Traceelements inbloodcollected frombirdsfeedingin theareaaround Do˜nanaNational Parkaffected bythe toxicspillfromtheAznalc´ollarmine
V.Benitoa,V.Devesaa,O.Mu˜noza,M.A.Su˜nera,R.Montoroa,*,R.
Baosb,F.Hiraldob, M.Ferrerb,M.Fern´andezc,M.J.Gonz´alezc
aInstitutodeAgroqu´ımicayTecnolog´ıadeAlimentos,IATA,(CSIC),ApdoCorreos73,46100Burjassot,Valencia,Spain
bEstaci´onBiolo´gicadeDo˜nana,Pabellondel Per´us/n(CSIC),Se illa,Spain
cDepartamentodeI.A.yQu´ımicaAmbiental,IQOG,CSIC,JuandelaCiera,3,28006-Madrid,Spain
Abstract
Along-term monitoring plan wasestablished to study ifbird populationsaround Do˜nana National Park were affected bythetoxicspillfromtheAznalc´ollarmine.TheconcentrationsofZn,Pb,As,Cu,Sb,Co,TlandCdinthe bloodof11birdspeciesfeedingintheareaweredetermined.Theparameterswhichmostaffecttheaccumulationof trace elements inthe birds studied are, firstly,species and, secondly, trophic position, sex,daysofexposure and weight.Insomeindividuals,ZnandCuoccurred athigher levelsthan thereferencevaluesforcontaminatedareas.
Concentrations ofPbandCdinaconsiderable number ofindividualswerehigherthanthosefoundinbirdsfrom uncontaminatedareas. The present data, together withthe lackofdata onblood metal concentrationprior tothe spill,donotofferanyconclusiveevidence oftheinfluence ofthespillonavianbloodmetal concentrations.
Keywords: Acidminewaste;Heavymetals; Arsenic; Do˜nanaNational Park; Blood;Birds
*Correspondingauthor. Tel.:+34-963900022;fax:+34-963636301.
E-mailaddress:oro.
8
1.Introduction
On the 25April 1998,adam retaining 5Hm3 of pyritic sludge gave way and released heavy metals totheGuardiamarRiver.Themostabun- dantmetals,considered highlytoxicforwildlife, were Zn, Pb,As,Cu,Sb,Co,Tland CdŽWood,
1974..Although great efforts havebeen made to
term monitoring plansamplinglivinganimalswas also established to identify the sublethal effects thatthiscontaminationepisodemayhaveonindi- vidualbirdsandfuture birdpopulations.
Toachievethese aims,concentrationofmetals inbloodweremeasured sincebloodreflectsthe current status of a toxic substance in the body and isincreasingly used inwildlifesampling pro-
remove the sludge,soilshavebeen contaminated
grams
ŽHenny and Meeker, 1981.. It is known
withvariable amounts ofthese elements ŽSimon et al., authors in this issue.. The contaminated surfacerepresentsarelativelysmallpercentageof
theprotectedarea ofDo˜nanawetlands,butsome of the affected locations are important feeding
andbreedinggroundsfor nestingandwintering waterfowl,thusasignificantpart ofthese popula- tions could be affected by the contamination. Land birds may not be affected to the same extentas aquaticbirds,sincetheirfeedingarea generally does not extend into the contaminated part, although species that coverlarge distances, suchassomeraptors, maybeaffected.
TheDo˜nanaNational Park isaBiosphere Re- serveŽComisi´ondeExpertos, 1992.andthewhole area is consideredto be of special interest for species ofwaterfowl ŽGarc´ıa etal.,1989;M´a˜nez,
1991; Garc´ıa-Novo, 1995..Do˜nanaisabreeding ground andwinteringarea forspeciessuchasthe
that theconcentrationofCdandPbinbloodisa goodshort-term andlong-termindicator ofcon- taminationŽGarc´ıa-Fern´andez etal.,1995,1997.. AsfarasAsandCoareconcerned, there isno evidenceifbloodconcentrationofthese elements represents ashort-oralong-termexposure,al- though inhumans arsenic levelsinblood havea short biological half-life and reflect recent expo- sure ŽCornelisetal.,1995;NicolasandDescotes,
1996..Moreover, there is alsoagreatdifference betweenorganismsintheretention ofarsenicin erythrocytes ŽEisler, 1994..Large amounts ofZn are initiallystored inthe human liver,butitalso tendstoaccumulateinredbloodcellsandbones, whilstcopper isinitiallybound toserum albumin and later more firmly bound to alpha-cerulop- lasmin,theliverandbonemarrowbeingthemain copper storage organs. Trivalent forms of anti- mony generally concentrate in red blood cells,
Spanish imperial eagle
ŽAquila adalberti., con-
whilepentavalentcompoundsarefoundinplasma.
sidered inthe EuropeanUnion tobeone ofthe
ŽHammondandBeliles,1980..
birds in greatest danger of extinction
ŽFerrer,
The toxic effects of the various heavy metals
1993..Inaddition,somesedentary birds,andbirds
that winter inDo˜nana are hunted inneighbour- ingareas, and this mayprovide alink bywhich
contaminationcouldreach human populations.
andarseniconbirdshavebeendocumented.Lead cancause birdmortality ŽRamoetal.,1992;Ma- teoetal.,1998.,havesublethal effectsŽOchiaiet al., 1992., or a negative effect on reproduction
The Scientific Research Council
ŽCSIC.
was
ŽBurger,1995.,depending onthedose.Highdoses
instructedbytheSpanishGovernmenttomonitor theeffectthatcontaminationbyheavymetalsand arsenic mayhaveonwildlifeintheecosystemsof Do˜nana. For the reasons stated above, and be- cause aquatic birds are consideredgood indica- torsofcontaminationinwetlandssincetheyaccu- mulate metals and As in various organs, they were included among the various groups ofani-
ofCdcancausedeath, andsublethal effectshave beendescribedin birdsatlowerconcentrations ŽBokori etal.,1995..Arsenic,especiallyinits inorganicforms,canbringaboutthedeathof an individual,producesublethaleffectsand affect reproductionŽEisler, 1994..There are noknown studiesontheeffectsofSb andTlonbirds,but sublethal effects have been described in both
malsstudied. Intheshort term, theaimofmoni-
adult rats and fetuses
ŽAlkhawajah et al., 1996;
toring wasto determinewhether contamination by metalsaffectedsummer avianmortality ŽHern´andez etal.,authors inthisissue.. Along-
Barroso Moguel etal.,1996;Albiser etal.,1997; Poonetal., 1998..Zn,CuandCoareessential elements forbirds, but theycanbetoxicinhigh
concentrations,andvarioussublethal effectshave
whichincludes species, number ofindividuals of
been described
ŽRamey and Sterner, 1995;
each speciessampled, position inthe foodchain,
Mart´ınezandD´ıaz,1996;Ewingetal.,1998.. Moreover, when studying the effects of an
episodeoflocalcontaminationonvariousspecies, onemustbearin mindthattherearenumerous sourcesofindividualvariation. Amongthosemost
weight,daysofexposuretothespill,sex, age, samplingpoint andsocialandecologicalinterest. Theweightsassignedtoeachoneof thestudied speciesandintroducedintheanalysisascontinu- ousvariable weretaken directlyfromthebibliog-
frequently quoted are: physiological characteris-
raphy
ŽCramp and Simmons, 1977, 1980, 1983,
ticsrelated toratesofabsorptionandassimilation ŽPeterle,1991.;trophic position andweightŽPain and Amiard-Triquet,1993;Garc´ıa-Fern´andez et
1985.,selecting the weightofindividuals belong- ingtothe analysed species,sampled inthe same
seasonŽspring—summer.inplacesnexttoDo˜nana.
al., 1997.; sex
ŽFinley et al., 1976., and age
Days of exposure were calculated as the differ-
ŽGarc´ıa-Fern´andez et al.,1996,1997..Moreover, the distance of breeding grounds from the spill
and the length of time that birds have been exposed to its effects may affect the degree of aviancontamination.
The aim of this paper is to present the first data on metal concentrations in the blood of birds found inDo˜nana during the breeding sea- son,andtodeterminewhether these levelscorre- spondtoareas contaminatedbyheavymetalsand arsenic ortouncontaminatedareas. Finally,Gen- eralised Linear ModelsŽNelderandWedderburn,
1972;McCullagh andNelder, 1983.wereusedto determinetheimportancethatthevarioussources ofvariationmentioned abovemayhaveinex- plainingthecontaminationobserved inindividual specimens.
2.Materialsandmethods
2.1.Samplingprocedure
The birds initially included in the sampling mainlyfed inthe wetlands, since itwasunlikely thatother birdswouldbeaffected by thiscon- tamination episode. Sufficient species were se- lected to represent the various trophic levels foundinthewetland birdcommunity ofDo˜nana. Breeding hadalreadystarted, andthusthecollec- tion ofsamples to some extent wasconditioned. Formanyofthespeciesselected significantnum- bersofsampleswerenotobtained, andresultsfor
encebetween thedates ofcapture andtoxicspill. Thelocations ofthe sampling stations are shown inFig.1.
Amolecular techniquebased onthe structural difference ofsexchromosomes wasusedtode- termine the sexofthe specimens captured.Un- likemammals,malebirdshavetwo identicalsex chromosomesŽZZ.,whereas thefemales are het- erogameticŽZW.ŽLesselsandMateman,1996.; consequently, theDNAsequences oftheWchro- mosome belong exclusivelyto females. Griffiths etal.Ž1996.described agene ŽCDH-W.whichis foundintheWchromosomeinfemales, withthe exceptionof ostrichesandrelated species;this knowledge wasapplied and the specimens were sexedby meansofPolymerase ChainReaction ŽPCR..
Whendrawingup thesamplingprogram,we attemptedtosample aparticularspeciesonvari- ous datesandatvariousdistancesfromthearea directlyaffected by thespill.Thelatter wasim- practical,sincemanyofthespeciesliveincolonies andonlyasinglecolonywasaccessibleduringthe sample collection period, thus itwasnot possible toconsiderthisvariablewhensourcesofvariation werestudied.
Thegreater flamingoŽPhoenicopterusruber.was
included inthe sampling although itsclosest breeding colony is135km from the site of the spill because monitoring ofmarked birds re- vealed that they frequently visitDo˜nana to feed duringthebreeding periodandthewinterseason. The specimens werecapturedinthe nest Žpul-
these species are thereforenot included in this
lets.
or in their feeding grounds
Žyoung birds;
paper. ThespeciesforwhichtheconcentrationofmetalsinbloodwasanalysedaregiveninTable1,
adults., usingvarioustrapping methods. Foreach specimen,2mlofbloodwastakenfromtheradial
Fig.1.Mapoftheareastudiedandthesamplingstations.
1.Dehesa deAbajo / 6.LaAlgaida. / 11.ElPinar deSanAgust´ın.2.Matasgordas / 7.Ca˜nodeGuadiamar. / 12.ElPuntal.
3.Ca˜nodelaEscupidera. / 8.Maril´opez. / 13.Odiel, Ayamonte. Huelva.
4.Veta Hornitos.
5.LaFAO. / 9.Veta LaPalma
10.LasPajareras. / 14.Fuente delaPiedra. M´alaga.
Table 1
CharacteristicsofwaterfowlspeciesinDo˜nanaNational Park studied inthisworka
SpeciesIndividualsTrophiclevelWeightb
Days ofSexAgeSamplingpointSE
sampled
Žg.
exposurec
ŽindividualsIntereste
sampled at
eachlocation.d
PodicepscristatusŽgreat7Fishpredator9201067N.K.7Žpull.
crested grebe
3Ž7.
Protected
ArdeacinereaŽgreyheron.
20Fishandcrayfish150543—113f
predator
3M;14F;3N.K.17Žpull.; 3ŽN.K..
4Ž4.;10Ž14.;12Ž2.
Protected
CiconiaciconiaŽwhitestork.
30Fishandcrayfish35711616M;14F30Žpull.
predator
1Ž30.
Protected
Platalea leucorodiaŽspoonbill. / 30 / Fishandcrayfish / 1960 / 3930—87f / 15M;12F;3N.K. / 28Žpull.;2Žadult. / 4Ž1.;10Ž15.;13Ž14. / Protected
predator
PlegadisfalcinellusŽglossy / 10 / Invertebrate / 557 / 25 / 4M;6F / 10Žpull. / 5Ž10. / Protected
ibis. / predator
PhoenicopterusruberŽgreater / 20 / Invertebrate / 3579 / 111 / 14M;6F / 20Žpull. / 20Ž14. / Protected
flamingo.
AnasplatyrhynchosŽmallard. / 8 / predator
Omnivorous / 1216 / 65—107f / 2M;1F;6N.K. / 3Žpull.; 5Žadult.; / 4Ž2.; 7Ž1.;8Ž2.; / Consumption
AnasstreperaŽgadwall. / 5 / Omnivorous / 659 / 74—86f / 3M;2F / 1ŽN.K..
4Žpull.; 1ŽN.K.. / 9Ž4.
4Ž1.;8Ž4. / Consumption
AythyaferinaŽpochard. / 10 / Omnivorous / 849 / 82—88f / 10N.K. / 10Žpull. / 4Ž1.;8Ž9. / Consumption
MiliusmigransŽblackkite. / 25 / Birdandmammal / 807 / 45—78f / 13M;7F;4N.K. / 23Žpull.; 2Žadult. / 1Ž2.;2Ž9.; 6Ž3.;11Ž2.; / Protected
predator / N.K.Ž9.
LaruscachinnansŽyellow- / 11 / Carrion consumer / 971.5 / 46—108f / 2M;9N.K. / 10Žpull.;1ŽN.K.. / 9Ž7.;N.K.Ž3. / Protected
leggedgull.
aM,male; F,female;N.K., not known;Pull, pullet.
bThe weightsquoted are typicalweightsreferredbyCramp and Simmons Ž1977,1980,1983,1985..
cDays ofexposurewere calculatedasthe difference betweencapturedateand toxic spill date.
dSamplingpoint:see Fig. 1tolocatethe samplingpointon the map.
eSEInterest:Social and EcologicalInterest.
fDays exposureintervalinwhich the captureswere carriedout.
vein. Thespecimenscollectedweremarkedwith plastic rings that could be identified individually atadistance toallowmonitoring ofindividualsin thefuture.
2.2.Treatmentofsamples
Blood was kept frozen from collection to preparation.BloodŽ1ml.wastaken andadded to
9mlof0.5%Žw/v.Triton X-100ŽMerckFarma y
Qu´ımica,S.A.,Valencia, Spain..Thesolution was mechanically shaken for 30 s and then shaken
vigorouslyinanultrasoundbath for30min.The mixture wascentrifuged at2000rev.min—1 for5 min.Aliquots of1mlweretaken andtheprecipi- tate wasdiscarded. The aliquots obtained were
glassware,Eppendorf vials,anddisposable sam- plingcupsweretreatedwith10%nitricacidfor1 week, and then rinsed three times with Milli-Q water, before use. Between uses, glassware was placedin10%nitricacidfor24h.Method detec- tionlimitswere:AsŽ0.006mgl—1.,CdŽ0.0001mg l—1.,CoŽ0.001 mgl—1.,CuŽ0.013 mgl—1.,PbŽ
0.002mgl—1.,SbŽ0.002mgl—1.andTlŽ0.002mg l—1..Theprecision asmeasuredby triplicate anal- ysis,expressed asthe relative standarddeviation ŽR.S.D..wasasfollows:As4%,Cd5%,Co5%, Cu2%,Pb4%,Sb 3%andTl3%.Therangeof recoveryevaluated byspikingbloodsamples with eachoftheelements was85—115%.
Accuracy for Pb and Cd was established by analysing a certified referencesample ofbovine
stored at —20°C prior to elementalanalysis. An
blood ŽCRM-195.
from the Institute for Refer-
MSE Minor centrifuge
ŽPacisa, S.A., Madrid,
ence Materials andMeasurementsŽIRMM..The
Spain., aVortexMS2Minishaker
ŽIKA
values found
ŽPb: 413±17; Cd: 5.40±0.69 pg
Labortechnik,MerckFarma y Qu´ımica,S.A., Barcelona, Spain.andaSelecta UltrasonsPbath ŽJ.P.Selecta, S.A.,Barcelona, Spain.wereused.
l—1.overlappedwith the interval found for the certified values ŽPb: 416±9;Cd: 5.37±0.24pg l—1.. For As, Cu and Co, a CRM sample or TORT-2 Žlobster hepatopancreas,Canada Natio-
2.3.DeterminationofAs,Cd,Co,Cu,Pb,SbandTl
nal Research Council, CNRC.
was used. The
The determinationofAs,Cd, Co, Cu, Pb, Sb andTlinwholebloodsamples wasaccomplished by graphite furnace Zeeman-effectatomicab- sorption spectroscopy. APerkin-ElmerŽPE.lon- gitudinal ACZeemanŽAAnalyst 600.atomic ab- sorption spectrometer, equipped withatrans- versely heated graphite atomiser and a built-in fully computer-controlled AS-800 autosampler
values obtainedfor these elements inour CRM
analyseswereconsistent withthecertified values.
2.4.Determinationofzinc
InthedeterminationofZnbyflow injection- flame-atomic absorption spectrometry,a PE Model3300atomicabsorption spectrometer equipped withaPEflowinjection analysissystem
ŽPerkin Elmer Hispania, S.A., Madrid, Spain.,
for atomic spectroscopy
ŽFIAS-400.
and an au-
andPEpyroliticgraphitecoatedtubeswith an inserted L’vovplatform were used. Calibration was performed with the method of Additions
tosampler ŽPEAS-90.wereused.Precisionwas established byanalysisofanA-13referencesam- pleofanimalbloodobtainedfromtheInternatio-
StandardCurve. Deionised water
Ž18 M cm.
nalAtomicEnergyAgencyŽIAEA..Therecorded
obtainedwithaMilli-Q water system ŽMillipore
values
Ž13.6±0.6 pg g—1.
overlappedwith the
Inc., Millipore Ib´erica, Madrid, Spain. wasused forthepreparationofreagents andstandards.All
chemicals including standardsandsolutions were ofpro analysi quality or higher: nitric acid Žp=
certifiedinterval forthiselement Ž13±1pgg—1.. Theevaluationof Znrecoveryinspikedblood sampleswas95%.Themethod detection limitfor thiselement was0.23mgl—1.
1.38gml—1.,1000mgl—1
standardsofAs,Cd,
Co,Cu,Pb,Sb,Tl,palladium powder,magnesium nitratehexahydrateand ammoniumdihydrogen phosphateas matrix modifiers ŽMerck Farma y
2.5.Statisticalanalysis
AGeneralisedLinear Model ŽGLM., ŽNelder
Qu´ımica, S.A., Valencia, Spain.
were used. All
andWedderburn, 1972; Dobson,1983; McCullagh
andNelder,1983.wasusedtoderiveamathemat- icaldescriptionofindividualvariations inconcen- trations ofmetalsandarsenicinblood.Modelsof thiskindare used whendeterminingthe individ- ualeffectofseveralvariablesonagiven pheno- menon ŽDon´azar et al.,1993;Bustamante,1997;
Forero et al., 1999; Tella et al., 1999.. Gener-
alisedLinear Modelscanbeconsidered asa particularcaseofmultiplelinearregression.Three componentsmustbedefinedforaGLM:alinear predictor,anerror function andalinkfunction. A linear predictorŽLP.isdefined asLP=a+b*x
+c*x+... where a is a constant to be esti- mated; b,c, ... are parameterstobeestimated
different models which mostly converged into a singlemodel orasetofmodelsfromwhichsimi- larrelationshipscouldbeinferred.
Theexpression ofthemodelbecomes:
[metal]=ea+b x+c x+... Ž1.
3.Results
Table 2showsthetotal bloodmetal content of thespeciessampled,andTable3gives theper- centage ofindividualsofeachspecieswithhigher
from observed data; and x1, x2
... are the ex-
levelsofmetals than those reportedinthelitera-
planatory variables. The error function depends
ture for birds in uncontaminatedareas
ŽPb, As
on the nature ofthe data. The concentration of
and Cd.
or contaminated areas
ŽCu and Zn..
the various metals in blood wasln-transformed and anormal error distributionwasassumed for the models. An identity link was used as link function. In this case the model does not differ fromamultiplelinearregressionwith aln-trans- formed-dependentvariable.Theexplanatory vari- ablesconsideredwereintroducedinto the model asfactors Žspecies, sex,trophic position. orcont- inuousvariablesŽdaysofexposure tothespilland weight.. Although age isrelevant as a factor of
Data forbirdsinuncontaminatedareas werenot found for the latter elements. Reference values were not found for Co, Sband Tllevelsin the bloodof birds.Foreachelement, inthecaseof measurementsbelowthelimitofdetection ŽLOD., the valueintroducedinthe databasewasthat of LOD.
3.1.Zinc
individual variation
ŽGarc´ıa-Fern´andez et al.,
The zinccontents found inthe species studied
1996,1997.,itwasnot included inthe statistical
analysis since most of the individuals sampled werepulletsŽTable1..Wefittedeachexplanatory variable tothe observations usingthe GLM pro- gramŽBaker, 1987.followingamodificationofa traditional forwardstepwiseprocedure.Each variable wastested in turn for significance, and
ŽTable2.variedbetween 0.3and8.6mgl—1.The
species that presentedthe highest mean levelof this metal wasthe gadwall,followed bythe yel- low-leggedgull.The mean value found inglossy ibiswasnotably low.Twentypercent ofthe gad- wallindividualsanalysed ŽTable3.presentedval- ues even higher than those established for birds
only those significant at the 5% level were in-
in contaminatedareas Ž7.5 mgkg—1
wet weight
cluded in the model. Recent papers have criti- cised automaticstepwise proceduresasthey are not necessarily abletoselectthe mostinfluential
Žww., Falandyszetal.,1988.,thisbeingtheonly speciesinwhichanindividualexceeded therefer- encevalues.
variable from a subset
ŽJames and McCulloch,
1990..Themodification ofthestepwisemodelling procedureinvolvedtesting thealternative models that wereobtainedwhenthe second orthe third most significant variable wasincluded Žprovided that itwassignificantatthe5%level.,instead of the most significant one at each step. This For- ward Stepwise Branching Modelling Procedure
3.2.Lead
The levels of Pb ranged between 0.002 and
0.454 mg l—1 ŽTable 2.. The species with the highest mean levels of lead in blood was the mallard, followedbythegadwall.Someindividual specimens ofmallard had levelsclose to 0.5mg
ŽDon´azar et al., 1993.
eventually gave a set of
kg—1,avalueindicativeofleadpoisoninginswans
Table 2
Metal andarsenic concentrationsinbloodofwildbirdsfeedinginthearea around Do˜nanaNational Park a
SpeciesNo.TotalZnTotalPbTotalAsTotalCuTotalSbTotalCoTotalTlTotalCd
specimens
Žmg l—1.Žmgl—1.Žmgl—1.Žmgl—1.Žmgl—1.Žmgl—1.Žmgl—1.Žmgl—1.
CiconiaciconiaN=301.9b
0.071b
0.019b
0.586b
0.002b
0.007b
0.002b
0.0015b
Žwhitestork.
0.8—2.8c
0.002—0.320c
0.006—0.121c
0.180—1.530c
0.002c
0.001—0.016c
0.002c
0.0001—0.0090c
PlegadisfalcinellusN=100.9b
0.061b
0.008b
0.133b
0.002b
0.029b
0.002b
0.0110b
Žglossyibis.
0.7—1.3c
0.020—0.233c
0.006—0.017c
0.067—0.241c
0.002c
0.019—0.041c
0.002c
0.0060—0.0150c
MiliusmigransN=253.3b
0.054b
0.008b
0.211b
0.002b
0.004b
0.002b
0.0068b
Žblackkite.
2.3—4.5c
0.002—0.179c
0.006—0.035c
0.120—0.303c
0.002c
0.001—0.013c
0.002c
0.0010—0.0140c
AythyaferinaN=103.7b
0.073b
0.006b
0.203b
0.002b
0.011b
0.002b
0.0009b
Žpochard.
2.5—6.0c
0.025—0.274c
0.006c
0.151—0.396c
0.002c
0.005—0.016c
0.002c
0.0001—0.0030c
AnasplatyrhynchosN=93.3b0.208b0.011b0.258b0.002b0.006b0.002b0.0048b
Žmallard.
1.3—4.0c
0.045—0.454c
0.006—0.042c
0.142—0.361c
0.002c
0.001—0.015c
0.002c
0.0001—0.0190c
ArdeacinereaN=202.2b
0.015b
0.006b
0.352b
0.002b
0.019b
0.002b
0.0002b
Žgreyheron.
1.5—3.3c
0.002—0.089c
0.006c
0.204—0.650c
0.002c
0.008—0.025c
0.002c
0.0001—0.0010c
Phoenicopterusruber N=201.7b
0.076b
0.006b
0.334b
0.002b
0.054b
0.002b
0.0006b
Žgreaterflamingo.
0.3—2.6c
0.035—0.120c
0.006c
0.187—0.531c
0.002c
0.039—0.070c
0.002c
0.0001—0.0010c
AnasstreperaN=55.9b
0.120b
0.017b
0.526b
0.002b
0.027b
0.002b
0.0006b
Žgadwall.
3.5—8.6c
0.069—0.174c
0.006—0.029c
0.346—0.753c
0.002c
0.005—0.047c
0.002c
0.0001—0.0290c
PlatalealeucorodiaN=303.2b
0.008b
0.019b
0.307b
0.002b
0.048b
0.002b
0.0003b
Žspoonbill.
1.4—5.5c
0.002—0.034c
0.006—0.181c
0.190—0.569c
0.002c
0.001—0.110c
0.002c
0.0001—0.0010c
PodicepscristatusN=72.2b0.002b0.006b0.395b0.002b0.001b0.002b0.0001b
Žgreatcrested grebe.
1.3—2.9c
0.002c
0.006c
0.327—0.488c
0.002c
0.001—0.002c
0.002c
0.0001c
LaruscachinnansN=104.4b
0.020b
0.006b
0.429b
0.002b
0.040b
0.002b
0.0002b
Žyellow-leggedgull.
3.1—5.2c
0.009—0.032c
0.006c
0.271—0.535c
0.002c
0.015—0.070c
0.002c
0.0001—0.0005c
aInthosecasesinwhichthevaluewasnotdetectable,thedetection limitwastaken ŽZn0.23mgl—1;Pb0.002mgl—1;As0.006mgl—1;Cu0.013mgl—1;Sb0.002 mgl—1;Tl0.002mgl—1;Co0.001mgl—1;Cd0.0001mgl—1..
bMean ofnvalues.
cRange ofnvalues.
valuesfoundintheliteratureforcontaminatedanduncontaminatedareas
SpeciesNo.ofZnPbAsCuCd individuals
CiconiaciconiaN=300a
Žwhitestork.
PlegadisfalcinellusN=100a
Žglossyibis.
MiliusmigransN=250a
Žblackkite.
AythyaferinaN=100a
Žpochard.
AnasplatyrhynchosN=90a
Žmallard.
ArdeacinereaN=200a
Žgreyheron.
PhoenicopterusruberN=200a
Žgreaterflamingo.
AnasstreperaN=520a
Žgadwall.
PlatalealeucorodiaN=300a
Žspoonbill.
PodicepscristatusN=70a
Žgreatcrested grebe.
LaruscachinnansN=100a
Žyellow-leggedgull.
33a
20a
32a
30a
78a
5a
70a
100a
0a
0a
0a
17a3a
0a0a
8a0a
0a0a
11a0a
0a0a
0a0a
40a0a
20a0a
0a0a
0a0a
33a
100a
92a
20a
78a
0a
0a
0a
0a
0a
0a
Referencevalue
Ž7.5mgkg—1 ph.b
Ž0.062mgl—1.c
Ž0.020mgl—1.d
Ž1.15mgkg—1 ph.e
Ž0.001mgl—1.f
a%Žno.ofspecimens withvaluehigher than referencevaluefoundintheliterature/no.ofspecimens analysed.
bReferencevalueofZnforbirdsincontaminatedareas ŽFalandyszetal.,1988..
cReferencevalueofPbforbirdsinuncontaminatedareas ŽDieteretal.,1976..
dReferencevalueofAsforbirdsinuncontaminatedareas ŽBurgerandGochfeld, 1997..
eReferencevalueofCuforbirdsincontaminatedareas ŽVanEeden andSchoonbee, 1996..
fReferencevalueofCdforbirdsinuncontaminatedareas ŽGarc´ıa-Fern´andezetal.,1995..
ŽBlus etal.,1991..Thevaluesfoundinducksin uncontaminated areas by other authors are around 0.062 mgl—1 ŽDieteretal.,1976..No individualsofspoonbill,yellow-leggedgull orgreat crested grebe exceeded the referencevalue ŽTa- ble3..
3.3.Arsenic
Levels ofAsranged between 0.006and 0.181 mg l—1 ŽTable 2.. The highest mean contents were found in the white stork, spoonbill and gadwall,inallcasestheywerelessthan therefer-
Eeden and Schoonbee, 1996..Only afewexam- ples of white stork had levels above or around thisvalueŽTable3..
3.5.Antimonyandthallium
Noneof thespeciesanalysedpresentedvalues whichexceeded thedetection limitsŽ0.002mgl—1 forboth metals..
3.6.Cobalt
The Corange varied between 0.001and 0.110
ence value for uncontaminatedareas
Ž0.02 mg
mg l—1 ŽTable 2., with particularly high levels
l—1, Burger and Gochfeld, 1997.. Nevertheless,
someindividualsoffive speciesstudied presented levels that exceeded the referencevalue. There was greatvariabilityinarseniclevelsbetween specimens ofthe samespecies.Thehighnumber of species with levels below the detection limit wasalsonoteworthy Žpochard,greyheron, greater flamingo, great crested grebe and yellow-legged gull..
3.4.Copper
found inthe greater flamingo,spoonbill and yel- low-leggedgull.Thespecieswiththelowestmean levelwasthegreat crested grebe.
3.7.Cadmium
Cd concentrations were between 0.0001 and
0.029mgl—1 ŽTable 2..The species that had the highest mean concentrationsweretheglossyibis, blackkiteandmallard,indescendingorder.When theindividualvaluesforeachofthespecieswere comparedwiththe referencevaluesfound inthe
LevelsofCucontent found wasbetween 0.067
literature for uncontaminated areas
Ž0.001 mg
and1.530mgl—1 ŽTable2..Thespecieswiththe highestcopperbloodlevels werethewhitestork, gadwalland yellow-leggedgull.These mean val- ues are lowerthan those found inthe literature
l—1;Garc´ıa-Fern´andez et al.,1995., onlyfiveof theelevenspeciesincluded inthestudyhadindi-
vidualsthatexceeded thereferencevalue.Itmust be emphasised, however, that in three of these
for contaminatedareas Ž1.15 mgkg—1
ww;Van
five species the percentage of individuals with
Table 4
Variables Žfactorsandcontinuous.that are significantat5%intheGeneralisedLinear Models
Element / Model1a / Model2bAccounted devianceŽ%. / Factors / Continuous variable / Accounted devianceŽ%. / Factors / Continuous variable
Zn / 71.98 / Species / — / 52.77 / Trophic / —
Pb / 57.48 / Species / — / 24.93 / Sex / —
Trophic
Cd / 79.53 / Species / — / 69.19 / Trophic / Days
Weight
Cu / 49.08 / Species / — / 40.98 / Trophic / Weight
Co / 74.93 / Species / — / 50.74 / Trophic / —
As / 15.90 / Species / — / 9.27 / Trophic / Days
aModel 1:allthevariables studied havebeen included Žspecies;trophic level,sex,daysofexposure andweight..
bModel 2:thevariable ‘species’hasbeen omitted inallcases.
Allthevariables studied havebeen included Žspecies,trophic level,sex,daysofexposure andweight.
gadwall
greater flamingo
Plegadisfalcinellus—0.73720.09550.34040.41152.21800.1922—1.47800.15221.66000.24710.01730.3002 glossyibis
Residual deviance / 308.73 / 115.87 / 1418.90 / 317.83 / 126.29 / 151.81d.f. / 157 / 165 / 158 / 161 / 165 / 165
S.E.:StandardError. d.f.:degrees offreedom
levelsabove the referencevalues wasveryhigh, andinthecaseoftheglossyibisitwas100%.
4.Discussion
Twomodel groups werefittedforAsandeach ofthemetalsexceptSb andTl,whichwerenot- detectedinmostofthebirdsstudied. InModel1 all the variables were included Žspecies, trophic level,weight,timeelapsedsincethespillandsex., whileinModel2thespeciesvariablewasomitted inallcases.Varioussignificantmodelswerefound inbothgroups.Ineachcaseonlythebestmodels wereselected, i.e.thosewiththesmallestresidual deviance ŽTable4..
Whenweconsidered Model1,modelswere obtainedthat explained ahighpercentageofthe original deviance for all the metals except As ŽTable 5..Moreover, for allthe metals the best model included only one variable, species, and noneof theotherfactorsorcontinuousvariables considered was significant at the level chosen Ž5%.. Thevariousmodelspredictadifferent be- haviour for each species according to the metal beingconsidered.Forexample,theglossyibiswas thespeciesleastcontaminatedbyCuorZn;simi- larly, the white stork was the species with the highestconcentrationofCuinblood,whereasZn was theninthintheorderofZnconcentration comingafterthegadwall,yellow-legged gull, pochard,blackkite,mallard,spoonbill,greyheron, andgreat crested grebe.
In the models that did not include species ŽModel 2., the factors and continuous variables that are significant at the 5% level varied from
andtwocontinuousvariables,daysandweight,as significant,withmetalconcentrationincreasing as weightandpossibledays ofexposuretothespill decrease. ForCu,Model2 includesonefactor, trophic position,andonecontinuousvariable, weight;withineachtrophiclevel, theCuconcen- tration inbloodtendstoincrease withbirdweight. ForCo,onlytrophic positionissignificant;Model
2predicts the greatest concentrationin inverte- brate predatorsandthesmallest concentrationin fishpredators. ForAs,Model2includestwo significant variables, trophic position as a factor anddaysasacontinuousvariable,withthemetal concentration increasingasthepossibledays of exposuretothespill decrease.TheGeneralised Linear Models show that when the species to whichanindividualbelongsis consideredsepa- ratelyfromits trophicposition,inall themetals studied species is the variable that most con- tributes toexplainingthevariabilityofmetal con- centration inblood.Thismaybelinkedwiththe physiologicalcharacteristicsof speciesrelated to their ability to absorb and excrete metals. In somespeciesofmammals thelevelofmetal con- tamination hasbeenrelated moretoinabilityto excretemetalsthantotheirtrophic position.Nev- ertheless, wecannot discard the possibility that theimportanceofspeciesislinkedwithecological factors such as longevity, ability to move else- where,differences inmicrohabitat, andfeeding habits, which were not consideredin this study andyetare knowntoaffectthelevelofcontami- nation ofanindividual.
5.Conclusions
one metal to another
ŽTables 4 and 6.. In all
cases the models explained a good part of the deviance, except inthe caseofAs.Nevertheless,
Amongthe variousmetals studied, someŽsuch as Zn and Cu. tended to be present in all the
for all the metals the deviance explained was
individuals, whereas others
Žsuch as Sb and Tl.
alwayslessinModel2than Model1.ForZn,the
were not detectedin any of the specimens ex-
Model2includedonlytrophic position,predicting
amined. In the case of Zn and Cu
Žthe only
the highest values inthe carrion consumers and thelowestvaluesintheinvertebratepredators.In thecaseofPb,Model2included trophic position andsexassignificant;thismodelpredicts alower concentrationofPbinfemales fromeachtrophic level.For Cd,Model 2includes trophic position
metalsforwhichbloodvaluesproceeding from studies of contaminatedareas were found. only two speciesshowedindividualswithhigherlevels than thereferencevalues.Thereferencelimitfor Zn wasexceeded by20% of the gadwall speci- mens,while3%ofthewhitestorkspecimens had