Effects of hydrological variability on fish assemblages in small Mediterranean streams: implications on ecological assessment.

(1)

and education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling or

licensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of the

article (e.g. in Word or Tex form) to their personal website or

institutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies are

encouraged to visit:

(2)

ContentslistsavailableatSciVerseScienceDirect

Ecological

Indicators

j ou rn a l h o m e pa g e :w w w . e l s e v i e r . c o m / l o c a t e / e c o l i n d

Effects

of

natural

hydrological

variability

on

ﬁsh

assemblages

in

small

Mediterranean

streams:

Implications

for

ecological

assessment

Paula

Matono

a,b,∗

_,

_João

_M.

_Bernardo

a

_,

_Thierry

_Oberdorff

c

_,

_Maria

_Ilhéu

a,b

a_Departamento_de_Paisagem,_Ambiente_e_Ordenamento,_Escola_de_Ciências_e_Tecnologia,_Universidade_de_Évora,_Rua_Romão_Ramalho_59,_7000-671_Évora,_Portugal b_Instituto_de_Ciências_Agrárias_e_Ambientais_{Mediterrânicas}_(ICAAM),_Universidade_de_Évora,_Núcleo_da_Mitra,_apartado_94,_7002-774_Évora,_Portugal

c_UMR_BOREA,_IRD_207,_Département_Milieux_et_Peuplements_Aquatiques,_Muséum_National_d’Histoire_Naturelle,₄₃_rue_Cuvier,₇₅₂₃₁_Paris_cedex,_France

a

r

t

i

c

l

e

i

n

f

o

Articlehistory:

Received26December2011 Receivedinrevisedform22April2012 Accepted22April2012 Keywords: Intermittentstreams Naturaldisturbance Human-induceddegradation Fishassemblages Ecologicalassessment Portugal

a

b

s

t

r

a

c

t

SmallMediterraneanstreamsareshapedbypredictableseasonaleventsoffloodinganddryingoveran annualcycle,andpresentastronginterandintra-annualvariationinflowregime.Nativefishassemblages inthesestreamsareadaptedtothisnaturalenvironmentalvariability.Thedistinctionofhuman-induced disturbancesfromthenaturalonesisthusacrucialstepbeforeassessingtheecologicalstatusofthese streams.Inthisaim,thepresentstudyevaluatestheeffectsofnaturalhydrologicalvariabilityonfish assemblagesfromdisturbedandleastdisturbedsitesinsmallintermittentstreamsofsouthPortugal. Datawerecollectedoverthelasttwodecades(1996–2011)in14siteslocatedintheGuadianaandSado riverbasins.Highvariabilityoffishassemblageswasstronglydependentonhuman-induced distur-bances,particularlynutrient/organicloadandsedimentload,andonnaturalhydrologicalvariability. Naturalhydrologicalvariabilitycanactjointlywithanthropogenicdisturbances,producingchanges onfishassemblagesstructureofsmallintermittentstreams.Inleastdisturbedsites,despitethe nat-uraldisturbancescausedbyinter-annualrainfallvariations(includingdroughtandfloodevents),fish assemblagesmaintainedalong-termstabilityandrevealedahighresilience.Onthecontrary,disturbed sitespresentedsignificantlyhighervariabilityonfishassemblagesandashortandlong-term instabil-ity,reflectingadecreaseontheresistanceandresilienceoffishassemblages.Undertheseconditions, fishfaunaintegrityisparticularlyvulnerableandtheecologicalassessmentmaybeinfluencedbynatural hydrologicalvariations.Highhydrologicalvariability(especiallyifitentailshighfrequencyofdryeryears andmeaningfulcumulativewaterdeficit)mayaffecttheimpactofthehumanpressureswithsignificant andconsistentconsequencesonfishassemblagecompositionandintegrity.Inthisstudy,fishmetrics thatmaximizethedetectionofhumandegradationandminimizetheresponsetonaturalvariability werebasedontherelativeabundanceofnativespecies(insectivorousspecies,eurytopicspecies,water columnspecies,nativelithophilicspecies),relativeabundanceofspecieswithintermediatetolerance andrelativenumberofexoticspecies.Resultshighlighttheimportanceofassessingtemporalvariability onstreambiomonitoringprogramsandemphasizetheneedtoimprovetheassessmenttools, account-ingforlong-termchangesinfishassemblages,namelybyselectingthemostappropriatefishmetrics thatrespondtoanthropogenicdisturbancesbutexhibitlownaturaltemporalvariability,essentialboth inthecharacterizationofthebiologicalreferenceconditionsandinthedevelopmentoffishindexesin intermittentstreams.

1. Introduction

Mediterranean streams and their endemic ﬁsh fauna are amongstthemost threatenedecosystemsand biota worldwide

(Smith and Darwall, 2006; Hermoso and Clavero, 2011). The

∗ Correspondingauthorat:DepartamentodePaisagem,AmbienteeOrdenamento, EscoladeCiênciaseTecnologia,UniversidadedeÉvora,RuaRomãoRamalho59, 7000-671Évora,Portugal.Tel.:+351266745338;fax:+351266745395.

E-mailaddress:pmatono@uevora.pt(P.Matono).

Mediterraneanregionisoneoftheareasintheworldwhere land-scapes have most experienced a simultaneous burst of human activities, suchas intensiveirrigated agriculture, extensive sin-glecropping,tourismandurbanization.Thesouthernregions(e.g. PortugalandSpain)areparticularlyvulnerablebecauseof increas-ingwaterdemands(GasithandResh,1999;Laraus,2004)andland usechanges(e.g.Laraus,2004;Pe ˜naetal.,2007;Symeonakisetal., 2007).InPortugal,besideswaterquality(e.g.organicandnutrients inputs)andmorphologicalhumaninducedpressures,the hydro-logicaldisturbancecausedbylargeimpoundments,likeAlqueva reservoirintheGuadianacatchment,haveconsiderableimpacts

(3)

onthenativebiota,namelyfishfauna(Santosetal.,2004;Morais, 2008).Flowregulationandothermodificationsofflowregimeand runoffdecreasecausedbydammingandwaterabstractionarewell knownhumaninduced hydrologicalpressures withmajor con-sequencesinfishspeciesabundance,structureandintegrity(e.g.

GormanandKarr,1978;OrthandMaughan,1982;Freemanetal.,

2001;Murchieetal.,2008;Benejametal.,2009).Conversely,

nat-uralﬂowregimesarevitaltothemaintenanceofahealthybiotain manyriverecosystems(e.g.Poffetal.,1997;LytleandPoff,2004;

Arthingtonetal.,2006).

MostriversinPortugalarestronglyinfluencedbythe Mediter-raneanclimateand thehydrologicalregimeis stronglyaffected byprecipitationbothatintraandinter-annualscales.Asa conse-quence,especiallyinsouthernregions,manyriversaretemporary, exhibitingreducedorzeroflowduringthedryseason.This inter-mittentcharacterstronglyshapestheenvironmentalcontextto whichthebiotaissubjected.Theparticularcharacteristicsofthe Mediterraneanclimatearealsoreflectedintheirregularand unpre-dictableannualvariationofriverdischarge.ManylowlandIberian andsouthernEuropeanrivershavehugeinterandintraannualflow variationscausedbyirregularprecipitationandfrequenteventsof lowflowsrelatedtodryperiods.

Highflowshavebeenregardedasimportantnaturalprocesses maintainingacertaincommunitystructureandfunctioninriver systems(CumminsandSpengler,1978;Reshetal.,1988).Many studiesreportedthatanincreaseinstreamflowleadtoanincrease in fish density and fish assemblage diversity (Pegg and Pierce,

2002;Aarts et al.,2004; Xenopoulos and Lodge,2006).On the

contrary,low ﬂow conditions represent an important negative impactonﬁshassemblagestructure(e.g.Gehrkeetal.,1999;Pegg

andPierce,2002;Sagawaetal.,2007)andsomestudiesreported

thatreductionsinﬂow resultinimpoverishedﬁshassemblages oftendominatedbyintroducedspecies(GehrkeandHarris,2001;

Bernardoetal.,2003).

Small Mediterranean intermittent streams are particularly affectedbyhydrologicalvariability,asavailablehabitats,andtheir suitabilitytodifferentlifehistorystages,mayﬂuctuate dramati-cally.Consequently,ﬁshassemblagestendtopresenthightemporal variability(Magalhãesetal.,2002a;Bernardoetal.,2003;Clavero

etal.,2005;Mesquitaetal.,2006).Moreover,ﬁshfaunaexhibit

rel-ativelyhighresistancetoharshenvironmentalconditionsasthese speciesevolvedinachangeableandsometimesextreme

environ-ment(Almac¸a,1995).Thus,ﬁshassemblagesmayberesilientin

theshort-termtohumanpressure,whoseinﬂuencecaneasilybe misinterpreted,but eventuallyquitevulnerableonthelongrun tocontinuousandcumulativehuman-inducedchanges(Matthews

andMarsh-Matthews,2003).

Theassessmentoftheecologicalintegrityinwaterbodiesisa centralissuetowaterpoliciesandtonatureconservationingeneral. InEurope,theimplementationoftheWaterFrameworkDirective (WFD)(EuropeanCommission,2000)requirestheevaluation of ecologicalintegrityofaquaticsystemsusingbiota.Accordingly,all MemberStateshavetodeveloporadoptsuitablebioticindexes anddemonstratetheyrespondtohumandisturbances,i.e.to pres-sures,inaneffectiveway.Toattainthis goal,itisimportantto discriminatebetween theeffectsof natural vs. human-induced environmentalvariability(Oberdorffetal.,2002;Pontetal.,2006). In the context of Mediterranean rivers, a relevant issue is whetherornotthenaturalenvironmentaldisturbance,particularly thehydrologicalone,cancauseanundesirablebias,decreasingthe ecologicalassessment accuracyinsmallMediterraneanstreams. ThecrucialquestionisthenDoesthenaturalhydrological variabil-ityaffectﬁshassemblagesresponsetohumanpressureleadingtoless accurateassessmentoftheecologicalstatus?

This paperpresentsdatafrom leastdisturbedand disturbed sitesduringseveralyears,evaluatingtheeffectsofhydrological

500 1000 -500 0 Rainfall (mm) -1000 2010 2000 1990 1980 1970 1960

Cumulative surplus Cumulative deficit Total annual rainfall Average rainfall

Fig.1. Timeseriesoftotalannualrainfall,cumulativerainfallsurplusanddeﬁcit (deviationsfromthelong-termaveragerainfall)betweenthehydrologicyearsof 1960/61and2010/11foragaugingstationlocatedinDegeberiver(Guadianabasin). Thispatternisroughlyrepresentativeofthegeneralrainfallpatternobservedinall thegaugingstationsused,althoughabsolutevaluesmaydiffer.

variabilityonfishassemblagesstructureanddiscussingits possi-bleimplicationsonstreamsecologicalassessment.Specifically,the mainobjectivesare:(i)tocomparetheeffectofnatural hydrologi-calvariabilityonfishassemblagesindisturbedandleastdisturbed sites;(ii)toevaluatetherelationshipbetweenhydrological vari-abilityandtheimpactofhuman-inducedpressures;(iii)toidentify fishmetricsthatmaximizethedetectionofhumaninduced degra-dationandminimizetheresponsetonaturalvariability.

2. Methods

2.1. Studyarea

StudysiteswerelocatedinGuadianaandSadoriverbasins,two ofthemainriversintheSouthofPortugal.Thisregionis character-izedbyaMediterraneanclimate,presentinghighsusceptibilityto droughtevents(e.g.Pereiraetal.,2006).Theprecipitationregime ishighly irregularinthespatialandtemporal domains,namely regardingtheamountanddistributionofrainfall(Daveau,1977). Flowisstronglydependentontheseasonaldistributionofrain, mainlyconcentratedinOctober–March.Thehydrologicalregimes ofsampledsitesareveryvariable,withseveredroughtsandﬂoods. Thesesmallstreamsareparticularlyaffectedduringthesummer dry season(June–September),when theyarecompletelydry or reducedtoisolatedpools(inthiscase,over50%ofthestreambed maydryup).BasedontheexistinghydrologicrecordsfortheSouth ofPortugal,theinter-annualvariationofdischargereachesaratio ofapproximately100–1(ARHA,2011).

Duringthelastdecades,severedroughteventswithareturn periodof25–50yearsoccurredin1994/95,2004/05and2008/09 (Fig. 1). Other droughts were registered in 1991/92, 1992/93, 1998/99and 2003/04, allwith a return period of 10–25years. Important events with above the average rainfall occurred in 1995/96–1997/98 and 2000/01. Over the last 50-year records, frequencyand intensityof wetteranddryerperiods(deviations from the long-term average rainfall) were quite similar. How-ever, the duration of dry events was clearly higher, resulting in extendeddrought periods (cumulativedeﬁcit) (e.g.1983/84, 1994/95, 2008/09). Therefore, in this study, sampling period (1996–2011)encompassedawiderangeofhydrologicalconditions. Owingtotherelativelylowmeanannualprecipitation(Fig.1) andtheconstantincreaseonwaterdemandforpublicsupplyand agriculture,numerousimpoundmentswerebuiltinthissouthern region:67 inSado catchmentand1643in Guadianacatchment

(ARHA, 2011).Other humanimpactsonrivers aremainly

(4)

Fig.2. LocationofsamplingsitesinsmallsouthernPortuguesestreamsform Gua-dianaandSadoriverbasins:1–Cbx,2–Alg,3–Mos,4–Asm,5–Gar,6–Fqm,7– Sdg,8–Azb,9–Pec,10–Ami,11–Val,12–Saf,13–Mtg,14–Vas.

extraction and water abstraction. All these factors have been responsibleformajorchangesinaquaticecosystems,threatening thenativeﬁshfauna.

Fishassemblagesgenerally presentlow speciesrichnessand includemanyendemicspecieswithhighconservationstatus, par-ticularlyinGuadianariver(Cabraletal.,2005).Inthisbasin,the mostabundantandfrequentspeciesisroach(Squaliusalburnoides Steindachner),followedby barbelspecies (Barbusmicrocephalus Almac¸a,BarbuscomizoSteindachner,BarbussteindachneriAlmac¸a) andGuadiananase(PseudochondrostomawillkommiiSteindachner), allendemicspecies.

TheSadoriverbasiniscomparativelypoorer,presentinghigh abundanceofnon-nativespecies,namelypumpkinseed(Lepomis gibbosus L.).Themost signiﬁcantendemic species areroach (S. alburnoides),commonbarbel(BarbusbocageiSteindachner),Iberian nase(PseudochondrostomapolylepisSteindachner)andPortuguese nase(IberochondrostomalusitanicumCollares-Pereira).

2.2. Sampling

Datawerecollectedin14leastdisturbedand disturbedsites locatedinsouthernPortuguesesmallintermittentstreams(Fig.2). Sitesweresampledovera4–11yearsperiodbetween1996and 2011,thoughnotalwaysinconsecutiveyears.Siteselection cri-teriaprivilegedtheavailabilityoflong-termdataforeachsite,in ordertoallowsigniﬁcanttemporalanalysis.Onlyonesite(Val)was locatedintheSadoriverbasin,beingalltheothersiteslocatedin theGuadianariverbasin.

Sampling took place in earlyspring, following the protocol developedandadoptedbythePortugueseWaterAgency(INAG)

forPortugueserivers(INAG,2008)undertheimplementationof theWFD,followingalsoCENprotocol(CEN,2003).Surveyswere carriedoutalwaysinflowingwaterconditions,immediatelyafter the floodsand previously tothe strongreduction of flow dur-ingthesummerperiod,inordertoensurehighhabitatdiversity in the streams. Ateach site/year one streamsection was sam-pled,encompassingalltheexistingphysicallyhomogeneousunits (mesohabitats)–pool,runandriffle.Thelengthofthesampled sectionwasdefinedas20 timesthemeanwidthof thestream, withamaximumof150mlong.Fishwerecollectedusing back-packbattery-poweredelectrofishingequipment(IG200/2B,PDC Hans-GrasslGmbH,SchönauamKönigssee,Germany),wadingin shallowreaches(<1.2m)orfromaboatindeeperareas.Captured fisheswereidentifiedtothespecieslevel,measuredandreturned alivetothestream.Thesamplingmethodwasmaintainedalongthe studysites/years,althoughtheofficialprotocolswereonly pub-lishedinrecentyears.Indeed,theprotocoldevelopedunderthe

WFD(INAG,2008)incorporatedexactlythesesamplingprocedures

andCEN(CEN,2003).

Regionalvariableswereobtainedfromdigitalcartographywith freeInternetaccessandincludedaltitudeofthesite(m),drainage areaofthebasin upstreamofthesite (km2₎_and_distance_from source(km).Localvariableswereassessedduringtheﬁeld sam-pling procedure: water temperature(◦C), conductivity(␮S/cm), pH,dissolvedoxygen(mg/L),meanwaterdepth(m),meancurrent velocity(m/s)anddominantsubstrateclass(adaptedfrom Went-worthscale(GillerandMalmqvist,1998):1–mudandsand;2– gravel;3–pebble;4–cobble;5–boulders;6–boulderslarger than50cm).

Ineachsampling,humandisturbancelevelwasevaluatedusing 10semi-quantitativevariables(formerlydevelopedwithinthe EU-projectFAME(2004),availableathttp://fame.boku.ac.at):landuse, urbanarea, riparianvegetation,longitudinalconnectivityofthe riversegment,sedimentload,hydrologicalregime,morphological condition,presenceofartiﬁciallentic waterbodies,toxicity and acidiﬁcationlevels,andnutrient/organicload.Eachvariablewas scoredfrom1(minimumdisturbance)to5(maximumdisturbance)

(Table1)andthesumofthesescoresrepresentedthetotalhuman

pressurein eachsite.Siteswithscores1and/or2andonlyone variablewitha3wereconsideredleastdisturbed.

Severaladditionalphysicochemicalparameterscomplemented and supportedtheevaluationof human-induceddisturbancein each site (mainly organic/nutrient enrichment – Table 1) after laboratory measurements and analyses according to the Stan-dard Methods for the Examination of Water and Wastewater

(Clescerietal.,1998):ﬁvedaybiologicaloxygendemand–BOD5

(mg/L), chemical oxygen demand – COD (mg/L), phosphate – P2O5 (mg/L), totaldissolved phosphorous – P (mg/L), nitrite – NO2−(mg/L),nitrate–NO3−(mg/L),ammonium–NH4+ (mg/L) and totaldissolvednitrogen– N (mg/L).Theseparameters also reﬂect the water quality of the sampled sites, as their values interpretationwasbasedonthewaterfeaturesformultipleuses, accordingtothePortugueseWaterAgencyguidelines(available

at http://snirh.pt/snirh/dadossintese/qualidadeanuario/boletim/

tabelaclasses.php).

Forsamplespriorto2004,thehumanpressurevariableswere scoredusingalltheinformationexhaustivelygatheredduring sam-pling inallthesites(e.g.ﬁeldrecords,photos,physicochemical parameters),andalsofromdigitalcartographyandonline infor-mation(e.g.SNIRH,availableathttp://snirh.pt).

2.3. Dataanalysis

Inordertoevaluatehydrologicalvariabilityofsites,recordsfrom elevengaugingstations,geographicallyclosetothesamplingsites,

(5)

Table1

Description,assessmentscaleandmethods,andscoringcriteriaofthe10variablesusedtoevaluatethelevelofanthropogenicdisturbanceinsampledsites.

Variables Description Assessmentscale Score Criteria Methods

Landuse Impactoffarming/forestry

practices

Riversegment 5 >40%agriculturaluse(intensiveagriculture),

verysevereimpact(riceﬁeld)

Localexpertassessment

complementedwithCorine

LandCover(2000,2006)a

4 >40%strongimpact(areawithstrongforestry, includingclearcuts)

3 <40%moderateimpact(subsistencegardens, pastures)

2 <40%smallimpact(corkandholmoaks, high-growthforest)

1 <10%nosigniﬁcantimpacts(naturalforestand bush)

Landcoverandbankface characterization

Local 5 Irrigatedcropsand/orhighstocking

4 Horticulturalcrops,semi-intensivegrazing 3 Extensivecultures(e.g.pastures,cerealcrops,

pine,eucalyptus),extensivegrazing

2 Corkandholmoaks

1 Natural

Urbanarea Impactofurbanareas Riversegment 5 Verysevere(locationnearacitywithbasic

sanitationneeds)

Localexpertassessment complementedwithCorine LandCover(2000,2006)a

4 Town

3 Village

2 Hamlet

1 Negligible(isolateddwellings)

Riparian vegetation

Deviationfromthenatural stateoftheriparianzone

Riversegment 5 Lackofriparianshrubsandtrees(onlythe

presenceofannualplants)

4 Fragmentedvegetationwithbushesand/orthe

presenceofreed

3 Secondreplacementstep(dominanceofdense

brushwood)

2 Firstreplacementstep(presenceofshrubor treestratawithsomelevelofpreservation) 1 Potentialvegetation(presenceofshruband

treestrataaccordingtothegeo-series) Morphological

condition

Deviationfromthenatural stateofthestreambedand banks

Local 5 Transverseandlongitudinalproﬁleofthe

channelcompletelychanged,withveryfew habitats

4 Channelizedsector,missingmostofthe naturalhabitats

3 Channelizedsector,missingsometypesof naturalhabitats,butmaintainingmuchofthe shapeofthenaturalchannel

2 Poorlychangedsector,closetothenatural mosaicofhabitats.

1 Morphologicalchangesabsentornegligible

Sedimentload Deviationfromthenatural sedimentload(bothcarriedin thewatercolumnand depositedontheriverbed)

Riversegment andlocal

5 >75%ofcoarseparticlesofthestreambedare coveredwithﬁnesediments(sand,silt,clay)

Localexpertassessment 4 50–75%ofcoarseparticlesofthestreambed

arecoveredwithﬁnesediments(sand,silt, clay)

3 25–50%ofcoarseparticlesofthestreambed arecoveredwithﬁnesediments(sand,silt, clay)

2 5–25%ofcoarseparticlesofthebedare coveredwithﬁnesediments(sand,silt,clay) 1 <5%ofcoarseparticlesofthestreambedare coveredwithﬁnesediments(sand,silt,clay) Hydrological

regime

Deviationfromthenatural hydrologicalregime(ﬂow patternand/orquantity). Includesallsourcesof hydrologicalteration,suchas signiﬁcantwaterabstraction

Local 5 <50%andstrongdeviationfromthenatural

variabilityoftheﬂowregime

Localexpertassessment complementedwithSNIRH 4 <50%andmoderatedeviationfromthenatural

variabilityoftheﬂowregime

3 >50%anddurationofﬂoodperiodsclosetothe natural

2 >75%anddurationofﬂoodperiodsclosetothe natural

1 >90%andnormaldurationofnaturalﬂood periods

Local 5 <10%ofmeanannualdischarge

4 <15%ofmeanannualdischarge 3 >15%ofmeanannualdischarge 2 >30%ofmeanannualdischarge 1 >90%ofmeanannualdischarge Toxicand

acidiﬁcation levels

Deviationfromthenatural stateoftoxicityconditions, includingacidiﬁcationand oxygenlevels

Local 5 Constantforlongperiods(months)orfrequent

occurrenceofstrongdeviationsfromnatural conditions(e.g.pH<5.0,DO<30%)

Localexpertassessment complementedwithSNIRH

(6)

Table1(Continued)

Variables Description Assessmentscale Score Criteria Methods

4 Constantforlongperiods(months)orfrequent

occurrenceofstrongdeviationsfromnatural conditions(e.g.pH<5.5,DO<30–50%)

3 Occasionaldeviations(singlemeasurementsor

episodic)inrelationtonaturalconditions(e.g. pH<5.5,DO<30–50%)

2 Occasionaldeviations(singlemeasurementsor

episodic)inrelationtonaturalconditions(e.g. pH<6.0)

1 Conditionswithinthenormalrangeof

variation

Organicand

nutrientloads

Deviationfromthenormal valuesofBOD,COD,

ammonium,nitrateand

phosphateconcentrations

Local 5 >20%ofvaluesinclassesDorE SNIRH(classiﬁcationofwater

qualityformultipleuses, accordingtotheguidelines fromtheWaterNational Instituteb_),_complemented

withlocalexpertassessment 4 >10%ofvaluesinclassesDorE

3 >10%ofvaluesinclassC

2 Noobviousortoosmallsignsofeutrophication andorganicloading

1 Nosignsofeutrophicationandorganicloading Artiﬁciallentic

waterbodies

Impactrelatedtothepresence ofartiﬁciallenticwaterbodies upstreamand/ordownstream ofthesite(upstreamchangein thermalandﬂowregimes; downstreaminvasionbyexotic speciesoflenticcharacter)

Local 5 Localimmediatelydownstreamofalarge

reservoirorwithintheinﬂuenceareaofits backwater

SNIRHandavailable cartography

4 Localimmediatelydownstreamofa

mini-hydroorwithintheinﬂuenceareaofits backwater

3 Localdownstreamofareservoirorwithinthe inﬂuenceareaofthereservoir

2 Localdownstreamofamini-hydroorwithin

theinﬂuenceareaofitsbackwater

1 Noinﬂuenceofreservoirs

Connectivity Impactofartiﬁcialbarriersto ﬁshmigration

Riverbasinand segment

5 Permanentartiﬁcialbarrier SNIRH,availablecartography,

documentaldataandlocal expertassessment

4 Occasionalpassageofsomespecies

3 Passageofcertainspeciesoronlyincertain years

2 Passageofmostspeciesinmostyears 1 Nobarriersorexistenceofaneffective

pass-throughdevice

a_Caetano_et_al.₍₂₀₀₉₎_.

b _Information_available_at_{http://snirh.pt/snirh/}_{dadossintese/qualidadeanuario/boletim/tabela}_classes.php_.

insteadofflowdatabecausetherewerefewgaugingstationswith longtimeseries.Furthermore,asinsmallintermittentstreamsflow ismoredependentonrainfallpatternsthaninanyotherstreams, theuseofrainfalldataensuresacorrectassessmentofhydrological variability.Missingperiodsofrecordswerefilledusingpredictions ofregressionmodels.Thisprocedureinvolveddevelopingalinear equationtopredictmonthlyrainfallforthemissingperiodfrom nearbystations.

Usingdatafrommeteorologicalstations,andinsitu measure-mentsseveralvariableswerederived:meantotalannualrainfall, coefficient of variation (CV−standarddeviation/mean×100) of totalannualrainfall,relativefrequencyofdryerandwetteryears, meancumulativerainfallsurplusand deficit,CVofmeanwater depth,CVofmeancurrentvelocityandCVofdominantsubstrate class.Variabilityofwaterdepth,currentvelocityanddominant sub-stratewereusedasameasureofhabitatmodificationsfollowing hydrologicaldisturbances,asthesevariableshavealreadyshown ecologicalrelevanceinfishassemblagespatternsanddistribution inmostofthesampledstreams(Ilhéu,2004).Dryerandwetter yearswereidentifiedbasedonannualrainfalldeviationsfromthe long-termaveragevalue(50-yearrecords).Cumulativerainfall sur-plusanddeficitwerecalculatedusingthetheoryofruns(Yevjevich, 1967).Arunisdefinedasaportionofatimeseriesofvariablexi,in whichallvaluesareeitherbeloworaboveachosencriticallevel, yc.Consideringadiscretetimeseries,x1,x2,...,xt,...,xn,a neg-ativerunoccurswhenxtislessthanycconsecutively,duringone ormoretime intervals.Negativerunsinrainfalltimeseriesare relatedtodroughtcharacteristicsandthedifferencebetweenyc andxtisreferredasdeficit.Accordingly,positiverunsarereferred toasrainfallsurplus.Cumulativevalueswerethencalculatedas

therun-sumsofconsecutivenegative(deﬁcit)orpositive(surplus) deviationsfromthelong-termaveragerainfall,withchangesinthe signalofdeviationsresettozero.Thesecumulativevalueswere standardizedbythetotalannualrainfallineachsite,toallow com-parisonsbetweensites.

Individualhumanpressurevariablesand totalpressurewere averagedforeachsite.Theincreaseinhumanpressurealongthe studyperiodwasalsoquantifiedforeachsite,calculatingthe dif-ferenceintotalpressurebetweenthefirstandthelastsampling occasions.Accordingtotheincreaseinhumandisturbancelevel, somesiteschangedtheirclassificationfromleastdisturbedto dis-turbedduringthestudyperiod.Inthesecases,siteswereanalysed accordingtotheirclassification,dependingonthesampling occa-sion.

Fish captures werestandardized to anareaof 100m2 _in _all samplesand expressedasdensity(numberof ﬁsh/100m2_)._The studyperiod waslongenoughtoencompassat leastonemean generationtimeforallthespeciescaptured,thusallowing mean-ingfuljudgmentsconcerningpatternsofﬁshassemblagevariability

(Grossmanetal.,1990).

Different approacheswere used to access variability in ﬁsh assemblagescompositionovertime:

(i)Assemblagepersistence(P)describestherepeatedextinction and immigration of populations in ecological assemblages

(Oberdorffetal.,2001)andwasquantiﬁedas(P=1−T)where

T is the species turnover rate. FollowingEby et al. (2003),

Magalhãesetal.(2007)andOberdorffetal.(2001),turnover

ratewasdeﬁnedasT=(C+E)/(S1+S2),whereCandEarethe numberofspeciesthatcolonizedorwereextirpatedbetween

(7)

twosamplingoccasionsandS1andS2arethenumberofspecies presentineachoccasion.Trangesfrom0(noturnover)to1 (completeturnover),sopersistencerangesfrom0(no persis-tence)to1(completepersistence).Foreachsite,sequentialP valuesalongthesampledyearswereaveraged.

(ii)Stabilityoffishassemblagescompositionconsideringspecies relativeabundancewasquantifiedwithBray–Curtissimilarity coefficient(ClarkeandWarwick,1994)betweeneach possi-ble pairofsampledyearsin eachsite. Meansimilaritywas subsequentlycalculatedforeachsite.

(iii)Nonmetric Multidimensional Scaling (MDS) (Clarke and

Warwick,1994)ofsamples,basedontheBray–Curtis

similar-itymatricescalculatedin(ii)foreachsitewiththetemporal trajectoryoverlaid.Thedisplacementpatternofpointsinthe plots allowedidentifyingchangesinﬁshassemblagesalong theyears(ClarkeandWarwick,1994).Fishassemblagescould showchangesfollowedbyareturntoanearlierstate (indicat-ingacyclicpatternofvariabilitywithnolong-termdirectional shift)orprogressivedisplacementsfurtherawayfromthe orig-inalposition(indicatingadirectionalshiftincomposition). (iv)Time lag regression analysis (see Collins, 2000; Eby et al.,

2003; Magalhães etal., 2007)complemented theMDS and

wasusedtoexaminewhetherﬁshassemblageswere under-goingalong-termdirectionalchange(linearregression).The method involves the calculation of the Euclidean Distance (ED)betweeneachpossiblepairofannualsamples.Regression modelswherethenconductedbetweenEDandthesquareroot ofthetimelagseparatingthesamples.Thesquareroot trans-formationreducespossiblebiasintheanalysisresultingfrom fewdatapointsatlargertimelags.

Possible changes in ecological integrity of fish assemblages along the study period in each site were evaluated through variability in structural and functional features of fish assem-blages.Thesecommunityattributes,orfishmetrics,includedfish density,speciesrichnessanddiversity(Shannon–WienerIndex), relativeabundanceofnon-native,potamodromousandlonglived speciesandfunctionalguildsrelatedtohabitat(relativeabundance ofrheophilic, limnophilic,eurytopic,benthicand water column species), breeding (relative abundance of lithophilic and phy-tophilicspecies),feeding(relativeabundanceofomnivorousand insectivorousspecies)andtolerance(relativeabundanceof intoler-ant,tolerantandintermediatetolerancespecies)towhichcaptured specieswere assignedaccordingto publishedliterature(FAME,

2004;Ilhéu,2004;Cabraletal.,2005;Holzer,2008;Magalhãesetal.,

2008)andexpertjudgmentbasedontheavailableknowledge. A totalof 117metrics was initially considered for analysis, expressedin abundanceand numberof species,in relative and absolutetermsandcalculatedfortotalfishassemblagesandfor native species assemblages. Individual fish metrics were then screenedfornaturalvariabilitybycalculatingtheirCVinleast dis-turbedanddisturbedsites,andfortheirresponsetoanthropogenic disturbancesusingSpearmanrankcorrelationcoefficient(|r|≥0.5; P<0.05) between fish metrics and human pressure variables. Mann–Whitneytestwasusedtodetectsignificantdifferencesin fishmetricsCVsbetweenleastdisturbedanddisturbedsites(Siegel

andCastellan,1988).Onlynon-redundantmetrics,presenting

sig-nificantlylownaturalvariability(CV<50%)(Grossmanetal.,1990) underleastdisturbedconditionsovertimecomparedtodisturbed sites,andsignificantresponsestohumanpressurevariableswere selected.Asforassemblagescomposition,variabilityinfishmetrics wasquantifiedwithBray–Curtissimilaritycoefficientsforeachsite. Relationships between fish assemblages variability, human disturbances and hydrological variability were assessed with Spearmanrankcorrelationcoefficient(|r|≥0.5;P<0.05)andusing RedundancyAnalysis (RDA) (Jongman et al., 1987). The model

wastestedwithMonteCarlotest(999permutations).Correlations largerthan|0.4|wereusedingradientsinterpretation.Toaccount formulticollinearity,variablesweremaintainedinthemodelsonly iftheiradditiondidnotcauseanyVariationInflationFactor(VIF) exceeding3.Alinearordinationmethodwasselectedasa prelim-inaryDetrendedCorrespondenceAnalysishasshowna gradient lengthsmallerthan3SD(terBraakandSmilauer,1998).The pos-sibleinfluenceofspeciesrichness,totalfishdensityandlandscape gradientsinfishassemblagesvariabilitywastakenintoaccountin theseanalyses.

ForsequentialSpearmanrankcorrelationsandMann–Whitney teststheobtainedP-valueswereadjustedusingtheBonferroni cor-rection(seeWright,1992)andthesigniﬁcancelevelwassetat 0.05.

Alldatawere eitherlog(x+1)(linear measurements)or arc-sin[sqrt(x)] (percentages) transformed (Legendre and Legendre, 1998)beforecalculatingBray–Curtissimilaritiesandperforming MDS, Spearman rank correlations and RDA. Statistical analyses were performedusing the softwareStatistica 6.0 (StatSoft Inc., 2001),Primer6.0(Clarkeand Gorley,2006)andCanoco4.5(ter

BraakandSmilauer,2002).

3. Results

3.1. Environmentalandanthropogeniccharacterizationofsites Sampled streams are representative of small southern Por-tuguesestreams,showingastrongintermittentcharacter.Table2

summarizestheenvironmentalandanthropogenic characteriza-tionofsampledsitesconsideringtheperiodandyearsofsampling. Drainage area of streams ranged between 10.2km2 _and 192.9km2_._Alg,_Asm,_Cbx,_Fqm,_Gar_and_Mos,_with_drainage_areas lowerthan60km2_,_represented_the_smallest_streams._Mtg,_Ami, Pec, Saf and Vas have drainage areas around 100km2_. _Finally, Azb,SdgandValpresentedthelargerdrainageareas,withvalues between150km2_and₂₀₀_km2_._Distance_from_source_was_highly correlatedwithdrainage areaand wasthereforeexcludedfrom theanalyses.Asexpected,altitudeofsitesshowedanegative rela-tionwithdrainagearea(r=−0.74;P<0.01),withAlg,CbxandMos locatedinthehighestaltitudes,followedbyAsmandVas.Lowest altitudeswereregisteredforSafandVal.

Considering the hydrological variables, annual rainfall was stronglycorrelatedwithaltitude(r=0.71;adjP<0.01),with partic-ularlyhighvaluesoccurringinAlgandCbx(higherthan900mm). Asm and Vas showed slightly lower values (around 700 and 800mm) and Saf registered the lowest annual rainfall. In all theothersitesvalues rangedsomewherebetween500mmand 600mmrainfall.Cumulativerainfallsurpluswasalsocorrelated withaltitude(r=0.64;adjP<0.05)andcumulativerainfalldeﬁcit wasrelatedtotherelativefrequencyofdryeryears(r=0.79;adj P<0.01)and wetteryears (r=−0.79;adjP<0.01)and maximum valueswereobservedinCbx,Fqm,SdgandVas.Overall hydrolog-icalvariability,i.e.CVofannualrainfallwashighestinGar(>50%), followedbyFqm,Mos,Azb,Pec,SdgandVal(between30%and 50%).ThelowesthydrologicalvariabilitywasobservedinMtgand Vas(<20%).

Regardinglocalvariables,currentvelocityshowedhigh variabil-ityingeneraland waterdepthpresentedamoderatevariability inmostsites.Siteswiththinnersubstratetendedtoshowhigher substratevariabilityovertime(e.g.AzbandAmi)thansiteswith coarsersubstrate(e.g.ValandVas).

Duringthesamplingyears,almostallsitesregisteredanincrease inhumanpressure,exceptVal.Accordingly,mostoftheleast dis-turbed sites have changed theirabiotic classiﬁcation alongthe years(Asm,Fqm,Mtg,Ami,Azb).Allthesesitessufferedamarked

(8)

Table 2 Summary of the main environmental characteristics of sampling sites, anthropogenic disturbance level and related classification. Sites Alg Asm Cbx Fqm Gar Mos Mtg Ami Azb Pec Sdg Saf Val Vas No. years 5 7 6 8 9 11 5 6 8 7 10 6 6 4 Altitude (m) 460 282 365 230 249 329 220 180 168 175 225 135 157 270 Drainage area of the basin (km 2) 10.24 47.20 30.62 42.30 59.99 46.44 112.40 100.00 192.93 98.50 167.17 99.61 178.11 107.68 Total annual rainfall (mm) (mean ± SD) 922.00 ± 23.17 715.62 ± 26.77 922.56 ± 28.17 556.63 ± 39.35 517.54 ± 51.62 615.27 ± 34.14 509.55 ± 11.28 626.21 ± 26.18 520.21 ± 33.34 523.79 ± 35.70 529.89 ± 38.40 438.48 ± 24.62 570.86 ± 34.59 816.89 ± 16.49 Total annual rainfall (CV) 23.17 26.77 28.17 39.35 51.62 34.14 11.28 26.18 33.34 35.70 38.40 24.62 34.59 16.49 Cumulative rainfall surplus (%) (mean ± SD) 29.61 ± 37.05 36.34 ± 41.68 3.31 ± 6.53 26.25 ± 39.11 26.29 ± 39.85 23.12 ± 37.13 22.22 ± 28.23 17.57 ± 24.51 15.07 ± 25.49 17.22 ± 26.74 21.00 ± 36.22 1.17 ± 2.87 4.47 ± 7.82 25.63 ± 51.25 Cumulative rainfall deficit (%) (mean ± SD) 17.03 ± 32.16 6.06 ± 11.53 69.21 ± 91.32 146.31 ± 167.03 97.12 ± 153.76 25.74 ± 34.99 38.13 ± 52.33 14.77 ± 20.46 62.92 ± 76.67 65.95 ± 82.29 127.07 ± 152.79 69.50 ± 60.77 80.26 ± 99.97 152.02 ± 142.62 Relative frequency of wet years 0.60 0.71 0.33 0.38 0.44 0.45 0.60 0.50 0.38 0.43 0.30 0.17 0.33 0.25 Relative frequency of dry years 0.40 0.29 0.67 0.63 0.56 0.55 0.40 0.50 0.63 0.57 0.70 0.83 0.67 0.75 Water depth (CV) 24.42 24.72 36.26 34.71 35.41 40.38 62.87 17.03 37.30 8.04 47.33 72.18 31.62 17.82 Current velocity (CV) 62.47 59.43 27.03 84.07 157.95 36.54 39.56 87.76 51.13 95.59 56.44 92.97 56.53 67.09 Dominant substrate Gravel– pebble Cobble– boulders Pebble– cobble Sand– gravel Mud– sand Pebble– cobble Cobble– boulders Gravel– pebble Mud– sand Sand Sand– gravel Sand– gravel Cobble– boulders Boulders Dominant substrate (CV) 24.64 10.50 9.21 23.82 15.95 14.67 12.79 33.45 59.31 20.42 24.22 22.37 5.12 7.52 Total human pressure (mean (min–max)) 11.8 (11–13) 18.6 (14–20) 14.3 (14–15) 18.4 (16–21) 24 (21–29) 12.7 (11–14) 15 (13–18) 13 (11–17) 16.9 (12–24) 19.6 (17–22) 21 (20–22) 18.7 (18–21) 18 15 (14–16) Abiotic classification Least disturbed Least disturbed/ disturbed Least disturbed Least disturbed/ disturbed

Disturbed

Least disturbed Least disturbed/ disturbed Least disturbed/ disturbed Least disturbed/ disturbed

Disturbed

Least disturbed

increaseinhumanpressureandtwoevenbecamesomeofthemost disturbedones(AzbandFqm).OnlyAlg,Cbx,MosandVasremained leastdisturbed.Asthesesitesarelocatedathighaltitudes,a neg-ativecorrelationbetweentotalpressureandaltitudeofsiteswas observed(r=−0.5;adjP<0.05).Onthecontrary,Gar,Pec,Sdgand Saf werealwayssubjectedto meaningfulanthropogenic distur-bances,whichcontinuedtoincreasealongtheyears,especiallyin Gar.

Total human pressure and its increase were mainly due to landuse/landusechanges(r=0.78;adjP<0.01),namely through-out agriculture and livestock intensiﬁcation in the last years, degradationofriparianvegetation(r=0.94;adjP<0.001), nutri-ent/organicinput(r=0.74;adjP<0.05)andsedimentload(r=0.85; adj P<0.001). Considering that thesevariables were all signiﬁ-cantly intercorrelated(r>0.5;adjP<0.01),ripariandegradation mayoccurindependentlyorasaconsequenceoflandusepractices, leadingalltogethertoincreasesinnutrient/organicinputand sed-imentloadsinsampledstreams.Theremainingpressurevariables werefairlystableandrangedmostlybetweenhigh(score1)and good(score2)condition.

3.2. Composition,persistenceandstabilityoffishassemblages Elevennativespeciesandeightnon-nativespecieswere cap-tured during the study (Table 3). S. alburnoides was the most abundantandfrequentspeciesinmostofthesites.Squalius pyre-naicuswasparticularlyabundantandfrequentinAlg,CbxandVas. IberochondrostomalemmingiiandCobitispaludicatendedtooccur insiteswithfinersubstrate,especiallyinFqm,GarandSaf,which werealsosomeofthemosthumandisturbedsites.Nevertheless, C. paludicaalso occurredwithhighrelative abundancein Asm, MosandVas.Barbelswereallwidespreadspecies,thoughhigher abundancesand frequenciesofoccurrence wereobservedforB. microcephalusandjuvenileindividuals(Barbusspp.).B.bocageionly occurredinVal,theonlysitesampledintheSadoriverbasin,due todifferentandrestricteddistributionareasofbarbelspecies.P. willkommiishowedhigheroccurrenceandabundanceinMos,Ami andCbx.TheoccurrenceofAnaecyprishispanica,asmallendemic andendangeredspecieswasreducedtothreesitesandonlyshowed meaningful valuesinAsm.Overall,fishassemblageswere dom-inated bynativespecies intheleastdisturbed sites.Non-native speciesweremainlypresentinhumandisturbedsites,particularly inGar,Pec,Sdg,SafandVal,wheretheyrepresentedalarge per-centageofthetotalfishspeciesorevendominatedtheassemblages. Non-nativesweremostlyrepresentedbyL.gibbosusandGambusia holbrooki.

Totalnumberofﬁshspecieswithinsitesrangedfrom2to9, withlowermeanvaluesoccurringinAlgandVal.Inthe remain-ing sites,mean speciesrichnesswas verysimilar.As such, this metricwasfairly stablealongtheyears inleast disturbedsites (meanCV=21.8%)andmoderatelystableindisturbedones(mean CV=40.4%).Fishdensityshowedhighvariabilityamongstsampled sites(meanCVbetween61%and188%) withnosigniﬁcant dif-ferencesbetweendisturbedandleastdisturbedsites(adjP>0.05)

(Table4).

Persistenceandstability offishassemblagescomposition fol-lowed a similarpattern in sampledsites (r=0.57; P<0.05), yet persistencevalueswerelessvariablethanstabilitybetweensites. PersistenceregisteredthehighestvaluesinAlg,VasandCbx,and thelowestinFqmandMtg.Assemblagestabilitywashighestin AlgandCbx,andlowestinFqmandPec.Neitherpersistencenor variability in fishassemblages composition revealed significant correlations(adjP>0.05)withspeciesrichnessorwithtotalfish density(Table4).

MultidimensionalScalingandtimelagregressionanalysis pro-vided complementary results, showing different trends in ﬁsh

(9)

Table3

Listofcapturedspecies(scientiﬁcandcommonnames),occurrencetype(EndIb–Iberianendemism;EndBas–Basinendemism)inGuadianaand/orSadoriverbasinsand respectivefrequencyofoccurrence.

Species Commonname Occurrence

Type Guadianabasin Sadobasin Frequency

Anaecyprishispanica SpanishMinnowCarp EndBas x 0.05

Barbusspp.(juveniles) juvenileBarbels EndIb x 0.52

Barbusbocagei CommonBarbel EndIb x 0.05

Barbussteindachneri SteindachnerBarbel EndIb x 0.12

Barbuscomizo IberianGudgeon EndIb x 0.17

Barbussclateri SouthBarbel EndIb x 0.02

Barbusmicrocephalus Small-headBarbel EndBas x 0.52

Squaliuspyrenaicus IberianChub EndIb x x 0.51

Squaliusalburnoides Roach EndIb x x 0.86

Pseudochondrostomawillkommii GuadianaNase EndIb x 0.38

Iberochondrostomalemmingii Arched-mouthNase EndIb x 0.41

Cobitispaludica SouthStoneLoach EndIb x x 0.58

Lepomisgibbosus Pumpkinseed Nnat x x 0.57

Cyprinuscarpio CommonCarp Nnat x x 0.11

Carassiusauratus Goldﬁsh Nnat x x 0.01

Micropterussalmoides LargemouthBass Nnat x x 0.08

Herichtysfacetum ChamaleonCichlid Nnat x x 0.01

Ameiurusmelas BlackBullhead Nnat x x 0.02

Alburnusalburnus Bleak Nnat x x 0.08

Gambusiaholbrooki Mosquitoﬁsh Nnat x x 0.32

Table4

Speciesrichness(mean;min-max),ﬁshdensity(ind/100m2₎_(mean_±_SD),_assemblage_persistence_(P)_and_stability_(mean_{Bray–Curtis}_similarity_{coefﬁcient)}_in_each_site.

Sites Speciesrichness

(mean(min–max))

Fishdensity(ind/100m2₎

(mean±SD) Assemblage persistence(P) Assemblagestability (meansimilarity) Alg 2 68.85±129.45 1 93.4 Asm 6(3–9) 100.78±96.05 0.75 66.00 Cbx 4.8(3–6) 129.20±100.97 0.80 81.40 Fqm 4.6(1–7) 16.84±13.95 0.44 60.00 Gar 5.1(4–7) 60.36±50.99 0.83 71.00 Mos 4.7(3–6) 87.34±118.04 0.72 75.60 Mtg 3.8(1–9) 9.80±13.17 0.46 71.60 Ami 5.3(4–7) 24.18±14.82 0.70 64.30 Azb 5.1(4–8) 64.04±106.62 0.61 65.10 Pec 5.2(3–8) 43.16±50.64 0.63 57.90 Sdg 6.8(3–9) 28.74±20.16 0.73 63.60 Saf 5.3(4–7) 71.80±65.94 0.60 60.90 Val 3(2–5) 11.69±11.22 0.54 72.40 Vas 5.8(5–6) 20.35±31.26 0.88 77.60

assemblagesvariabilityforsamplingsites(Table5,Figs.3and4).

Fqm,Gar,Ami,Azb,Pec,Sdg,SafandValshowedsignificantlinear patternsofchangesinfishassemblagesconsideringbothMDSand timelagregressionresults.Despitethelowvaluesobtainedforthe determinationcoefficient(R2₎_in_several_sites,_particularly_in_Fqm andAzb,theregressionmodelsandalltheequationscoefficients

Table5

ResultsfromMultidimensionalScaling(MDS)andtimelagregressionanalysisin eachsite.

Sites Timelagregression(linear) MDS

R2 _Trend _Trajectory

Alg n.s. – Cyclic

Asm n.s. – Random

Cbx n.s. – Cyclic

Fqm 0.10;P<0.05 Directional Directional

Gar 0.24;P<0.01 Directional Directional

Mos n.s. – Cyclic

Mtg n.s. – Random

Ami 0.52;P<0.01 Directional Directional

Azb 0.17;P<0.05 Directional Directional

Pec 0.29;P<0.01 Directional Directional

Sdg 0.23;P<0.001 Directional Directional

Saf 0.29;P<0.05 Directional Directional

Val 0.31;P<0.05 Directional Directional

Vas n.s. – Cyclic

weresigniﬁcant,conﬁrmingtheexistenceofalinearpattern

under-lyingthedata.Alg,Cbx,MosandVasdidnotshowsigniﬁcantresults

fortimelagregression,butMDSplotsrevealedtheexistenceof

cyclicpatternsinﬁshassemblages.ForAsmandMtgresults

sug-gestarandompattern,withnoevidenttrendinﬁshassemblages

changes.Overall,resultssuggestatrendawayfromcyclicpatterns

towardslinearpatternswiththeincreaseintotalhumanpressure.

3.3. Structuralandfunctionalchangesinﬁshassemblages

ThemeanCVsofﬁshmetricsexpressedintermsofabsolute

abundancewereextremelyhigh(>100%).Onthecontrary,ﬁsh

met-ricsbasedonrelativeabundance,absoluteandrelativenumberof

speciesrevealedlowervariability,particularlyinundisturbedsites

(<50%).32ﬁshmetricswithlowtemporalvariabilityrevealed

sig-niﬁcantdifferencesbetweenleastdisturbedanddisturbedsites.

Afterfurtherscreeningforsigniﬁcantresponsestohuman

pres-surevariables,sevenmetricswereselectedtoevaluatepossible

ecologicalchanges inﬁshassemblages(Table6):relative

abun-danceofnativespecies,relativeabundanceofnativeinsectivorous species, relative abundanceofnativeeurytopic species,relative abundance of nativewater column species, relative abundance of nativelithophilicspecies,relative abundanceofspecies with intermediatetoleranceandrelativenumberofnon-nativespecies. Despitetheobservedredundancybetweenmetricscalculatedfor

(10)

P. Matono et al. / Ecological Indicators 23 (2012) 467–481 475 Table6

Fishmetricsselectedtoevaluatestructuralandfunctionalvariabilityoffishassemblagesinsites.Differencesinthecoefficientsofvariation(mean±SD)betweenleastdisturbedanddisturbedsitesaresignificantatP<0.001,and significantcorrelations(|r|≥0.5;adjP<0.05)withhumanpressurevariablesarelisted.

Fishmetrics Coefﬁcientofvariation Signiﬁcantcorrelationswithhuman pressure

Alg Asm Cbx Fqm Gar Mos Mtg Ami Azb Pec Sdg Saf Val Vas

Leastdisturbed Disturbed Relativeabundanceof

nativespecies

1.5±2.3 54.8±30.1 Totalhumanpressure −0.65 100 84.42±17.76 100 74.59±32.90 61.79±32.10 99.28±1.67 98.55±3.24 82.28±33.51 76.14±17.52 67.05±27.9749.93±37.11 47.46±43.08 27.60±22.26100 Landuse −0.54

Sedimentload −0.64 Organic/nutrientload −0.60 Ripariandegradation −0.59 Relativeabundanceof

nativeinsectivorousspecies

21.0±14.1 81.4±35.7 Totalhumanpressure −0.50 100 49.77±27.26 85.61±9.17 46.42±33.67 41.34±27.17 71.74±24.38 87.14±14.24 44.82±30.11 52.75±28.37 58.53±29.8938.90±33.88 21.72±23.32 7.55±11.14 70.44±16.52 Landuse −0.50

Sedimentload −0.50 Relativeabundanceof

nativeeurytopicspecies

24.7±9.2 81.1±41.3 Totalhumanpressure −0.55 79.27±11.4959.81±30.64 76.63±13.7939.93±,34.1729.32±23.68 75.84±19.14 83.47±16.80 62.25±27.52 62.77±18.97 34.89±34.0642.45±32.46 28.90±27.49 6.76±10.49 59.47±23.87 Landuse −0.56

nativewatercolumn species

12.3±9.7 76.2±39.3 Landuse −0.50 100 65.96±30.33 87.16±8.49 65.36±33.70 54.01±33.10 74.33±17.58 92.99±7.42 51.58±24.89 65.09±25.27 36.34±36.5444.35±34.70 26.77±27.12 7.30±11.32 75.78±10.85 Sedimentload −0.50

Organic/nutrientload −0.50 Relativeabundanceof

nativelithophilicspecies

5.9±3.6 65.7±29.7 Totalhumanpressure −0.63 100 63.51±31.91 95.79±6.01 67.26±34.42 54.11±33.18 92.64±8.82 96.62±4.64 80.42±32.66 76.14±17.52 39.61±37.8449.38±36.69 37.32±35.59 27.35±22.6286.48±10.09 Landuse −0.57

specieswithintermediate tolerance

5.9±3.6 73.0±40.9 Totalhumanpressure −0.62 100 63.51±31.91 95.79±6.01 67.26±34.42 54.11±33.18 92.64±8.82 96.62±4.64 80.42±32.66 76.14±17.52 39.61±37.8449.38±36.69 37.32±35.59 7.30±11.32 86.48±10.09 Landuse −0.55 Sedimentload −0.60 Organic/nutrientload −0.61 Ripariandegradation −0.52 Relativenumberof non-nativespecies

27.7±60.5 62.4±37.1 Totalhumanpressure 0.68 0 19.40±11.69 0 16.61±11.97 35.58±10.42 5.61±9.78 4.44±9.94 14.29±23.90 29.11±11.08 42.26±22.1044.90±26.15 31.75±25.90 51.11±8.61 0 Landuse 0.58

Sedimentload 0.65 Organic/nutrientload 0.62 Ripariandegradation 0.55 Morphologicalcondition 0.50

(11)

1 5 2 (Mos) n.s. 0 5 1 R2 024 2 R2= 0,24 5 1 5 (Gar) Euclidean distance 0 4 3 2 1 0 Time lag (√t)

Fig.3.TimelagregressionanalysisforsampledsitesMosandGar,showing exam-plesofabsentanddirectional(linearregression)trendsinﬁshassemblagesoverthe studyperiod. 2D Stress: 0

(Alg)

1997

200

4 2010

(Alg

1997

199

8 2011

2000

2D Stress: 0

(Ami)

1996

199

7 1999

201

0

201

1

199

6

Fig.4. 2-dimensionalMDSconﬁgurationplotsforsampledsitesAlgandAmiwith superimposedtemporaltrajectory,showingexamplesofdirectional(progressive movefurtherawayfromtheinitialposition)andcyclic(displacementfollowedby areturntoanearlierposition)changesinﬁshassemblagesoverthestudyperiod. Thestressvaluesobtained(<0.05)indicateanexcellentordinationdiagram,withno prospectofmisinterpretation(ClarkeandWarwick,1994).

totalﬁshassemblagesandfornativeassemblages,nativemetrics behavedbetter,regardingtheselectioncriteria.

Alltheselectedmetricsshowedsigniﬁcantcorrelations(|r|>0.5; adjP<0.001)mainlywithtotalhumanpressure,sedimentloadand organic/nutrientenrichment,althoughmeaningfulrelationswere alsoobservedwithlanduse,degradationofriparianvegetation,and morphologicalcondition(Table6).Thenon-nativemetricwas pos-itivelyrelatedwithhumanpressurevariables,whereasthenative metricsshowedandinverseresponse.

Meanvaluesoffishmetricsineachsitereflectedtheobserved relationshipwithhumandisturbance(Table6).Assuch,least dis-turbedsitespresentedatrendtowardshighproportionofnative metricsandlowproportionofnon-nativespecies,whilstdisturbed sitesshowedanoppositetrendalongthehumanpressuregradient. Accordingly,Alg,Cbx,Mos,VasandMtgshouldpresenthigherfish assemblagesintegrityoverthestudyperiodthantheothersites, particularlyGar,Pec,Sdg,SafandVal.

Meansimilaritiesoffishmetricsinsampledsitesfollowedthe patternalreadyobservedforfishassemblagescomposition,though withhigher values. Furthermore, metrics similaritiestended to presenthigherdifferencesbetweenleastdisturbedanddisturbed sitesthansimilaritiesoffishcomposition.Asforfishcomposition, fishmetricsdidnotpresentsignificantcorrelationswithspecies richnessandtotalfishdensity(adjP>0.05).

3.4. Relationshipsbetweenﬁshassemblagesvariability,human pressureandhydrologicalvariability

Variabilityinfishassemblageswasassociatedtohuman pres-sure and hydrological variability. Fish composition similarities showedsignificantcorrelationswithaltitude(r=0.64;adjP<0.05), annual rainfall(r=0.61;adjP<0.05)and CVof current velocity (r=−0.51; adj P<0.05), but alsowith total pressure (r=−0.66; corrected P<0.01),ripariandegradation (r=−0.59; adjP<0.05), sedimentload(r=−0.7;adjP<0.05)andorganic/nutrient enrich-ment (r=−0.5; adj P<0.05) (Table 7). Likewise, fish metrics similaritiesrevealedsignificantcorrelationswithaltitude(r=0.69; adjP<0.01),annualrainfall (r=0.58;adjP<0.05),CVof annual rainfall (r=−0.53; adj P<0.05), total pressure (r=−0.72; adj P<0.01), land use (r=−0.59; adjP<0.05), riparian degradation (r=−0.62;adjP<0.01),sedimentload(r=−0.76;adjP<0.01)and organic/nutrient input(r=_−0.50;adj P<0.05) (Table 7).On the otherhand,assemblagepersistencewasonlydependenton alti-tude(r=0.71;adjP<0.01)andannualrainfall(r=0.60;adjP<0.05) ofsites.Despitetheabsenceofsignificantresults,atrendtowards lowervalueswasobservedinmostdegradedsites.

However, total human pressure was also related to hydro-logical variability, showing signiﬁcantcorrelations with annual rainfall(r=−0.61;adjP<0.05)andCVofannualrainfall(r=0.59; adjP<0.05).Moreover,sedimentloadwasrelatedwiththeCVof annualrainfall(r=0.72;adjP<0.01)andnutrient/organic enrich-ment decreased with the relative frequency of wetter years (r=−0.64;adjP<0.05)andincreasedwiththerelativefrequency ofdryeryears(r=0.64;adjP<0.05)andcumulativerainfalldeﬁcit (r=0.61;adjP<0.05).

RDAresultswereinaccordancewiththeobservedcorrelations, furthercontributingtothegeneralinterpretationofmajor influ-encesinassemblagesvariabilityandpossiblechangesinecological integrity. The first two axes (1=0.71; 2=0.07) accountedfor 78%ofthetotalvarianceinfishassemblagesvariability (consid-eringbothcompositionandstructuralandfunctionalfeatures)and ninesignificant(P<0.05)variableswereincludedintheordination model.Accordingtocanonicalcoefficientsand inter-set correla-tions,axis1wasmainlydefinedbyannualrainfall(r=−0.76)and totalhumanpressure(r=0.70),andtoalessextentbyCVofcurrent

(12)

Table7

Mostrelevantcorrelations(|r|≥0.5;adjP<0.05)betweenfishassemblagesvariability(fishcompositionsimilarities,fishmetricssimilaritiesandassemblagepersistence) andhydrologicalandanthropogenicvariables.

Variables(hydrologicalandanthropogenic) Correlations

Fishcompositionsimilarities Fishmetricssimilarities Assemblagepersistence

Altitude r=0.64;P<0.05 r=0.69;P<0.01 r=0.71;P<0.01

Annualrainfall r=0.61;P<0.05 r=0.58;P<0.05 r=0.60;P<0.05

CVofannualrainfall n.s. r=−0.53;P<0.05 n.s.

CVofcurrentvelocity r=−0.51;P<0.05 n.s. n.s.

Totalhumanpressure r=−0.66;P<0.01 r=−0.72;P<0.01 n.s.

Landuse n.s. r=−0.59;P<0.05 n.s.

Ripariandegradation r=−0.59;P<0.05 r=−0.62;P<0.01 n.s.

Sedimentload r=−0.70;P<0.05 r=−0.76;P<0.01 n.s.

Organic/nutrientload r=−0.50;P<0.05 r=0.50;P<0.05 n.s.

velocity(r=0.54)andCVofannualrainfall(r=0.41),whileaxis2

wasmorerelatedtotheincreaseinhumanpressure(r=0.63).

Ordination diagram (biplot of sites×human pressure and

hydrologicalvariables)(Fig.5)showedagoodspatialdispersion

ofsites,especiallyalongthefirstaxis.Leastdisturbedsites(Alg, Cbx,MosandVas)showedthehighestfishassemblagesstability andweremainlyassociatedtohighannualrainfallandcumulative surplus.Onthecontrary,mostdisturbedsites,especiallyGar,Sdg, SafandPec,experiencedmeaningfulchangesinfishassemblages andsimultaneouslysufferedtheinfluenceofhighhydrological vari-abilityrepresentedbyCVofannualrainfallandhabitatdisturbance (CVofcurrentvelocityandsubstrate).Cumulativerainfalldeficit wasoneoftheleastimportantvariables.However,thisvariable wasmoreassociatedtoassemblageschangesinthemostdisturbed sites,eventhoughhighvalueshavebeenalsoregisteredinleast disturbedones.

4. Discussion

Underthe need for developing reliable biological indicators basedonﬁshassemblagesin a hydrologicallyvariable environ-mentasMediterraneanclimatestreams,thisstudyanalysedthe

.0

1

.

pressure inc

Mtg Azb Pec surplus subst (cv) Asm Mos Ami V rainfall rainfall (cv) deficit depth (cv) _{cvel (cv}₎ pressure Cbx Fqm Vas Gar Sdg Saf 0 . 1 0 . 1 --1 .0 Alg Val

Fig.5. OrdinationdiagramofRedundancyAnalysis(RDA)betweenﬁshassemblages variability(Bray–Curtissimilaritiesofsitesalongtheyearsusingﬁshcomposition andmetrics),humanpressurevariablesandhydrologicalvariables.

roleofhydrologicalvariabilityontheresponsesofﬁshassemblages tohumandisturbanceinsmallintermittentstreams.

Inthiscontext,spatialandinter-annualvariabilityoffish assem-blageswasanalysedbasedonrelativeabundanceoffishspecies andstructuralandfunctionalassemblageattributes(i.e.fish met-rics).Fishassemblageattributes,whichrespondtoanthropogenic disturbances but exhibit low natural temporal variability, are potentially themostsensitiveindicatorsofhumanpressuresto useinbioassessment(Paller,2002).Sevenmetricsfulfilledthose criteria and werethus selected toindicate possible changes in fishassemblesintegrityineachsite:relativeabundanceofnative species,relativeabundanceofnativeinsectivorousspecies,relative abundanceofnativeeurytopicspecies,relativeabundanceofnative water column species, relative abundance of native lithophilic species,relativeabundanceofspecieswithintermediatetolerance andrelativenumberofexoticspecies.Thesemetricsarethenthe mostappropriatetoincludeinapossiblemultimetricindexbased onfishfaunainsmallintermittentMediterraneanstreams.

Highvariabilityoffishassemblageswasmainlyassociatedwith thehuman-induceddisturbance,particularlynutrient/organicload and sediment load. Both correlations and RDA results showed strongassociationsofBray–Curtissimilaritycoefficientsbasedon fishcompositionandmetricswithhumanpressurevariables. Dif-fusepollutionfromagricultureispossiblythemajorhumanimpact ontheaquaticsystemsofsouthernPortugal,thoughlivestockmay haveanimportantcontributiontotheorganicloading.Theabsence of fencesalongthestreams is verycommon,allowingcattleto invadethestreams,oftenwithdestructionofriparianvegetation, factorsthattogetherleadtoanincreaseinwatereutrophication andecosystemdegradation(e.g.Vidonetal.,2008).

Fish assemblages variability was also related to hydrologi-cal variability, mainlyexpressedby CVof meanannual rainfall and alterations in habitat conditions following hydrological disturbances, particularly current velocity and substrate type. TheseresultsagreewithpreviousstudiesonMediterranean-type streams,presentingawideinter-annualvariationonﬁsh assem-blagesassociatedtothetypicalhydrologicalﬂuctuationsobserved

(Filipeetal.,2002;Magalhãesetal.,2002a,2007;Bernardoetal.,

2003; Claveroetal.,2005;Mesquitaet al.,2006;Ferreiraetal.,

2007).In leastdisturbed sites,despite thenatural disturbances causedbyinter-annualrainfallvariations(includingdroughtand floodevents),long-termassemblagestability seemstobe main-tained.ThiswasfurtherconfirmedbytimelagregressionandMDS analyses,asallleastdisturbedsitesshowedcyclictemporal pat-ternsofchangesinfishassemblagesthatunderpintheexistence ofadrift-and-recoverypatternandahighresiliencetonatural dis-turbances(Magalhãesetal.,2007).Onthecontrary,indisturbed sites,adirectionaltrendpatternwasobserved,reflectingadecrease onthe resistanceand resilienceof fishassemblages.In moder-atedisturbanceconditions,noevidenttrendinfishassemblages changeswasobserved.Thehighlyvariablefloodsanddryingevents

(13)

Natural va

riabili

ty

(long-term stability)

Natural Streamflow

Disturbance

Water quality

DO and pH variations turbidity

Human-induced

Disturbance

Degradation

Standing vs flowing waters Changes of spatial

heterogeneity, Sediment flushing vs. siltation

Eutrophication Toxicity

High

va

riabili

ty

( short and long-term instability)

Dilution, sediment washing vs Desiccation, water stagnation , nutrient concentration

reproduction, recruitment, mortality, alien species invasion

Habitat

Reproductive habitats Reproductive migrations Water volume in dry season Longitudinal connectivity

Poss

ible Change

s

Dilution, debris washing vs

Water stagnation , Organic and nutrient concentration

reproduction, recruitment, mortality, alien species invasion

No change

s

Fish Asse

mblages

Asse

mblages I

ntegrit

y

Fig.6.Potentialanthropogenicandhydrologicalmechanismsdeterminingvariationinﬁshassemblages.

inMediterraneanstreamsmayhaveactedasselective environmen-talﬁltersoververylongtimescales,therebyreducingthesubsetof speciestothoseevolutionarilyadaptedtocopewiththeprevailing harshenvironmentalpatterns(sensuPoff,1997).Onthecontrary, disturbedsitespresentedmuchhighervariabilityofﬁsh composi-tionandmetrics,anddirectionaltemporalchanges,henceashort andlong-terminstabilitypossiblyentailingchangesinassemblages integrity.Similarresultswerefoundinotherstudies(e.g.Karretal.,

1987;Schlosser,1990;Tayloretal.,1996;Paller,2002)showing

thatﬁshassemblagesexhibithighertemporal variabilityin dis-turbedenvironments.

Indryeryears,lowflowconditionsdirectlyaffectthe availabil-ityofandaccesstohabitats,namelythereproductiveones,and, decreasingthewatervolumeandwettedarea,affecttosomeextent thelifeoffish,aswell.Asaresult,reproductionandrecruitment ofthenativefisharenegativelyaffectedandsusceptibilityto pre-dationincreasescausedbyconfinementinsmallerwaterbodies. For non-nativespecies withbenthic affinities,e.g. centrarchids, carp,ormosquitofish,streamconditions indryer yearsbecome moresuitable.Butinhumandisturbedsystems,especiallyin nutri-ent/organicenrichedstreams,duringdryperiodstheaquaticbiota faceadditionalpressures.Lowflow causestheintensificationof eutrophication,asnowashingoutoffinerandorganicsediments andnodilution ofnutrientsand biomasstakeplaceand slower runningwatersaremorefavourabletoprimaryproduction.Under thesecircumstances,thefishassemblagesaresubjecttothe natu-ralhydrologicaldisturbanceandtoitsconsequencesonthewater qualitylevel,aswell.

Intheleastdisturbedsites,fishassemblageshavetocopewith thehabitatchangescausedbylowflowsandsubsequentlyrecover, whichcausestheobservedcyclicpattern(alternatingsomeloss ofintegrityandrecovery).Inthemoredisturbedsites,lowflow conditionsenhancethepressuresexerteduponfishassemblages andnosimilarrecoveryisobserved,originatingaquitedifferent patternofresponseovertime.

Inthisstudyneitherthenumberofspeciesnorthetotalfish den-sitywereshowntoinfluencefishassemblagesvariability,leadingto possibleconfoundingresults,assiteswithhigherspeciesrichness andabundancearelikelytobelesssusceptibletoenvironmental disturbancesthansiteswithimpoverishedfishfauna,whichare morepronetolocalextinctions(Tilmanetal.,1998).

Alongthestudyyears,mostsampledstreamsweresubjectedto anincreaseinhumanpressure,especiallythoselocatedinstreams nearbymorepopulatedareas.Onthecontrary,streamslocatedat higheraltitudes,inisolatedareaswithlimitedordifficultaccess, sufferedthelowestincreaseinhumandisturbanceandmaintained leastdisturbedabioticclassificationoverthestudyperiod.Thisfact wasdeterminantintheobservednegativerelationbetweenhuman pressureandaltitude,andconsequentlywithtotalannualrainfall, whichnormallypresenthighvaluesinaltitudeareas.Therefore, theexistenceofalandscapegradientinfishassemblages variabil-itydefinedbyaltitudeactuallyreflectsaprevailinghumanpressure gradient.Regardingotherpossibleconfoundinglandscapefactors ininterpretingtheimportanceofhumanpressureandhydrological variabilityinfishassemblagesvariability,resultsalsodidnotshow anydirectinfluence ofdrainagearea(Horwitz,1978;Schlosser,

1982).

Consideringthatinter-annualhydrologicalvariabilitytendedto belowinhigheraltitudestreams,thediscussedincreasein anthro-pogenic disturbance observed in many lowland sites was also partiallyresponsible forthepositive relationobservedbetween totalhumanpressure andhydrologicalvariability. Nevertheless, resultsalsosuggestthathydrologicalvariabilitymayaggravatethe impactofhumanpressures(andtoacertainextenttheir evalu-ation)infishassemblages,asreflectedbytherelationsbetween sedimentloadandtheCVofannualrainfall,andalsobetween nutri-ent/organicloadand therelative frequencyofwetterand dryer yearsandcumulativerainfalldeficit.

Accordingly,insiteswithhighannualrainfallandlower inter-annual hydrological variability high ﬂows are more frequent,

(14)

leading to habitat rearrangement and flush out of fine sedi-ments and organic material. In fact, the mentioned negative correlationbetweenthe relative frequency ofwetter years and organic/nutrientpressureisrelatedtothefactthathighflow condi-tionsmaypromotenutrientdilution,improvingthewaterquality and reducingthe impactof organic/nutrient load onthebiota. Moreover,wetteryearsfavourthereproductionofnativespecies

(Bernardoetal.,2003;Hilletal.,1991)aswellaspromoting

post-summerrecolonizations and spawningmigrations upthe river systems(BernardoandAlves,1999;Magalhãesetal.,2002b;Ilhéu, 2004).Severalstudiesreportedincreased recruitmentof native juvenilesfollowingﬂowdisturbance,causingrapidrecoveryofﬁsh assemblages(e.g.Bohnsack,1983;ClossandLake,1996;

Lobón-Cerviá,1996).

Onthecontrary,insiteswithlowerannualrainfallandhigh hydrologicalvariability,thefrequentandcumulativeabsenceof floodscanactattwolevels(Bernardoetal.,2003):(i)decreasing habitatcomplexityanddiversity,promotingsiltationand degra-dationof thewater quality, negatively affectingthenative fish habitatsincludingthereproductiveonesand thoserequiredfor earlylife-historystages;(ii)favouringconditionsforthe recruit-ment and no flush out of the alien limnophilic species, thus increasingthepressureonthenativespeciesthroughcompetition andpredation.

Results from the sites where abiotic classiﬁcation changed duringthestudyperiod (Mtg,Asm,Ami, Azband Fqm)further reinforcedtheseconclusions.Infact,changesinﬁshassemblages werelowerwhenhydrologicalvariabilitywasalsolower(Asmand Mtg),despitethemoreorlesssimilarincreaseinhumanpressure registeredinthesesites.

In sum, results suggest that hydrological variability can act jointlywithanthropogenicdisturbances,producingbothdirectand indirecteffectsonfishassemblagesinsmallintermittentstreams (Fig.6).Inhumandisturbedsituations,highhydrologicalvariability (especiallyifitentailshighfrequencyofdryeryearsand meaning-fulcumulativewaterdeficit)mayaffecttheimpactofthehuman pressuresbothonhabitatandonwaterquality,withsignificantand consistentvariationsoffishassemblagescompositionandintegrity thatmayinfluencetheecologicalassessment.Indeed,insiteswith variablehydrologicalregimesanddisturbancehistories, substan-tiallydifferentconclusions frombioassessment of asite can be obtainedfromdifferentyears,evenifdatawascollectedidentically oneachsamplingoccasionandnochangeinhumandisturbance wasregistered(Wilcoxetal.,2002).

Thesynergetic effectsof hydrological variability and human pressures on fish assemblages, acquires particular importance underaclimaticchangescenario(seeSantosandMiranda,2006) thatforeseessignificantimpactsnotonlytowardsanincreasein temperature,butalsotowardsreductionsinrainfallandincreases ininter-annualflowvariabilityinMediterraneanstreams. Conse-quently,theseaquaticecosystemsarelikelytobehighlyvulnerable tolong-termchangesinhydrologicregimeinthefuture.The selec-tion of metrics for the fish multi-metric indexes should avoid possiblebiasrelatedtometrics,whichmightrespondtonatural hydrologicalvariations.

5. Conclusions

Highhydrologicalvariabilitymayaffectﬁshassemblages com-position,butnativeﬁsh speciesseem toexhibit highresilience tonaturaldisturbancewhenhydrologicalandhabitatconditions resemble the pre-disturbed state. Despite the relatively small numberofsitesanalysed,thisstudycontributestohighlightthe importanceofassessingtemporalvariabilityonstream biomon-itoring programs. This approach makes possible to distinguish

substantive ecologicalchanges from natural fluctuationsof fish assemblage structure.In this study,thefishmetricsselectedto assess changes in assemblages integrityare based on the rela-tiveabundanceofnativespecies(insectivorousspecies,eurytopic species, water column species, nativelithophilic species), rela-tiveabundanceofspecieswithintermediatetoleranceandrelative number of exotic species. These metrics are the most appro-priate for inclusionin a multimetricindex tobe developed for thesestreamsbecausetheyrespondtoanthropogenicdisturbances but exhibit low natural temporal variability. Nevertheless, re-evaluationofthisissuewithlargerdatasetsfromdifferentriver types and longer time series including extreme natural events wouldhelptoimproveandexpandthepresentresults.Moreover, acorrectionofclassboundariesforthedevelopedbioticindexes mayalsobeapossiblesolutioninordertoimprovetheassessment accuracyinsmallMediterraneanclimatestreams.

Acknowledgements

This study gathered data collected under several different programmes: (i) assessment of the ecological flow in Guadana river basin, partially funded by the National Water Agency (INAG, Instituto Nacional da Água); (ii) the implementation of theWaterFrameworkDirective,partiallyfundedbytheNational Water agency; and (iii) climate change and fish communities of Mediterranean-type streams – potential impact on the bio-integrity and implications onthe ecological status assessment, fundedbyFCT(FundaçãoparaaCiênciaeTecnologia).P.Matono wassupportedbyagrantfromFCTandlaterbyagrantfromIIFA (InstitutodeInvestigaçãoeFormaçãoAvançada,Universidadede Évora).Specialthanksareduetoallthecolleaguesandvolunteers fromtheUniversity ofÉvora whocollaboratedin thefieldwork alongthestudyyears.WealsothankEng.ManuelaCorreiafor help-ingwiththeproductionofthestudyareamap.TheNationalForest Authorityprovidedthenecessaryfishingpermits.

References

Aarts,B.G.W.,VandenBrink,F.W.B.,Nienhuis,P.H.,2004.Habitatlossasthemain causeofslowrecoveryofﬁshfaunasofregulatedlargeriversinEurope:the transversalﬂoodplaingradient.RiverRes.Appl.20,3–23.

Almac¸a,C.,1995.FreshwaterﬁshandtheirconservationinPortugal.Biol.Conserv. 72,125–127.

ARHA,2011.PlanodeGestãodasBaciashidrográficasintegradasnasRegiões Hidro-gráficas6e7.MinistériodoAmbiente,doOrdenamentodoTerritórioedo DesenvolvimentoRegional.AdministraçãodaRegiãoHidrográficadoAlentejo, Évora.

Arthington,A.H.,Bunn,S.E.,Poff,N.L.,Naiman,R.,2006.Thechallengeofproviding environmentalﬂowrulestosustainriverecosystems.Ecol.Appl.16,1311–1318. Benejam,L.,Angermeier,P.,Garcia-Berthou,A.,Munne,E.,2009.Assessingeffectsof waterabstractiononﬁshassemblagesinMediterraneanstreams.Freshw.Biol. 55,628–642.

Bernardo,J.M.,Alves,M.H.,1999.Newperspectivesforecologicalﬂow determina-tioninsemi-aridregions:apreliminaryapproach.Regul.River15,221–229. Bernardo,J.M.,Ilhéu,M.,Matono,P.,Costa,A.M.,2003.Interannualvariationofﬁsh

assemblagestructureinaMediterraneanriver:implicationsofstreamﬂowon thedominanceofnativeorexoticspecies.RiverRes.Appl.19,1–12.

Bohnsack,J.A.,1983.ResiliencyofreefﬁshcommunitiesintheFloridaKeysfollowing aJanuary1977hypothermalﬁshkill.Environ.Biol.Fish.9,41–53.

Cabral,M.J.(coord.),Almeida,J.,Almeida,P.R.,Dellinger,T.R.,FerranddeAlmeida, N.,Oliveira,M.E.,Palmeirim,J.M.,Queiroz,A.I.,Rogado,L.,Santos-Reis,M.(Eds.), 2005.LivroVermelhodosVertebradosdePortugal.InstitutodaConservac¸ãoda Natureza,Lisboa.

Caetano,M.,Nunes,V.,Nunes,A.,2009.CORINELandCover2006forContinental Portugal.InstitutoGeográﬁcoPortuguês,Lisboa.

CEN,2003.WaterQuality–SamplingofFishwithElectricity.EuropeanStandard– EN14011:2003.EuropeanCommitteeforStandardization,Brussels.

Clarke,K.R.,Warwick,R.M.,1994.ChangeinMarineCommunities:AnApproachto StatisticalAnalysisandInterpretation.PlymouthMarineLaboratory,Plymouth, UK.

Clarke,K.R.,Gorley,R.N.,2006.Primerv6:UserManual/Tutorial.PlymouthMarine Laboratory,Plymouth,UK.