campbell,2014). Future studies off Iberia should include the separation of the
differentMesodinium species inmonitoring counts aswell asa moredetailedstudyofprey-predatorrelationships,includingthe delayinbloominitiationatdifferentlatitudes.
4.2. InterannualvariabilityofDinophysisbloomsanddevelopmentof anomalousconditions
AsobservedinpreviousdecadesoffIberia(Regueraetal.,1995; Palmaetal.,1998;Pazosetal.,1995),inthisstudy(2004–2013) Dinophysis species exhibited a great interannual variability, characterizedbyyearsdominatedbyoneofthespecies,oryears when both species occurred simultaneously.The present study showedhow meteorological anomalies are importantin deter-miningeachyear’sdynamicsofIberianDinophysisevents.
Contrasting climatological conditions were highlighted by Escalera et al. (2006) for theselection of Dinophysis acuta and Dinophysis acuminata in Rı´a de Vigoin previous years, namely thermohalinestratification combinedwith persistentupwelling conditionsfavoredD.acuminatain2002,whileaveryhotsummer withthelowestUIofthepast50yearsfavoredD.acutainlate summer2003.Thepresentworkshowsthatthesefindingsbecome morecomplexwhenwejoinseveralyears.In2005,forexample, when D. acuta showed an historical maximum in 30 years of monitoringattheAveirocoast,despitethisbeingaverydryyear (Fig.2A,dashedline)withhighairtemperatureanomaliesoverthe mainland (see http://www.ipma.pt/pt/oclima/monitorizacao/ index.jsp?selTipo=m&selVar=tx&selAna=an&selAno=2005), SST only showed small positive anomalies on that coast during summer, as highlighted byEOF3 (Fig. 2C). In fact, there wasa persistency of upwelling favorable winds, higher than in other yearswithlowerblooms,forexample2004and2006(Fig.2A,thick line). The D. acuta bloom season of 2005 in NW Iberia was discussedbyEscaleraetal.(2010)whosuggestedthatthesudden occurrenceofoutbreaksinRiadeVigowasduetoanalongshore transportofsouthernpopulations.Fig.8alsoshowsthat,inthat year, the outer-shelf off AV was occupied by a poleward flow bounded by two southward upwelled water branches: one occupying the inner-shelf and another over the slope. The recirculation between the coastal flow and the mid-shelf may have acted as a pool for D. acuta growth during August and September in Aveiro. It is likely that D. acuta blooms were concentrated and distributed in thepycnocline in that pool as observed in 2003 (Moita et al., 2006), and were introduced inshorewardsinAVstationbytheupwelledwaters.
Theatypicalpredominanceofupwellingwindsduringwinter andpositiveSSTanomaliesweresuggestedbyDı´azetal.(2013)to favor the early detection of cells and initiation of Dinophysis acuminatabloomsinPO,in2002and 2012.Ourresultsconfirm thesefindingsbutindicatethatveryhighriverdischargesbefore the winter upwelling events, as observed in 2010 and 2011, removethespeciesfromthecoastalareaanddelaytheinitiationof blooms, despite their earlydetection. This proposed impact of anomalousrunoffdoesnotcontradictseveralworksthatshowed the importance of haline stratification associated with river plumes in the dynamics of D. acuminata blooms (Dı´az et al., 2013;Peperzaketal.,1996;Velo-Sua´rezetal.,2010),arelationship also detected in southern Portugal when D.acuminata intense blooms(7104cellsL1)appearedrelatedtotheGuadianariver plumeinJanuary1998(unpublishedmonitoringdatafromIPMA). Themostseveremeteorologicalanomaliesobservedduringthe decade were, however, associated with the intensification of upwellingfavorablewindsin2007and2008thathavedisturbed thenormalonsetanddevelopmentofDinophysisbloomsinthetwo areas. The veryintense upwelling conditions observed in 2007 duringthewholespringandsummer(Fig.2A,thickline)seemto havehadimportantconsequencesontheavailabilityofsufficient
innoculafortheinitiationofDinophysisacutabloomsduringthe followingsevenyearsinPO,conditionswhichonlyre-appearedin 2013.Thatanomalyalsodelayedthe1stbloomofD.acutainAV untilOctober2007,andtherewerenoconditionsforitsgrowthor advectionintoPO.In2008,similarconsequencesoccurredinAV afterthewholeautumn2008characterizedbyanomalousintense upwellingwindsandsouthwardcurrents(Figs.2and8)insteadof the typical downwelling events. The strong equatorward and offshorecurrentsseemedtohavesweptD.acutabloomsfromthe AVsystemuntil2013,despiteasmallone-weekoutbreakobserved in2010(Table3).Inthoseyears,D.acutapopulationswerepresent inthesystembutwerenotabletoreachathresholdconcentration for bloom initiation. The blooms reappeared in 2013, after an anomalousyearwithoutupwellingduringallsummer,andwith thehighestpositiveSSTanomalyintheAVarea(Fig.2C,arrow9). InOctobertherewereconditionsfortheirpolewardtransportfrom AVintotheGalicianRı´as(Fig.9B).
4.3. Bloominitiationandprediction
Dinophysis outbreaks were observed from mid February (Dinophysis acuminata in 2008, AV)until mid December (Dino-physisacutain2005,PO),atimewindowslightlywiderthanthat observedinpreviousdecadesofftheNWIberian,FrenchandIrish coasts (Pazos et al., 2005; Xie et al., 2007; Raine, 2014). Unexpectedly,thetemporal windowofeach ofthetwospecies wasmoresimilarwiththeirdistributioninFlødevigenBay,onthe Skagerrak coast of Norway (Cembella et al., 2005), with D. acuminata blooms extended in time, and D. acuta events that mainlydevelopedduringthesecondhalfoftheyear.
The different behavior of the NW Iberia populations of Dinophysis acuminata and Dinophysis acuta, both in space and time,stronglysuggeststhatthedevelopmentofpredictivetools mustconsiderthespeciesindependently,sotheapproachfollowed bySinghetal.(2014),whereallDinophysisspeciesareconsidered together,isnotappropriateforNWIberia.Wealsowerenotyet abletodeterminetemporalrelationshipswiththepresenceofits prey,Mesodiniumrubrum.
In this work, we useda combination of simple overlapping conditionsofreadilyaccessiblevariables,likeSSTandUI,together with the knowledge of the periods most favorable for bloom initiation,tolearnhowfarsuchanelementaryapproachcouldbe usedtopredictthefirstbloomeventseachyear.Reasonableresults (Table 2) were obtained in anticipating the first Dinophysis acuminataeventsinPOandwithDinophysisacutainAV,curiously theareasconsideredtobethedistributioncentersofthespecies. From2007until2012,thenonexistenceofD.acutabloomsinPO wasalsopredictedjustbecauseweimposedtheconditionthatthe presenceofcellsisrequiredbeforebloominitiation.Theblooms werenotanticipatedintwocases,2007(Dinophysisacuminatain PO)and 2013,(Dinophysis acutainPO) apparentlyforthesame reason.Thebloomsinitiatedtoorapidlysuggestingthe popula-tionscouldhavegrownelsewhereandwereadvectedtothearea. Thishypothesis is consistentwithlagrangian model resultsfor 2013whenD.acutabloomsweremostlikelytransportedfromAV intoPO.In2007,thefirstoutbreakofD.acuminataintheRı´ade Pontevedra (only in mid June, during an upwelling-relaxation event) was well discussed by Velo-Sua´rez et al. (2014) and appearedto betriggered byheterotrophic feeding concomitant withthefirstpeak intheseasonof Mesodiniumcf.rubrum. Our resultsfurthersuggestthatthereporteddelayonthedevelopment ofMesodiniumin2007wasduetotheobservedanomalyrelatedto highpersistentandintenseupwellingwinds.
Bloominitiationofbothspecieswaspredictedtooearlyin2010. Inthatyearriverrunoffwasthehighestoftheseriesduringthe interquartileperiodofbloominitiationandwecanconcludethere
werenoconditionsforthespeciesgrowth.Asimilarreasoncanbe evokedforthedelayinDinophysisacuminata initiationatAV in 2011.Salinityisthereforeavariablethatshouldbeintroducedin futureexercises.InthecaseoftheearlypredictionofD.acuminata in2006,wewerenotabletohaveanexplanationbasedonthe available data, including the presence of Mesodinium cells in stationAV.
4.4. BloomconnectivitybetweenPOandAV
Previous studies have presented evidence of a northward transportofHABsintheNWshelf(Sordoetal.,2001;Escaleraetal., 2010;Pitcheretal.,2010).Howeveritcanalsobearguedthatthis apparent transport reflects poleward changesin the conditions favoring these species rather than an effective poleward cell advection (Pitcher et al., 2010), or that downwelling is the oceanographic process that triggers a positive selection and sudden dominanceof autochthonouspopulationsof dinoflagel-latesintheGalicianRı´asBaixasinautumn(Crespoetal.,2006).The modelsolutionsshownhereforDinophysisacutatransportfromAV to PO, in late October 2005, support Escalera et al. (2010) suggestionsthatthecellincreaseinRı´adeVigo,intheabsence ofcellulardivision,wasduetoanalongshoretransportfromthe coastofPortugal.Inthesameway,inOctober2013,thesudden outbreakofD.acutainPO,afteracompleteabsenceofcellsduring thewholeyear,canbeexplainedbythebloomtransportfromthe AV region asdemonstrated by themodel. These events, which originated suddenbans in Galician mussels harvesting showed howimportantwillbetokeepweeklyexchangeofinformationand early warning between monitoring programs of neighboring countries.
TheconnectivitybetweenAVandPOinrespectofthepoleward transportofDinophysisacutaismorecomplexwhendealingwith thesouthwardtransportofDinophysisacuminatafromPOtoAV. The selectedexamplesweredeveloped duringactiveupwelling periods, when transport is offshore and southwards. The first studies on coast-wide connectivity patterns off W Iberia were carriedouttoinvestigatethecommonshorecrabCarcinusmaenas recruitment on western Iberia coast (Nolasco et al., 2013a,b, followingCarretal.,2007),andthatexpertisewasappliedinthe present work. The connectivity between PO and AV was demonstratedtobepossibleiftrackingparticleswereforcedto stayina5mlayerat10mdepth(2.5m).Thiswaysomeparticles remain closeto thecoastandtravelsouthward reachingAV.This resultcan beappliedto D.acuminataifthe specieshasan active behaviorinthewatercolumnmaintainingitsverticaldistributionata given depth. Itstrophicrelationship with Mesodiniumrubrum,an activephototrophicswimmer wellknown ascapable tohavediel verticalmigrations(CrawfordandLindholm,1997),reinforcesthat possibility. The evidence of D. acuminata and M. rubrum diel migrationsbetween10mandthesurfacewasalreadyobservedin Rı´adeVigo(Villarin˜oetal.,1995)andencouragefurtherstudieson theDinophysispositioninthewatercolumnalongtheDielcycleand onthedifferenttransportpathways accordingtodifferentvertical migrationscenarios.
4.5. LatitudinalchangesinDinophysisacutaepicenter
Theobservedchangesoftheupwellingpatternsthatoccurred in spring/summer 2007(persistencyof veryintense upwelling) and in autumn 2008 (intense autumn upwelling instead of downwelling) resulted in important latitudinal shifts of the Dinophysis epicentersoff NW Iberia.Thoseabnormalupwelling conditions changed the Dinophysis acuminata bloompattern in 2007and2008bymovingfromPOintoAVthebloominitiationand maxima.The bloomepicenterwasre-established in POonly in
2009.TheyalsostronglyaffectedtheepicenterofDinophysisacuta. Firstly,thechangesintheupwellingpatternsinterruptedforseven yearsthebloomsoreventhespeciesdetection(e.g.2007and2010) inPO.Theyalsoshiftedthespeciesepicenter,locateduntil2008on thecoastofAveiro-F.Foz,tomorethan200kmsouthwards,leading to unexpected intense blooms that initiated and spread from LisbonBayin2012and2013.In2013,anyearcharacterizedbythe lack of upwelling and, consequently, by the intensification of polewardcoastalcurrents,thebloomsinitiatedinLisboninJuly, reachingAVinSeptemberandPOinOctober.Afterthisexceptional year,D.acutaoutbreaksanchoredagaininAVwheretheinitiation andmaximawereobservedin2014.Changesintheepicenterof DinophysisspecieswerealsoreportedintheNorthseaassociated withachangeinwindpatternsinalargertemporalscalethanour study.Edwardsetal.(2006)usedCPRdatatodescribethat,over thelastfourdecades,theDinophysisepicenter(referredtoascenter ofdistribution)intheNorthSeahasmovedeastwardstowardthe coast of Norway in association with the establishment of predominantsoutherlywinds.IntheWcoastsofIreland,Shetland islands,andFrancechangesinwindpatterns,inparticularperiods dominatedbywesterlywindsinsteadoftheusualsoutherlies,gave risetorapidDinophysisincreasesinsidethebaysorinotherareasof thecoast(Raine,2014;Whyteetal.,2014;Batifoulieretal.,2012; Dı´azetal.,2013),althoughchangesinthedistributioncentersof Dinophysiswerenotdiscussed.
Thepresentworkused10yearsoftwotoxicDinophysisspecies observationsintwoselectedstationsofNWIberiaandwasableto highlight not only their seasonal variability but also the interannualvariabilityoftheupwellingintensity,runoffandSST patternswithemphasisonparticularmeteorologicaland oceano-graphicanomaliesthatwererelatedwithpronouncedchangeson thelocationandtimingofbloominitiationandintensity.Future worksshouldfocusonimprovingtheknowledgeoftheoriginof theinoculumpopulationsontheshelfandiftheyarerelatedto Dinophysis populations that develop during winter in semi-enclosedcoastallagoonsalongthePortuguesecoast,namelythe LagoadeO´ bidos(Pereiraetal.,2012)andLagoadeAlbufeirain Lisbonbay(Coutinho,pers.comm.).
Further studies on the relationship of the two Dinophysis species with their prey Mesodinium rubrum are also needed. Despite theobservations that Dinophysis blooms usually occur aftertheMesodiniumblooms,itwasstillnotpossibletoestablisha direct link, partly due to the insufficient identification of the Mesodiniumspecies(andpossiblyotherciliates)inIberiawaters. Suchstudieswillhelptoclarifyiftheycanbeusedasapredictorsof DinophysisbloomsoffIberia.A betterknowledgeofthevertical positionofbothpreyandpredatorverticalanddielmigrationin the water column will also be indispensable to appropriate simulationsofthealongshoretransportofbloomsandconnectivity betweenadjacentregions.
Thechanges in theDinophysisacutaepicenterreported here couldbeidentifiedbecausetheyweremainlyobservedalongthe Portuguesecoast.SimilarstudiesfocusedonDinophysisacuminata will need a join effort to combine the results from the HAB monitoringprogramsofPortugalandGaliciainordertocoverthe wholeNWIberiancoasts.
Acknowledgments
Thanks aredue to projectsEU FP7‘‘ASIMUTH’’ (7FP SPACE-261860), FCT ‘‘HAB-SPOT’’ (PTDC/MAR/100348/2008) and to INTECMAR(Galicia,Spain)andIPMA(Portugal)HABmonitoring programs.ThankstoVitorBettencourtforhislongandcontinuous helpontheweeklysamplingatAveiro,andtothecolleagueswho collaboratedonphytoplanktoncounts.
ThisstudyhasbeenconductedusingMyOcean,nowCopernicus –Marineenvironmentmonitoring service,Productsavailable at http://www.myocean.eu/. ECMWF ERA-Interim data have been obtainedfromtheECMWFdataserver.Theauthorsacknowledge theuseoftheFerretprogram,aproductofNOAA’sPacificMarine Environmental Laboratory (information is available at http:// ferret.pmel.noaa.gov/Ferret/).
ThisisacontributiontotheSCORandIOCGEOHABprogram, CoreResearchProjectson‘‘HABsinUpwellingSystems’’and‘‘HABs andStratification’’.[SS]
Annex 1. Interannual variability of dates of the first, last occurrence and bloom initiation of Dinophysis acuminata and Dinophysis acuta, and of dates and concentration of each species maxima
Years Dinophysis acuminata Dinophysis acuta
Aveiro – AV Pontevedra – PO Aveiro – AV Pontevedra – PO 1st occurr. Date of bloom initiation >200 cells/L Date of maxima Maxima cells/L Last occur. 1st occur. Date of bloom initiation >200 cells/L Date of maxima Maxima cells/L Last occur. 1st occur. Date of bloom initiation >200 cells/L Date of maxima Maxima cells/L Last occur. 1st occur. Date of bloom initiation >200 cells/L Date of maxima Maxima cells/L Last occur.
1st sampling on 18 May 2004 1st sampling on 18 May 2004
2004 N/A N/A 31-May 1080 04-Oct N/A N/A 9-Aug 680 08-Nov 31-May 19-Jul 27-Jul 1880 – – 0 PO with several peaks of 600
cells
30-Aug 30-Aug 1960 13-Dec 30-Aug 07-Sep 7-Sep 1200 20-Dec 2005 28-Mar 28-Mar 8-Apr 880 21-Mar 21-Mar 4-Apr 2880 08-Apr 1-Aug 22-Aug 13920 04-Apr 17-Aug 17-Aug 960
19-Sep 19-Sep 880 31-Oct 13-Jun 13-Jun 2120 07-Nov 5-Sepa 19-Sep 87160
17-Oct 14160 05-Dec 17-Oct 7-Nov 17560 19-Dec 2006 03-April – 25-Jul 80 15-May 17-Jul 01-Aug 1880 17-Apr 29-May 26-Jun 1360 24-Apr – – 0
11-Sep 11-Sep 1080 26-Sep 21-Aug 21-Aug 2160 20-Nov 21-Aug 18-Sep 1200 20-Nov 05-Sep 02-Oct 3000 30-Oct 2007 19-Feb 05-Jun 18-Jun 1480 13-Mar 18-Jun 18-Jun 2240 01-Oct 08-Oct 15-Oct 1520 11-Dec – – – 0 –
23-Jul 07-Aug 9360 20-Nov 17-Sep 17-Sep 1000 17-Dec
2008 22-Jan 11-Feb 11-Feb 280 27-Feb 12-May 26-May 2960 26-Feb 07-May 14-Jul 16240 17-Mar – 17-Mar 80 23-Jun 30-Jun 14-Jul 3600 21-Oct 01-Sep 01-Sep 1160 10-Nov 28-Jul 06-Aug 2080 28-Oct
2009 06-Apr 14-May 14-May 1720 06-Jul 02-Mar 16-Mar 29-Jun 3000 30-Nov 04-May – 04-May 160 01-Jun – 01-Jun 80
– 08-Oct 40 12-Nov – 26-Oct 80 26-Oct 2010 26-Mar 01-Jun 05-Jul 1900 22-Mar 07-Jun 12-Jul 27440 01-Jun 06-Sep 06-Sep 500 21-Sep – – – 0 –
06-Sep 21-Sep 280 20-Oct 23-Aug 30-Aug 1760 13-Dec
2011 28-Mar 11-Jul 11-Jul 260 21-Feb 29-Mar 09-May 1520 01-Aug – 01-Aug 150 17-Oct 06-Jun – 06-Jun 160 04-Jul 29-Aug 460 28-Nov 08-Aug 29-Aug 5200
17-Oct 31-Oct 1520 21-Nov
2012 1st sampling on 30 May 28-Feb 05-Mar 02-Apr 14640 1st sampling on 30 May 06-Aug – 06-Aug 80 06-Aug N/A N/A 04-June 480 09-Oct 21-May 11-June 1240 09-Oct 03-Jul – 03-Jul 140 01-Oct
2013 25-Mar 02-Apr 15-Apr 340 12-Feb 25-Mar 09-Apr 5880 15-Apr 09-Sep 17-Sep 4640 16-Dec 07-Oct 07-Oct 07-Oct 5000 16-Dec 16-Jul 17-Sep 900 16-Dec 08-Jul 15-Jul 4280
7-Oct 7-Oct 3320 16-Dec N/A – not applicable since sampling was considered to late for the species.
a
Date of a minimum (>200 cells/L) before another peak.
M.T. Moita et al. / Harmful Algae 53 (2016) 17–32 31
References
Batifoulier,F.,Lazure,P.,Velo-Sua´rez,L.,Maurer,D.,Bonneton,P.,Charria,G.,Dupuy,
C.,Gentien,P.,2012.DistributionofDinophysisspeciesintheBayofBiscayand
possibletransportpathwaystoArcachonBay.J.Mar.Syst.109–110,S273–S283.
Bakun,A.,1973.CoastalUpwellingIndices,WestCoastofNorthAmerica1946–
1971.NOAATech.Rept.NMFSSSRF-671,pp.103.
Carr,S.D.,Capet,X.J.,McWilliams,J.C.,Pennington,J.T.,Chavez,F.P.,2007.The
influenceofdielverticalmigrationonzooplanktontransportandrecruitmentin anupwellingregion:estimatesfromacoupledbehavioral-physicalmodel.Fish Oceanogr.17,1–15.
Cembella,A.D.,Ibarra,D.A.,Dioge`ne,J.,Dahl,E.,2005.Harmfulalgalbloomsandtheir
assessmentinfjordsandcoastalembayments.Oceanography18,158–171.
Crawford,D.W.,Lindholm,T.,1997.Someobservationsonverticaldistributionand
migrationofthephototrophicciliateMesodiniumrubrum(=Myrionectarubra)in astratifiedbrackishinlet.Aquat.Microb.Ecol.13,267–274.
Crespo,B.G.,Figueiras,F.G.,Porras,P.,Teixeira,I.G.,2006.Downwellingand
domi-nanceofautochthonousdinoflagellatesintheNWIberianmargintheexample oftheRı´adeVigo.HarmfulAlgae5,770–781.
Delmas,D.,Herbland,A.,Maestrini,S.Y.,1992.Environmentalconditionswhichlead
toincreaseincelldensityofthetoxicdinoflagellatesDinophysisspp.in nutrient-richandnutrient-poorwatersoftheFrenchAtlanticcoast.Mar.Ecol.Progr.Ser. 89,53–61.
Dı´az,P.A.,Reguera,B.,Ruiz-Villarreal,M.,Pazos,Y.,Velo-Sua´rez,L.,Berger,H.,
Sourisseau,M.,2013.Climatevariabilityandoceanographicsettingsassociated
withinterannualvariabilityintheinitiationofDinophysisacuminatablooms. Mar.Drugs11,2964–2981.
Edwards,M.,Johns,D.,Leterme,S.,Svendsen,E.,Richardson,A.,2006.Regional
climatechangeandharmfulalgalbloomsinthenortheastAtlantic.Limnol. Oceanogr.51(2),820–829.
Escalera,L.,Reguera,B.,Pazos,Y.,Moron˜ o,A.,Cabanas,J.M.,2006.Aredifferent
speciesofDinophysisselectedbyclimatologicalconditions?Afr.J.Mar.Sci.28, 283–288.
Escalera,L.,Reguera,B.,Moita,M.T.,Pazos,Y.,Cerejo,M.,Cabanas,J.M.,
Ruiz-Villarreal,M., 2010. Bloomdynamics ofDinophysis acutain anupwelling
system:insitugrowthversustransport.HarmfulAlgae9,312–322.
Figueiras,F.G.,Labarta,U.,Reiriz,M.J.F.,2002.Coastalupwelling,primary
produc-tionandmusselgrowthintheRı´asBaixasofGalicia.Dev.Hydrobiol.167,121– 131.
Fiu´za,A.,Macedo, M.E.,Guerreiro, M.R., 1982.Climatologicalspace and time
variationofthePortuguesecoastalupwelling.Oceanol.Acta5,31–40.
Frouin,R.,Fiu´za,A.F.G.,Ambar,I.,Boyd,T.J.,1990.Observationsofapoleward
surfacecurrentoffthecoastofPortugalandSpainduringwinter.J.Geophys. Res.95,679–691.
Garcia-Cuetos,L.,Moestrup,Ø.,Hansen,P.J.,2012.Studiesonthegenus
Mesodi-niumII.Ultrastructuralandmolecularinvestigationsoffivemarinespecieshelp clarifyingthetaxonomy.J.Eukaryot.Microbiol.59,374–400.
GEOHAB,2005.In:Pitcher,G.,Moita,M.T.,Figueiras,F.,Kudela,R.,Probyn,T.,
Trainer,V.(Eds.),GlobalEcologyandOceanographyofHarmfulAlgalBlooms, GEOHABCoreResearchProject:HABsinUpwellingSystems.SCORandIOC, BaltimoreandParis,p.88.
GEOHAB,2008.In:Gentien,P.,Reguera,B.,Yamazaki,H.,Fernand,L.,Berdalet,E.,
Raine,R.(Eds.),GlobalEcologyandOceanographyofHarmfulAlgalBlooms, GEOHABCoreResearchProject:HABsinStratifiedSystems.IOCandSCOR,Paris, France,andNewark,Delaware,USA,p.59.
Hankin,S.,Harrison,D.E.,Osborne,J.,Davison,J.,O’Brien,K.,1996.Astrategyanda
toolferret,forcloselyintegratedvisualizationandanalysis.J.Vis.Comput. Anim.7,149–157.
Harred,L.B.,Campbell,L.,2014.Predictingharmfulalgalblooms:acasestudywith
DinophysisovumintheGulfofMexico.J.PlanktonRes.36,1434–1445.
Kim,S.,Kang,Y.G.,Yih,W.,Coats,D.W.,Park,M.G.,2008.Growthandgrazing
responsesofthemixotrophicdinoflagellateDinophysisacuminataasfunctions oflightintensityandpreyconcentration.Aquat.Microb.Ecol.51,301–310.
Haynes,R.,Barton,E.D.,1990.ApolewardflowalongtheAtlanticcoastoftheIberian
Peninsula.J.Geophys.Res.95,11425–11441.
Moita,M.T.,Silva,A.J.,2001.DynamicsofDinophysisacuta,D.acuminata,D.tripos
andGymnodiniumcatenatumduringanupwellingeventofftheNorthwestcoast ofPortugal.In:Hallegraeff,G.M.,Blackburn,S.I.,Bolch,C.J.,Lewis,R.J.(Eds.), harmfulalgalblooms2000.IOCofUNESCO, pp.169–172.
Moita,M.T.,Sobrinho-Gonc¸alves,L.,Oliveira,P.B.,Palma,S.,Falca˜o,M.,2006.A
bloomofDinophysisacutainathinlayeroffNWPortugal.Afr.J.Mar.Sci.28, 265–269.
Nolasco,R.,Dubert,J.,Domingues,C.P.,CordeiroPires,A.,Queiroga,H.,2013a.
Model-derivedconnectivitypatternsalongthewesternIberianPeninsula:asymmetrical larvalflowandsource-sinkcell.Mar.Ecol.Prog.Ser.485,123–142.
Nolasco,R.,Cordeiro-Pires,A.,Cordeiro,N.,LeCann,B.,Dubert,J.,2013b.A
high-resolutionmodelingstudyoftheWesternIberianMarginmeanandseasonal upperoceancirculation.OceanDyn.63(9–10),1041–1062.
Oliveira,P.B.,Nolasco,R.,Dubert,J.,Moita,M.T.,Peliz,A.,2009.Surfacetemperature,
chlorophylland advectionpatterns duringa summer upwellingevent off centralPortugal.Cont.ShelfRes.29(5–6),759–774.
Palma,A.S.,Vilarinho,M.G.,Moita,M.T.,1998.Interannualtrendsinthelongshore
variationofDinophysisoffPortuguesecoast.In:Reguera,B.,Blanco,J., Ferna´n-dez,M.L.,Wyatt,T.(Eds.),HarmfulAlgae.XuntadeGaliciaandIOCofUNESCO Publishers, pp.124–127.
Pazos,Y.,Figueiras,F.G., AlvarezSalgado,X.A., Ro´son,G.,1995.Hydrographic
situationsand speciesassociatedwith theappearanceofDinophysisacuta andtheirprobablecystsinRı´adeArousa.In:Lassus,P.,Arzul,G.,Erard-Le Denn,E., Gentien, P., Marcaillou-Le Baut, C. (Eds.),Harmful Marine Algal Blooms.LavoisierPublishing,Paris, pp.651–656.
Pazos, Y., Moron˜ o, A., Maneiro,J., Escalera,L., Reguera, B.,2005. Dinophysis
acuminataandwatercolumnstratification.In:GEOHAB,OpenScienceMeeting
onHABsinStratifiedSystems.ProgrammeandBookofAbstracts,33..http://
www.geohab.info/images/stories/documents/OSM_A4.pdf.
Peperzak,L.,Snoeijer,G.J.,Dijkema,R.,Gieskes,W.W.C.,Joordens,J.,Peeters,J.C.H.,
Schol,C.,Vrieling,E.G.,Zevenboom,W.,1996.DevelopmentofaDinophysis
acuminatabloom intheriverRhineplume (NorthSea).In:Yasumoto,T., Oshima, Y., Fukuvo, V. (Eds.), Harmful and Toxic Algal Blooms. IOC of UNESCO, pp.273–276.
Pereira,P.,Botelho,M.J.,Cabrita,M.T.,Vale,C.,Moita,M.T.,2012.Winter–summer
nutrientcompositionlinkagetoalgaeproducedtoxinsinshellfishataeutrophic coastallagoon(O´ bidoslagoonPortugal).Estuar.Coast.ShelfSci.112,61–72.
Pitcher,G.C.,Figueiras,F.G.,Hickey,B.M.,Moita,M.T.,2010.Thephysical
oceanog-raphyofupwellingsystemsandthedevelopmentofharmfulalgalblooms.Prog. Oceanogr.85,5–32.
Raine,R.,2014.Areviewofthebiophysicalinteractionsrelevanttothepromotionof
HABsinstratifiedsystems:thecasestudyofIreland.DeepSeaRes.II101, 21–31.
Relvas,P.,Barton,E.D.,Dubert,J.,Oliveira,P.B.,Peliz,A.,Silva,J.C.B.,daSantos,A.M.,
2007.PhysicaloceanographyofthewesternIberiaecosystem:latestviewsand
challenges.Prog.Oceanogr.74,149–173.
RDevelopmentCoreTeam,2011.R:ALanguageandEnvironmentforStatistical
Computing.RFoundationforStatisticalComputing,Vienna,Austria.ISBN
3-900051-07-0,URLhttp://www.R-project.org/
Reguera,B.,Bravo,I.,Marin˜ o,J.,Campos,M.J.,Fraga,S.,Carbonell,A.,1993.Trendsin
theoccurrenceofDinophysisspp.inGalicianCoastalWaters.In:Smayda,T.J., Shimizu,Y.(Eds.),ToxicPhytoplanktonBloomsintheSea.Elsevier,Amsterdam, pp.559–564.
Reguera,B.,Bravo,I.,Fraga,S.,1995.Autoecologyandsomelifehistorystagesof
DinophysisacutaEhrenberg.J.PlanktonRes.17(5),999–1115.
Reguera,B.,Velo-Sua´rez,L.,Raine,R.,Park,M.G.,2012.HarmfulDinophysisspecies:a
review.HarmfulAlgae14,87–106.
Singh,A.,Harding,K.,Reddy,H.R.V.,Godhe,A.,2014.AnassessmentofDinophysis
bloomsinthecoastalArabianSea.HarmfulAlgae34,29–35.
Smith, R.L.,1981.In:Richards,F.A. (Ed.), AComparisonofthe Structureand
Variabilityofthe FlowFieldin ThreeCoastalUpwelling Regions:Oregon, NorthwestAfrica,andPeru.CoastalUpwelling.CoastalandEstuarineSciences 1,AmericanGeophysicalUnion,Washington, pp.107–118.
Sordo,I.,Barton,E.D.,Cotos,J.M.,Pazos,Y.,2001.Aninshorepolewardcurrentinthe
NWoftheIberianpeninsuladetectedfromsatelliteImages,anditsrelationwith G.catenatumandD.acuminatabloomsintheGalicianRias.Estuar.Coast.Shelf Sci.53,787–799.
Vale,P.,Botelho,M.J.,Rodrigues,S.M.,Gomes,S.,Sampayo,M.A.,2008.Twodecades
ofmarinebiotoxinmonitoringinbivalvesfromPortugal(1986–2006):areview ofexposureassessment.HarmfulAlgae7,11–25.
Velo-Sua´rez,L.,Gonza´lez-Gil,S.,Gentien,P.,Lunven,M.,Bechemin,C.,Fernand,L.,
Raine,R.,2008.ThinlayersofPseudo-nitzschiaspp.andthefateofDinophysis
acuminataduringanupwelling–downwellingcycleinaGalicianRı´a.Limnol. Oceanogr.53,1816–1834.
Velo-Sua´rez,L.,Reguera,B.,Garces,E.,Wyatt,T.,2009.Verticaldistributionof
divisionratesincoastaldinoflagellateDinophysisspp.populations:implications formodelling.Mar.Ecol.Prog.Ser.385,87–96.
Velo-Sua´rez,L.,Reguera,B.,Gonzalez-Gil,S.,Lunven,M.,Lazure,P.,Nezan,E.,
Gentien,P.,2010.Applicationofa3DLagrangianmodeltoexplainthedeclineof
aDinophysisacuminatabloomintheBayofBiscay.J.Mar.Syst.83(3–4), 242–252.
Velo-Sua´rez,L.,Gonzalez-Gil,S.,Pazos,Y.,Reguera,B.,2014.Thegrowthseasonof
Dinophysisacuminatainanupwellingsystemembayment:aconceptualmodel basedoninsitumeasurements.Deep-SeaRes.II101,141–151.
Villarin˜ o,M.L.,Figueiras,F.G.,Jones,K.J.,A´lvarez-Salgado,X.A.,Richard,J.,Edwards,
A.,1995.Evidenceofinsitudielverticalmigrationofared-tidemicroplankton
speciesinRı´adeVigo(NWSpain).Mar.Biol.123,607–617.
Whyte,C.,Swan,S.,Davidson,K.,2014.Changingwindpatternslinkedtounusually
highDinophysisbloomsaroundtheShetlandIslands,Scotland.HarmfulAlgae 39,365–373.
Wooster,W.,Bakun,A.,Mclain,D.,1976.Theseasonalupwellingcyclealongthe
easternboundaryoftheNorthAtlantic.J.Mar.Res.34,131–141.
Xie,H.,Lazure,P.,Gentien,P.,2007.SmallscaleretentivestructuresandDinophysis.
J.Mar.Syst.64,173–188.
Yih,W.,Kim,H.S.,Myung,G.,Park,J.W.,Yoo,Y.D.,Jeong,H.J.,2013.Thered-tide
ciliateMesodinium rubrumin Korean coastal waters. Harmful Algae 30, S53–S61.
