the role of nutrients in ecological classification of ...€¦ · general, and nutrients in...
TRANSCRIPT
1
Theroleofnutrientsinecologicalclassificationoffreshwaters:reportofECOSTATworkshop,Tallinn,29-30May2018
Introduction
Intheory,theroleofnutrientsintheclassificationoffreshwatersinvolvesastraightforwardapplicationofthetextinAnnexVoftheWFDwhichdescribeshowsupportingelementsingeneral,andnutrientsinparticular,shoulddeterminefinalstatusclassifications.Thenormativedefinitionsofecologicalstatusclassification(WFDAnnexV,Tables1.2.1–1.2.5)usesidenticalwordstodescribenutrientconcentrationsthatareconsistentwithhigh,goodandmoderatestatusirrespectiveofwaterbodytype.Theseare:
Highstatus:“Nutrientconcentrationsremainwithintherangenormallyassociatedwithundisturbedconditions”
Goodstatus:“Nutrientconcentrationsdonotexceedthelevelsestablishedsoastoensurethefunctioningoftheecosystemandtheachievementofthevaluesspecified[inthenormativedefinitions]forthebiologicalqualityelements.”
Moderatestatus:“Conditionsconsistentwiththeachievementofthevaluesspecified[inthenormativedefinitions]forthebiologicalqualityelements.”
The“oneout,allout”ruleappliestosupportingelementsaswellastobiologicalqualityelements,soanutrientconcentrationthatcorrespondedwiththenormativedefinitionformoderatestatusshouldleadtotheoveralldesignationforthewaterbodyinquestionbeingmoderatestatus,eveniftheBQEswereallathighorgoodstatus.
Inpractice,however,itisclearthatapproachestotheuseofsupportingelementsingeneral,andnutrientsinparticular,variesconsiderablyaroundtheEuropeanUnion.Forthisreason,akeyactivityofECOSTATinrecentyearshasbeentounderstandhowMemberStatesuseallsupportingelementsfortheassessmentofecologicalstatus,butwithaparticularfocusontheroleofnutrients.ThisstartedoutbyunderstandinghowMemberStatesestablishedthenutrientconcentrationsrequiredtosupportecologicalstatusandthen,havingnotedconsiderablevariationinapproaches,workingtowardsanunderstandingof“bestpractice”.This,inturn,requiredrecognitionofanumberofchallenges(notleastofwhichareweakrelationshipsbetweenbiologicalqualityelementsandinorganicnutrientsforavarietyofreasons,andtheroleofotherstressors)andhasresultedinaCISguidancedocumentduetobepublishedlaterin2018.
2
Therewas,however,recognitionearlyinthisprocessthattheprocessofsettingstandardsfornutrientconcentrationswastightlylinkedtoacountry’sregulatorymechanisms.Thisisparticularlytrueforinorganicnutrients,whereweakrelationshipsbetweenthepressurevariableandthebiologicalresponsecreateuncertaintybothfordecidingwhetherthereisagenuineimpactasaresultofelevatednutrients,andthatthereisahighlikelihoodofanimprovementifstepsaretakentoreducenutrientconcentrations.
Inbrief,ifaMemberState’spolicyistotakeactionassoonasaregulatorystandardisbreached,thenitmakessensetosetarelativelystrictnutrientboundaryinordertominimizethelikelihoodofa“falsepositive”.Ontheotherhand,ifaMemberStaterequiresbothbiologyandchemistrytofallbelowgoodstatusbeforetakingaction,thenastrictnutrientstandardwillincreasetheriskofa“falsenegative”and,instead,amorerelaxednutrientstandardmaybeappropriate(Theseissuesarediscussedinmoredetailinsection2.2oftheCISguidance).
Thisreport(basedondiscussionsheldattheECOSTATworkshopinTallinn)explorestheinteractionsbetweenscienceandregulationthatleadtofinalstatusclassifications.Thisinvolvesrecognizingboththatnutrientsareamajorreasonforfreshwatersnotachievinggoodstatus(and,therefore,compromisethelong-termsustainablewaterresource)andthatnutrientandbiologicaldataarehighlyvariable,whichmeansthatarrivingattherightdecisionforanyindividualwaterbodymaynotbestraightforward.
Theroleof“expertjudgment”
Thephrase“expertjudgment”occurredseveraltimesduringdiscussionsattheworkshop,andaninitialconsiderationofwhatismeantbythisterm,andhowitappliestotheclassificationofwaterbodiesisnecessarybeforeproceeding.
Therearefourreferencestotheuseof“expertjudgment”inthetextoftheWFD(AnnexII,1.3,AnnexV,1.3.4(twice)andAnnexV,1.4.1)althoughthefirstandlastinstances(selectionofreferenceandintercalibrationsites)havebothbeensupersededbymorerecentCISguidelines.Theothertworeferencesto“expertjudgment”refertosamplingfrequenciesforsurveillanceandoperationalmonitoring,andgiveMemberStatesscopetovary(withinlimits)theintensityatwhichdifferentqualityelementsareevaluated.Thisbriefanalysisindicatesthat,whilstthereareprecedentsfortheuseof“expertjudgment”intheimplementationoftheWFD,therearealsosituationswhereexperiencegainedthroughimplementationmeansthat“expertjudgment”canbereplacedbymorerigorous,evidence-basedprocedures.Thecaseofreferenceconditionsisuseful,inthepresentcontext,because,here,expertjudgmentisallowedonlywhereotherapproaches(spatiallybasedorbasedonmodelingoracombination)arenotpossible(AnnexII,1.3paragraph3).Expertjudgementisnot,strictlyspeaking,analternative;ratheralastresortinsituationswherestrongevidenceisnotavailable.
3
Thepresentdiscussionsdivideintotwobroadscenarios:
First,datafromalargenumberofspecificcases(i.e.individuallakesorrivers)wasusedtoinferageneralsituation(i.e.nutrientconcentrationsconsistentwithgoodstatus);
Second,thisgeneralsituationwasthenusedtopredictthenutrientstatusforindividuallakesandriverswhich,inturn,contributedtotheirfinalstatusassessment.
Inbothcases,thereshouldbeadirectpathwayfromtheevidencetotheoutcome.Inpractice,unaccountedvariation(partstochastic,partunmeasured)meansthatthereisrarelyanunambiguousoutcome,butratherarangeofprobabilities,recognizingthatthesamebiologicalstatecanoccuratseveralnutrientconcentrations.Theboundaryconcentrationissetfromamongstthesepossiblevaluesbased,partly,onanunderstandingofhowtheinformationwillbeusedintheregulatoryprocess(seech.7inCISGuidance).
Newcasesarethenevaluatedagainstthisboundarybut,again,variabilityinbothbiologicalandnutrientdatameansthataface-valueinterpretationmaybemisleading,especiallywheneitherthebiologicalorchemicalmeasurementsareclosetotheboundary.Inthissituation,adecisionmaynecessitatelookingbeyondthenarrowevidencebaseusedforformalassessments.
Bothsituationsrepresentthelegitimateuseof“expertjudgment”,recognizingthatitisnotpossibletocaptureallaspectsoftheinteractionsbetweennutrientsandbiologywithinthelimitedmeasurementsthatarepossiblewithinnationwidemonitoringnetworks.
Themeetingrecognizedthatexpertjudgmentcanplayanimportantrole,particularlywhenmakingdecisionsaboutindividualwaterbodieswheretherewaslowconfidenceinoutcomesfromformalqualityelements(i.e.ifanEQRwasclosetoastatusclassboundary)orifqualityelementsgaveconflictingresults.ThisexpertjudgmentmaytaketheformofadeeperunderstandingoftheecologyofparticularorganismsorcommunitiesthaniscapturedbytheBQEsorlocalknowledge.However,itisimportantthatthecriteriaonwhichsuchjudgmentsarebasedareclearandtheevidencerelevanttoeachcaseisdocumented.
Keyrecommendations:expertjudgment
• Genuinelyinformedexpertopinionisavaluableadjuncttoformalclassificationdata(BQEs,supportingelements)insomesituations.
• Useofexpertjudgmentneedstobeacknowledgedanddocumented.
• Thecriteriaonwhichjudgmentsarebasedneedtobeclear
• Mostappropriatewhenappliedatthelevelofindividualwaterbodies.
4
Settingnutrientcriteria
ThenewCISGuidancedocumentandtoolkitaddresseshowtodevelopnutrientboundariesinlinewiththenormativedefinitions.TheunderlyingassumptionisthataMemberStatehasatleastoneBQEwithastatisticallysignificant(presumedtobecausal)relationshipwithinorganicnutrients.ThisshouldallowMemberStatestocheckexistingboundariesanddevelopnewboundarieswhereappropriate;however,problemswillremainwhenthereisno(oronlyaweak)relationshipbetweentheBQEandnutrients,andwherenutrientsareinteractingwithotherpressures.
ChallengesarisewhenaMemberStatehasonlyasmallnumberofsiteswithinaparticulartype,orwherethenutrientgradientisshort.SomesolutionsaresuggestedintheCISGuidance,allattemptingtoproducelargerdatasetswith,hopefully,strongerrelationshipswithpressure.Thismayentailmergingtypeswithinacountryorlookingbeyondnationalborderstoobtaindatafromsimilarwaterbodieselsewhere.DatasetsdevelopedfortheintercalibrationexerciseofferasignificantopportunityforMemberStatestoshareinformation.Thesehavesummarydata(intercalibrationcommonmetrics,chemistry)aggregatedtointercalibrationtypesthatcouldbeutilizedbyMemberStateswhosenationaldatasetsmaynotyieldsufficientlystrongrelationshipsintheirownright.Theguidelinesforfittingnewormodifiedmethodstoexistingintercalibrationexercises(CISGuidance#30)willalsoenablebetween-countrydifferencestobeminimized(e.g.throughuseoflinearfixed-effectmodels).
AnalternativetoaMemberState“borrowing”datafromneighbourswithsimilarwaterbodytypesistodevelopgenericnutrientboundariesthatareapplicableacrossthesamebroadtypeinseveralMemberStates.ThishasbeenattemptedaspartofECOSTAT’snutrientactivities(seePhillipsetal.,2016)but,atpresent,boundariesareonlyavailableforafewbroadtypes.Oncethesecondroundofreportingiscomplete,thereshouldbemoredataavailablefromwhichsuchrelationshipscanbederived,allowingthenumberofbroadtypesforwhichboundariesareavailabletobeexpanded.
Broadtypeboundariesneedtobetreatedwithcaution,asEQRshavetobe“translated”fromnationalmetrictointercalibrationcommonmetric(ICM)scalesandtheremayalsobedifferencesinthesampling,analysisandstatisticalaggregationofchemicalvariablesbetweenMemberStates.Nonetheless,theseofferamajoropportunitytosupportMemberStatesintheireffortstodevelopeffectivenutrientclassificationcriteria.Expertjudgmentshouldbealastresortwhensettingnutrientboundaries
Keyrecommendations:nutrientcriteria
5
• EncourageuseofBPGforMemberStatestocheckanddeveloptheirownnutrientcriteria
• ProvidealibraryofICdatasetsthatMemberStatescanusetosupplementnationaldatawhendevelopingnutrientcriteria
• EncouragesharingdatabetweenneighbouringMemberStatestoenablestrongerrelationshipstobeestablished.
• Extendthecurrentrangeofnutrientboundariestocoverallbroadtype
Usingnutrientcriteriainclassification
Classificationincontext
Classificationisanadministrativesteponalongerjourneytowardsattaininggoodecologicalstatusand,thereby,securingasustainablewaterresourceforEurope.TheWFDandCISGuidancesetoutgeneralrulesforclassificationyetthereisalsoscopeforMemberStatestointerprettheserulesinamannerappropriatetotheirterritory.WhilstMemberStatesneedtoprovideasoundframeworkaroundwhichprogrammesofmeasurescanbeprioritizedanddesigned,theEUasawholeneedsconsistencyininterpretationoftheWFDtoenabletheEUtobettertrackprogressinWFDdeliveryandidentification.ThereisapotentialtensionbetweentheserightsandresponsibilitiesofMemberStatesandtheneedsoftheEuropeanCommission.
TherealsoappearstobeanissuewithimplementationofnutrientcriteriawithinMemberStates.Whilstallhaveproceduresforusingnutrientsintheirclassificationprocedures,thereisashortfallinthedatasubmittedtotheEEA,withdatamissingfor16%oflakes(byarea)and36%ofrivers(bylength).Whetherthisrepresentsanabsenceofdataorjustafailureincommunicationisnotclear.Theproportionofwaterbodieswith“unknown”statusfornutrientsvariesbetweenMemberStatesbutacomparisonofthenumberofwaterbodiesforwhichdataavailable,andthetotalareathattheserepresents,suggeststhat,ingeneral,smallerwaterbodiesarelesslikelytobemonitoredandreportedthanlargerones.
Afurtherissuewhencomparingclassificationresultsisthat,whilstBQEshave(largely)beenharmonizedasaresultoftheintercalibrationexercise,thereisnosuchconsistencyinsupportingelements.Differencesencompasschoiceofdeterminand(i.e.solubleversustotalnutrients),analyticalmethodsandsamplingfrequencyaswellasboundary-settingprocedures.Whilstthelatterisaddressedviathe“bestpracticeguide”andtoolkit,MemberStatesarenotboundtocomplytoanystandardapproach.Thereare,moreover,both
6
differencesandambiguityintherulesusedwhencombiningdatafromBQEsandsupportingelementstodetermineafinalclassification,aswellasalackofclarityinhowresultsareextrapolatedtocoverwaterbodiesthathavenotbeenmonitored.Atthesametime,itisclearthatclassificationmethodsareevolvingwithinMemberStatesinthelightofexperience,whichraisesfurtherchallengesofensuringcomparabilitybetweenresultsovertime(and,particularly,betweenRBMP)inordertobeabletotrackprogresstowardsWFDobjectives.
Classificationoutcomesmustbothprovideasoundframeworkaroundwhichprogrammesofmeasurescanbeprioritizedanddesigned
ConsistencyininterpretationofWFDwillenableEUtobettertrackprogressinWFDdeliveryandidentificationofbestpracticewillhelpMSachieveWFDgoals
7
Fig.1.Useofsupportingelementsinwaterbodyclassificationforlakes(top)andrivers
(bottom)in2ndRBMP.DatafromtheWISEFreshwatervisualizationtoolsummarizedby
area(lakes)orlength(rivers)forallMS(exceptGR,IE,LV)[finalversionswillberedrawn
andfocussolelyonnutrients]
Classificationinpractice
ECOSTATrepresentativeswerepresentedwiththreescenariosandaskedhowtheircountrywouldclassifyawaterbodyineachcase.Thescenariosandresponseswere:
• Biologyis“good”butnutrientsare“lessthangood”o Presentedwiththisscenario,mostMemberStatesclassifiedthewaterbody
as“lessthangood”status,withjustthree(AT,DE(freshwater),SK)allowingthehigh/goodbiologytooverrideevidenceofnutrientenrichment(Fig.2).
• biologyis“lessthangood”butnutrientsare“good”o TherewasgreaterunanimityamongstMemberStatesandbetweenwater
bodytypesforthisquestion,withallagreeingthatthissituationshouldresultina“lessthangood”finalclassification.
• bothbiologyandnutrientsare“lessthangood”o Asforthepreviousquestion,allMemberStateswouldclassifyawaterbody
withthesecharacteristicsaslessthangoodstatus.
ThefullresultsofthissurveyarepresentedinAppendix1.
Fig.2.MemberStateapproachestoclassificationofawaterbodywhenbiologyisathigh
orgoodstatuswhilstnutrientsarelessthangood.
SeveralMemberStatesaddedcaveatstotheiranswerswhichsuggestedareluctancetoactunlesstheywereconfidentintheface-valueclassificationgivenbyeithertheBQEor
0
4
8
12
16
20
high/good lessthangood other
8
supportingelement.Theirimmediateresponse,insuchcircumstances,wastotrytoraisetheconfidenceintheclassificationandthenextsectionswilldescribethewaysthatthisisattempted.However,therealsoappearstobedifferencesinthewayconfidenceinteractswithclassificationresults.Insomecountries,lowconfidenceinaBQEorsupportingelementmeansthat“oneoutallout”isoverriddensothatawaterbodymaybereportedtobeat“goodstatus”despitetheface-valueclassificationofoneormoreelementsbeingatmoderatestatus.Inothercountries,thefacevalueclassificationwillbereportedirrespectiveoftheconfidenceinthatclassification,butmeasuresappliedtothatwaterbodymaybelimitedifconfidenceislow.IncreasingconfidenceintheclassificationwascitedbysomeMemberStatesasanappropriatefirst“measure”incaseswhereBQEsandsupportingelementsgaveconflictingresults.
Increasingconfidenceinclassifications
Statisticalaspects
Theissueofconfidencearisesbecausetermssuchas“goodstatus”and“moderatestatus”arecategoricalexpressionsofprobabilitydistributionsderivedfromcontinuousdata(EQRs,nutrientconcentrations).AmeanEQRornutrientconcentrationrepresentsthepointofmaximumprobabilitybutthetailsoftheprobabilitydistributionmayextendintoneighbouringstatusclasses.Asprogrammesofmeasuresinevitablyplacefinancialburdensonbothregulatorsandstakeholders,itisreasonabletorequireahighlevelofconfidenceinaclassification,andparticularlyinthelinkbetweenecologicalstatusandpressures,beforeactionistaken.
Confidenceinaclassificationdevolvesintotwocomponents:proximitytoaboundaryandthereliabilityofthesamplemeanasanestimateofthe“true”(i.e.population)mean.Fig.3representshypotheticaldistributionswithidenticalmeans(0.56,assumedtoindicate“moderatestatus”)butdifferentstandarderrors.Theproximityofthemeanleadstopartofbothdistributionsextendingintotheadjacentstatusclass.Forthewiderofthese,thereisa33%chancethatthe“true”meanisingoodstatus(aswellas0.7%chancethatitis“poorstatus”).Bycontrast,thechanceofthe“true”meanoccurringingoodstatusdropsto19%inthecaseofthenarrowerdistribution.Asstandarderrorisafunctionofsamplesize,astraightforwardoption,whenpresentedwithasituationsuchasthisis,therefore,toincreasethequantityofdatauponwhichthestatuspredictionisbased.
9
Fig.3.ConfidencedensityofEQRvaluesforahypotheticalBQEassumingmean=0.56.
Curvesshowshapeofdistributionwhenthestandarderrorislarge(blue)andsmall
(green).Theverticalredlineindicatesthepositionofthegood/moderateboundary.
However,thoughtheriskofmisclassificationdropswithincreasingsamplesize,theproximityofthe“true”meantothegood/moderateboundaryinthiscasemeansthatitisunlikelytobeeliminatedentirely.Therefore,asecondoptionistolookatfurthereutrophicationindicatorstoseeifthese,too,indicateaproblem(Fig.4).Inthisexample,thecombinedlikelihoodofthetrueconditionofthewaterbodybeingatlessthangoodstatusdropsto16%.
RedrawFig.3has0.52not0.56–as1OAomeansthatFig.4isdifficulttointerpret.
BothstrategiesareusedbyMemberStatesinsituationswheretherearemismatchesbetweenstatusderivedfromBQEsandsupportingelementsandbothareclearlysensiblestrategiestoensuringconfidenceboththatawaterbodyislessthangoodstatusandthatnutrientsarethecause.However,whilstsomeMemberStatesseemtoapplytheserulestotheclassificationitself(i.e.requiring>1BQEto“fail”beforeawaterbodyisdowngraded),othersallow“oneoutallout”todeterminetheclassificationbutthenapplyrulessubsequentlytoensureconfidencebeforemeasuresareimplemented(effectively,raisingconfidenceistreatedasadefacto“measure”ininstanceswherethereareambiguousclassificationresults)
0
2
4
6
8
10
0 0.2 0.4 0.6 0.8 1
Confiden
ceden
sity
EQR
10
Fig.4.ConfidencedensityofEQRvaluesfortwohypotheticalBQEswithmean=0.56
(solidline)and0.52(dashedline).Theverticalredlineindicatesthepositionofthe
good/moderateboundary.
Keyrecommendations:increasingconfidenceinclassifications
• Instanceswhere“oneout,allout”isoverriddenshouldbedocumented.
• HarmonisedapproachestoclassificationinsituationswhenthereislowconfidenceinBQEsand/ornutrientcriteriawouldbehelpful.
• Agreementisneededonwhetherstepstoraiseconfidenceshouldaffectclassificationresultsorrepresenta“measure”isneeded
• MemberStatesneedflexibilitywhenapplyingclassificationrulesbutsharedexperiencesshouldallow“bestpractice”tobedisseminated.
Overall,itwasclearthatitismorecommonforMemberStatestooverride1OAOtoupgradeawaterbody(e.g.moveittogoodstatuswhenoneBQEorsupportingelementindicatesmoderatestatus)thantheopposite.MemberStatesregardedsuchupgradesasalegitimateresponsetosituationswheretherewasalowconfidenceinaclassification(e.g.whenthefailingBQEwasnaturallyvariable).Theoppositecase(i.e.whenawaterbodyisclassifiedasmoderatestatusdespiteallBQEsandsupportingelementsbeingatgoodstatus)wasregardedasappropriateinsituationswheretherewereknowntobepressurestowhichintercalibratedBQEswerenotsensitive.
0
1
2
3
4
5
6
0 0.2 0.4 0.6 0.8 1
Confiden
ceden
sity
EQR
11
Strengtheningtheevidence:BQEs
Theprevioussectionoutlinedtwobroadstrategiesadoptedincaseswhere1oAodictatedafailuretoachievegoodstatusbutwithlowconfidence:increasingthequantityofdatafortheBQEorsupportingelementconcernedortocollectabroaderrangeofBQEsandnon-biologicalevidence.
WhilstmonitoringthefullrangeofBQEsismandatoryforsurveillancemonitoring,MemberStateshavemoreflexibilitywhendesigningsamplingstrategiesforoperationalmonitoring.Inthecaseofasmall,homogeneouswaterbodywithasinglepressure,theuseofasingleBQEmightbedeemedtobearationaluseofalimitedbudget.However,eveniftheBQEmostsensitivetothepressureinquestionisadopted,therewillstillbealikelihoodofmismatchesbetweenBQEandsupportingelementclassifications,particularlyforsitesthatareclosetostatusclassboundaries.
Thismeansthat,exceptforafewcaseswhereimpactsareveryclear,monitoringmorethanoneBQEperwaterbodyprovidesaninsuranceagainstthescenariosoutlinedintheprevioussection.Ideally,theseBQEsshouldprovidecomplementaryecologicalinsightsintotheprincipalstressorgradient(presumedtobenutrients,inthiscase).WhilstKellyetal.(2016)suggestedadiminishingmarginalutilityfromemployingadditionalBQEsfortheassessmentoflakes(Fig.5),thisfocusedonbroad-scalesurveillancemonitoring,andthesituationwillbedifferentifthefocuswassolelyonlakesthatwereclosetokeyboundaries.Ontheotherhand,theincreaseinoverallstatisticalconfidenceasmoreBQEswereemployedmayberelativelymodest(##to##intheexampleinFig.#)andwilldependonboththerelativesensitivityofeachBQEandthestrengthofeachoftheirrelationshipstothepressuregradient.
Afinalcomplication,addressedlaterinthisreport,isthatBQEsmayberespondingtomorethanonepressure.Whilstthismaynotaffecttheoverallevaluationofstatus,itisaseriouscomplicationwhentryingtodemonstratethecausalrelationshipswithpressuresuponwhichprogrammesofmeasurescanbeplanned.
12
Fig.5.DiminishingmarginalutilityofemployingextraBQEsinlakeassessment:
simulationsassuminga1:1relationshipbetweeneachslopeandeachmetrichaving
identicalresponsetostresses(fromKellyetal.,2016).
Strengtheningtheevidence:supportingelements
Similarargumentsapplytosupportingelements:intheory,asingleexceedanceofasupportingelementstandardshouldleadtothestatusofawaterbodybeingdowngraded.Inpractice,however,MemberStatesadoptavarietyofapproaches.
Inthesimplestcases,classificationsarebasedonasinglenutrientparameterperwaterbodytype,basedonanaprioriassumptionaboutthelimitingnutrient.Thisiscombinedwithotherphysico-chemicalparametersfollowingthe“oneout,allout”rule.
Inmanycases,however,severalnutrientparametersareassessed,representingbothnitrogenandphosphorusfractions.Thesemaybeeachtreatedasindependentcomponentsofasuiteofphysico-chemicalelementswhicharethenevaluatedfollowingthe“oneout,allout”rule(e.g.Romanianfreshwaters),orareaggregatedintoasingle“nutrient”elementwhichisthencomparedwithotheraggregationstodeterminethefinalphysico-chemicalstatus.
InFrance,forexample,fiveseparatenutrientparametersaremeasuredinrivers(“PO4”,TP,“NH4”,“NO2”,“NO3”).Insuchcases,theprincipleofthedowngradingparameterisapplied(i.e.theworstofthesedeterminesstatus.However,inordertoreduceclassificationerrorswhenvaluesareneartheboundarybetween“high”and“good”statusorbetween“good”and“moderate”status,thefollowingrulesareadopted:
13
Ageneralphysico-chemicalqualityelementforwhichseveralparametersareinvolved,canbeclassifiedasHighif:
• AllBQEandtheothergeneralphysico-chemicalelementsclassifythewaterbodyasHighstatus;
• Onlyoneparameterisdowngradedto“Good”forthequalityelementinquestion.
Similarly,forGoodstatus:
Ageneralphysico-chemicalqualityelementforwhichseveralparametersareinvolved,canbeclassifiedasGoodifthefollowingconditionsaremet:
• AllBQEandtheothergeneralphysico-chemicalelementsareclassifiedasGood.
• Onlyoneparameterisdowngradedto“moderate”forthisqualityelement
Thisisnotvalidfornitratesinrivers:ifthe“nitrates”parameterisclassifiedlessthan"good"(concentration>50mg/l)thenthenutrientqualityelementisclassifiedasmoderate.
DenmarkusesnutrientdataonlyinlakeswhenoneormoreoftheBQEs(phytoplankton,macrophytesorfish)aremissing.IfthisisthecaseandtheoneortwomeasuredbiologicalelementsareinGES,butbothphosphorousandnitrogenexceedstheboundaries,thelakeisclassifiedasmoderatestatus.Demarkhasnotyetdevelopednutrientstandardsforrivers,butanewprojectisunderdevelopmentthat,basedonnutrienttoolkit,willtrytodevelopnutrients(andtheotherphysical-chemicalQEsaswell)inwatercourses.
Dissolvedoxygenisaparameterthatshould,intheory,providevaluablesupportingevidencewheneutrophicationissuspected.Inlakes,especially,hypolimneticdeoxygenationisasignificantconsequenceofeutrophication,butitcanbedifficulttogetmeasurementsthatarerobustenoughtoinfluenceclassifications.TherewaslittleevidenceinthequestionnairereturnsofMemberStatesusingthisroutinely,althoughGreecehaveanoptionforusing“expertjudgement”toincorporatehypolimneticoxygenintoclassifications.
Dissolvedoxygeninriversisprincipallyusedasanindicatoroforganicloading,ratherthanofconsequencesofinorganicnutrientsandoxygentargetsarebasedprimarilyonthetoleranceoffish.Nonetheless,recordsoffishkillsassociatedwithnighttimeanoxiaisaclear“undesirabledisturbance”resultingfromelevatednutrientconcentrationsandUKincorporatestheseaspartoftheirwider“weightofevidence”approachtoevaluatingeutrophicationimpactsinrivers.
14
Conclusions
CISGuidance##addressedtheissuesassociatedwithsettingobjectivenutrientboundariesthatwouldhelpMemberStatestoachievegoodstatusintheirwaterbodies.Thisdemonstratedthat,inmostsituations,therewasconsiderablevariationintherelationshipbetweenBQEsandnutrients.TheGuidanceaddressedthequestionofhowtoselectappropriateboundaryvaluesinlightofthisuncertainty.However,whilsttherateofclassificationmismatchescouldbereducedbyfollowingthesemethods,itcouldneverbeeliminatedentirely.ThesemismatchescontributetoareductionintheoverallconfidenceofawaterbodyclassificationunlessMemberStatestakestepsoverandabovethoseprescribedintheWFDitself.Thesestepscantakeoneofthreeforms:
• MoredatafortheBQEand/orsupportingelement(s)–intheory,reducingthestandarderrorandresultinginamoreaccuratepredictionofthetruestatus;
• DatafromadditionalBQEsand/orsupportingelements–inordertoobtainamoreholisticpictureofthestatusofthewaterbody;or,
• AdditionalsourcesofevidencetothatprovidedbyBQEsandsupportingelements(e.g.hypolimneticdeoxygenation,fishkills).
TheuncertaintiesinherentinestablishingsupportingelementstandardsmeanthatMemberStatesmusthavetheflexibilitytouseadditionalinformation(and,inparticular,localknowledge)toqualifyfacevalueclassificationresults.Therewasbroadagreementthatresponsibleuseofexpertjudgementisappropriateinsuchcircumstances,solongasthisiswell-documentedandtransparent.
Therewas,however,lessagreementonwhetherthisadditionalinformationshouldbeusedtooverridetheclassificationbasedonastraightforward“1oAo”orwhetheritshouldbeusedtoinformdiscussionsaboutappropriatemeasurestobeappliedinacatchment.IfseenasastepinalongerjourneytowardsachievementofWFDobjectivesthenwhetherornotclassificationitselfisaffectedmaybeoflittlerelevance.However,variationsinnationalapproachescouldhavesignificantimpactsonthecomparabilityofdatasubmittedtotheEuropeanCommissionaspartofRBMPreportingrequirements.
References
Borja,A.&Rodríguez,J.G.(2010).Problemsassociatedwiththe‘one-out,all-out’principle,whenusingmultipleecosystemcomponentsinassessingtheecologicalstatusofmarinewaters.MarinePollutionBulletin60:1143-1146.
Moe,S.J.,LycheSolheim,A.,Soszka,H.,Gołub,M.,Hutorowicz,A.,Kolada,A.,Picinska-Fałtynowicz,J.&Białokoz,W.(2015).Integratedassessmentofecologicalstatusand
15
misclassificationoflakes:Theroleofuncertaintyandindexcombinationrules.EcologicalIndicators48:605-615.
Prato,S.,LaValle,P.,DeLuca,E.,Lattanzi,L.,Migliore,G.,Morgana,J.G.,Munari,C.,Nicoletti,L.,Izzo,G.&Mistri,M.(2014).The‘‘one-out,all-out’’principleentailstheriskofimposingunnecessaryrestorationcosts:AstudycaseintwoMediterraneancoastallakes.MarinePollutionBulletin80:30-40.
Appendix:useofsupportingelementsinclassification–questionnaireresults
ECOSTATrepresentativeswerepresentedwiththreescenariosandaskedhowtheircountrywouldclassifyawaterbodyineachcase.Thescenarioswere:
Biologyis“good”butnutrientsare“lessthangood”biologyis“lessthangood”butnutrientsare“good”bothbiologyandnutrientsare“lessthangood”
Q1:Whatisthefinalstatusifbiologyis“good”butnutrients“lessthangood”?
19MemberStateshaveaunifiedapproachacrossbothfreshwaterandtransitional/coastalwaters(orarelandlocked)andafurtherfourdidnotspecifywhethertheirapproachappliedtoallwaterbodytypes.GermanyandItalyhavedifferentapproachestoclassificationinfreshwaterandtransitionalandcoastalwatersandtheremainingcountriesdidnotprovideinformationforallwaterbodytypes.
Presentedwiththisscenario,mostMemberStatesclassifiedthewaterbodyas“lessthangood”status,withjustthree(AT,DE(freshwater),SK)allowingthehigh/goodbiologytooverrideevidenceofnutrientenrichment(Fig.3).
16
Fig.3.Classificationofawaterbodywhenbiologyisathighorgoodstatuswhilstnutrientsarelessthangood.
Nineresponses,however,offeredmorenuancedinterpretationsofthesituation.Thesearesummarizedbelow:
DK(lakes):NoactiononlyifallBQEsaregood
FI:StatuswouldbegoodiftherearespatialandtemporalreplicatesoftheBQEsandthereislessreplicationofnutrients.However,statuswouldbemoderateiftherewerelessdataforBQEsbutmorefornutrients.
FR:Forrivers,statuswouldbemoderateifthiswasindicatedbytwophysico-chemicalparametersorifthenitratethreshold(50mgL-1)wasexceeded.Ifasinglephysico-chemicalparameter(otherthannitrate)wasinmoderatestatus,thenstatuswouldbegood.Forlakes,coastalandtransitionalwaters,statuswouldbemoderateifindicatedbytwophysico-chemicalparametersorifoneparameterindicatedthatthestatuswasbelowmoderate.Again,ifasinglephysic-chemicalparameterindicatedmoderatestatus,thenthefinalstatuswouldbegood.
IE:Forrivers,twonutrientstandardsmustfailwithconfidencebeforethewaterbodyisclassifiedasmoderatestatus;forlakes,ifanynutrientstandard/parameterfails,thelakeisclassifiedatmoderatestatus;forcoastalandtransitionalwaters,ifanutrientstandardfailstheTraCwaterbodyisclassifiedatmoderatestatus.
IT(coastalandtransitionalwaters):TRIXcombinesO2,nutrientsandchlorophyll.
LT:StatusislessthangoodifBQEsarehighandgoodbutoneormorephysico-chemicalparametersaremoderate.Inthiscase,theconfidenceofclassificationwouldbelow.
0
4
8
12
16
20
high/good lessthangood other
17
NO:StatuswouldbelessthangoodifBQEsindicatedgoodstatusbutgoodifBQEsindicatedhighstatus.
PL:Statuswouldbemoderate;however,ifonlyonephysico-chemicalparameterwasslightlybelowthemoderatestatusthresholditwouldbeignoredandstatuswouldbereportedasgood.
SI:Statuswouldbemoderate;however,iftherewaslimitednutrientdata,thewaterbodywouldbeclassifiedasgoodstatusandmonitoringfrequencywouldbeincreased.
Theseresponsesallfollowtheprincipleofclassifyingthewaterbodyaccordingtothelowestofbiologicalandphysico-chemicalcriteriabutoffervariousclausesthatincreaseconfidenceinthefinaloutcome.
Q2:Whatisthefinalstatusifbiologyis“lessthangood”butnutrientsare“good”?
TherewasgreaterunanimityamongstMemberStatesandbetweenwaterbodytypesforthisquestion,withalmostallagreeingthatthissituationshouldresultina“lessthangood”finalclassification(Fig.4).
Fig.4.Classificationofawaterbodywhenbiologyislessthangoodwhilstnutrientsareathighorgoodstatus.
Severalcountriesmadecommentsthatqualifiedtheirinterpretationofthisapproachandthesearelistedbelow.
BG:Ifthedifferenceispermanentitinitiatesinvestigativemonitoringand/orvalidationoftheclassificationsystem.
DE(coastalandtransitionalwaters):Thissituationhasneveroccurredasnutrientstandardsarestrict.
0
5
10
15
20
25
30
high/good lessthangood other
18
DK:InthepresentRBMPnutrientsareusedonlyfortheclassificationoflakes.
EE:Iftheaveragestatusbasedonnutrientsis“good”butTNorTPismoderate,thisistakenintoaccountatthepermittingstage.
NL:Inthissituation,measureswouldbeplannedtorestorebiology,andthesemayincludeadditionalmeasurestoreducenutrientsifnecessary
PL:IfoneBQEsdifferssignificantly(atleasttwoclasses)andtherearedoubts,itcanbediscarded
Q3.Whatisthefinalstatusifbothbiologyandnutrientsare“lessthangood”
AllMemberStateswouldclassifyawaterbodywiththesecharacteristicsaslessthangoodstatus.
Inwhichsituationsaremanagementactionstakenandhowdothesedifferinthefollowingsituations?
ThenextquestionattemptedtolinkclassificationstotheactionstakenbyMemberStateswhenfacedwithcombinationsofBQEandsupportingelementfailures.Therewasaconsensusontheneedtoinitiatemanagementactionswhenbothbiologyandnutrientsindicatedconditionswerelessthangoodstatus.However,theprocedurefordealingwithmoreambiguousoutcomesvaried.Manycountriesrecognizetheinherentuncertaintiesinbothbiologicalandchemicaldataasareasonforfurthermonitoringwhenjustonecomponentindicatedfailure.Thesituationoffailingbiologyassociatedwithnutrientsthatshouldsupporthighorgoodstatusledtosuggestionsofinvestigatingotherpotentialpressures.
19
Table1.Managementactivitiesassociatedwithdifferentoutcomesfrombiologicalandnutrientclassifications.
managementaction notes
BQE <good high/good <good
nutrients high/good <good <good
AT yes note yes moremonitoringdataneeded
BE-F yes yes yes
BE-W yes yes yes sometimesforHMWBtoo
BG note note yes investigatemonitoringbeforemeasures
CY yes yes yesMeasurestakenbasedonsignificantpressures,irrespectiveoffailedBQE
DE(freshwater)
note yes yes measuresforotherpressuresconsidered
DE(marine) note note yes thefirstcasehasoccurredjustoncewhilstthesecondisveryrare
DK yes note yes(lakesonly)NoactionistakenifallBQEs(phytoplankton,macrophytes,fish)aregood;actionisalsotaken(unlessexceptionsareused)iffewerBQEsareused,inconjunctionwithoperationalmonitoring.
EE note yes yes closerinvestigationrequiredifreasonforBQEfailureisnotclear.
20
managementaction notes
BQE <good high/good <good
nutrients high/good <good <good
ES note yes yes Researchisinitiatedifcausesoffailurearenotclear.
FI note yes yes casespecific:dependsonqualityofdata
FR note yes yes Finaldecisionconsidersbothecologicalstatusandnutrients
GR(freshwater)
yes yes yes
HR yes yes yes
HU note yes yes
Furthermonitoringandcloserinvestigationrequired.
Note,too,thatnutrientsarenotusedtoclassifysodiclakes;BQEsandconductivityareusedtoderivestatusforthese.
IE(freshwater) yes yes yesThereisalsoanoptionformanagementactionwhenbothareathigh/goodstatus,iftrendsinthedatapredictafailureinthefuture
IT(freshwater) yes yes yes
LT note yes yes investigativemonitoringtoidentifycauses
LU Yes yes yes
21
managementaction notes
BQE <good high/good <good
nutrients high/good <good <good
LV Yes yes yes
NO Yes yes yes
PL Yes yes yes
PT Note yes yes measuresfocusonappropriatepressures(e.g.hymooralienspecies)
RO Yes yes yes
SE Note no yesAgeneralrecommendationof10%nutrientreductionappliesifBQEsare<good
SI note(1) note(2) yes1)AdetailedanalysisofpressuresandimpactsisinitiatedincasesofBQEfailure;2)monitoringfrequencyisincreasedincasesoflowconfidence.
SK Yes note yes tofulfilrequirementsofUWWTDandNitratesDirective
UK note(1) note(2) yes1)Overall“weightofevidence”takenintoaccount(assessmentofclassificationandconfidence);2)Actiontakenwillberisk-based;low-risklow-costactionsmostlikely
22