118,Routedenarbonne,F-31062Toulouse,France;E-Mail

..

FRÉDÉRICSANTOUL*,1,ANNESOULARD1,JORDIFIGUEROLA2,RÉGISCÉRÉGHINO1and

SYLVAINMASTRORILLO1

1Laboratoired’EcologiedesHydrosystèmes, FRE2630CNRS/UPS,UniversitéPaulSabatier;

118,RoutedeNarbonne,F-31062Toulouse,France;e-mail:

2EstacionBiologicadeDon˜ana,AvenidadeMariaLuisas/n,PabellondelPeru,

41013Sevilla,Espan˜a

EnvironmentalFactorsInfluencingLocalFishSpeciesRichnessand

DifferencesbetweenHydroregionsinSouth-WesternFrance

keywords:fish,Garonnebasin,generallinearmodelling,hydroregions, speciesrichness

Abstract

Theaimofthisstudywastoinvestigatetheinfluenceof5 typologicalvariablesonthespatialdis- tributionpatternsoffishspeciesrichnessinsouth-westernFrance,and,subsequently,toanalysediffer- encesinthenumberofspeciesoccurringin6majorhydroregionslocatedwithintheoverallstudyarea. Thedatawerecollectedat329samplingsites.GeneralLinearModellingwasusedtoassesstheinflu- enceofeachtypologicalvariableonlocalfishspeciesrichness,andtodeterminethedifferencesinlocal fishspeciesrichnessbetweenthe6hydroregions. Localspeciesrichnesswassignificantly influenced by altitude,slopeandcatchmentarea,whereasdistancefromthesourceandstreamwidthshowedno significantrelationswithlocalrichness.TheCôteauxdeGascognehydroregionhadasignificantlylower speciesrichness,whereasnosignificantdifferences occurredamongotherneighbouring hydroregions. These results werecongruent withthespatial distribution patterns offreshwater invertebrate species richnessinthearea,whichwereanalysedinpreviousstudies.Atsucharegionalscale,wesuggestthat congruentpatternsbetweenfishandinvertebratespeciesrichnessarealmostcertainlyaresultofsimi- larresponsesbydifferenttaxatoenvironmental conditions,ratherthantobioticinteractions.

1. Introduction

Bothscientificstudiesandregionalsurveysofstreamecosystems haveprovidedlarge volumesofsite-specific data(e.g.,WHITTIERetal.,1988;CAYROUetal.,2000),fromwhich localandregionalspatialpatternsof biologicalcommunitiescanbederived(HAWKINS and NORRIS,2000;DETHIERandCASTELLA,2002).Severalauthorshavethusemphasizedthe importanceofgeographicdifferencesinbioticandabioticcharacteristics ofstreams(CULP and DAVIES,1982; ORENDT,2003), so that classification has become an integral part of effortstostudy,monitorandmanageecosystemsataregionalscale(WARREN,1979;OMER- NIK,1987;TATEandHEINY,1995;SANDINandJOHNSON,2000).Theidentificationofareas containingriverswithsimilarbiologicalandenvironmental characteristics mayhelpinsel- ectingreference riversforimpactassessments, orinselecting monitoring sitesfromwhich resultsmaybeextrapolated regionally(RHOMetal.,1987).Byknowingwhattheecosystem shouldbelikeinagiven geographiczone, ecologistscanthusdeterminethedegreetowhich humanactivityhasalteredit(HAWKINSetal.,2000).Suchapproachestoriverbioassessment wererecentlydevelopedinEurope(e.g.,RiverInvertebrate PredictionandClassification System),Australia(AustralianRiverAssessmentSystem),andCanada(BenthicAssessment ofSediment)(reviewedinWRIGHTetal.,2000).

* Corresponding author

Thepartitioning ofalargeareaintorelativelyhomogeneous regionsalsoprovidesspatial orderforassessingcomplexpatternsofecosystemvariation(HUANGandFERNG,1990).Con- cordances between the distribution of organisms or communities and such geographic regionstestourunderstandingof howorganismsrespondto combinationsof bioticandabio- ticfactors(OSWOODetal.,2000).Catchments arecommonly adoptedwhendefiningbound- ariesandpartitioningvarianceamongaquaticcommunities(e.g.JOHNSON,2000). Among aquatic organisms, stream fish communities offer excellent opportunities to examine the relativeinfluenceoflocalandregionalfactorsonspeciesrichness.Understanding thepat- ternsofgeographicvariationinthestructureoffishassemblagesis thereforeofcrucial importancetodevelopacomprehensive assessmentofstreamconditions.Thenumberoffish species occurring inagiven area isrelated tonumerous local factors (OBERDORFF etal.,

1993;HOLCÍK,2003;KRUKandPENCZAK,2003),andspeciesrichnesspatternsareimportant biodiversityindicators(GASTON,1996).Specifically,intheRiverGaronnedrainagebasin (S.W.France), MASTRORILLOetal.(1996)observedthattherewerenaturaldifferencesin thespatialdistribution offishspeciesrichness,butdidnotassesstheinfluence ofenviron- mentalfactorsontheobservedpatterns.

Fishdistribution patternsmightnotcorrespond exactlytothosesuggestedbyhydroregions (VANSICKLEandHUGHES,2000).However,LARSENetal.(1986)foundthatthecorrespondence isencouraging. Therefore,theobjectivesofthisstudyweretoinvestigatetheinfluenceoffive environmental variablesonlocalfishspeciesrichnessinsouth-westernFrance,and,toanalyse differencesinthenumberofspeciesoccurringin6neighbouring hydroregionswithintheover- allstudyarea.Wethusattempted toassess thepotentialoffishspeciesrichnesstodifferentiate hydroregions. Theresultswereanalysed withreference togeographic differences inenviron- mentalconditions,andtopreviousstudiesinthesameriverbasin(CEREGHINOetal.,2001,2003) whichyieldedregionalisations ofstreaminvertebrates communities inthestreamsystem.

2. Methods

2.1. StudyArea

TheRiverGaronnehasitssourceintheMaladettaGlacier(Spain),anditslopesfromthesoutheast tothenorth-west,whereitreachestheAtlanticoceanthroughtheGirondeestuary.TheRiverGaronne drainsanareaofabout57000km2 anditstotallengthis525km.Themeanannualdischargeamounts toabout545m3 s–1.ComparedwithotherFrenchrivers(e.g.theSeineriverandtheRhôneriver),the Garonneriverislessdisturbedbyindustrialpollution.However,itsnaturalflowhasbeenmodifiedby thepresenceofseveraldams,resultinginanimalandvegetalcommunityfragmentationwithintheriver channelandthealluvialfloodplain(DÉCAMPS etal.,1988).Theclimateoftheregionisinfluencedby oceanicprocesses,butthislessenstothesoutheastwhereitundergoestheMediterraneaninfluencewith drywindsandweakerpluviometry.

2.2. DataCollection

Tobuildthemodel,weuseddatafrom329fishsamplingsitesevenlydistributedthroughoutthepied- montzoneoftheGaronneriverbasin,whichcorresponds tothegraylingzone(Thymallus thymallus) andthecommonbarbelzone(Barbusbarbus)accordingtoHuet’szonation(1949).Wefocusedonthe piedmontzoneinordertopreventdifferencesduetonaturallongitudinalgradientsinfishspeciesrich- ness (MASTRORILLO, 1997). Data for local fish species richness were collected between 1980 and

2000.All sites were sampled once by electro fishing, during low-flow periods, using standardized methods(two-passremovalsampling,DELURY1947;SEBERandLECREN 1967).

Eachsitewascharacterisedwithsix environmentalvariablesandonebiologicalvariable.Theen- vironmental variables werechosen torelate thelocation ofsampling siteswithin theoverall stream system:altitude(a.s.l.m),distancefromthesource(km),catchment area(km2)andhydroregion data wereobtainedfromaGeographicInformationSystem,whereasslope(perthousand)andstreamwidth (m)wheremeasuredonthefield.Thebiologicalvariablewasthelocalfishspeciesrichness.

2.2.1. DefiningHydroregions

Understanding theroleofregionalinfluencesincommunityorganization requirescarefuldefinition ofregionalboundaries.Theclassificationintohydroregionsconsistedofsixsub-drainagebasins:CGA (CôteauxdeGascogne,numberofsamplingsites:n=45),GAR(Garonneriver,n=58),ARI(Ariège river, n=42),TAR (Tarn river, n=56),AVE (Aveyron river, n=70)andLOT(Lotriver, n=58). Hydroregions (Fig.1) were defined as the whole drainage basin area of the 6largest rivers inthe area (indexed by the French Water Agency: see also SOULARD,

2000).

2.2.2.DataAnalyses

Thedependentvariableusedinourstudy(speciesrichness)correspondstocountdata.Theanalysis ofthistypeofdataisoftenproblematic withusualANOVAandstandardregression methodsdueto the violation of the assumption of normally distributed errors of the dependent variable. However, GeneralLinearModelling(GLM)allowsamoreversatileanalysisofcorrelationthanstandardregres- sionmethods,becausetheerrordistributionofthedependentvariableandthefunctionlinkingpredic- torstoitcanbeadjustedtothecharacteristics ofthedata.Foranalysing speciesrichness(CRAWLEY,

1993)wefittedmodelsusingaPoissondistribution andaloglinkfunction.Riverwasincludedasa

randomfactorinthemodeltocontrolpseudoreplication duetotheinclusion ofmorethanonepoint fromeachriver.Datawereanalysed withtheGLIMMIX macroforSAS8.2(SASINSTITUTE 2000), fittingamixedeffectsGeneralLinearModel(riverasarandomvariableandenvironmental variables asfixedvariables).MaineffectswerefittedusingtypeIIItestsandastepwisebackwardsremovalpro- cedurewasusedtoobtainafinalmodelcontainingonlysignificantfactors.

Figure1. Mapofriverbasinincluding thesixhydroregions andsampling sitesonthemainrivers,

3-lettercodesindicatehydroregion(seetext).

3. Results

Themodelexplained87.2%ofthetotalvarianceinspeciesrichness(cf.Table1),asesti- mated by the deviation of the final model (F=200.28) and that of the null model (F=1561.27).Fourvariablessignificantlyinfluencedlocalspeciesrichness(Table2and3): altitude(p0.0001),slope(p0.0001),catchmentsarea(p0.008),and,toalesserdegree, hydroregion(p=0.05).Conversely,streamwidthand distancefromthesourcedid not influ- encethelocalspeciesrichness. Whenhydroregions werecompared, Côteaux deGascogne had significantly lower cumulative species richness than Garonne (p0.003) and Tarn (p0.004)hydroregions.Otherpairwisecontrastswerenotsignificant(Fig.2).Thus,local speciesrichness washigheratsiteswithlowaltitudeandweakslopesandinlarger catchments. Nevertheless, ifinthecaseofCôteaux deGascogne, thehydroregion seemed tosignificantlyinfluencelocalspeciesrichness,otherhydroregions showedaweakinfluen- ceonspeciesrichness.

4. Discussion

Ourmodelsdidshowthatafewpertinentvariablescouldexplainspatialvariationsin localfishrichness, andthat,subsequently, theuseofsimplevariables inasuccessful final modelcouldreducetheeffortandcostofdatacollectionforbasicinvestigationsand/or watermanagementapplications.Eventhoughourstudysiteswerelocatedwithina given biogeographic area(i.e.,thepiedmont–ortransition–zone),altituderemainedamajoreco- logicalfactor.Forexample,MORINandNAIMAN(1990)showedthatthenumberoffishspe- ciesinstreamsofnorthernQuebecwasnegativelycorrelatedwith streamaltitude,whichfits withourresults. Theliterature supports theideathataltitude influences localfishspecies richnessthroughwatertemperature(e.g.HUET, 1949).It is indeedcommonlyrecognizedthat

Table1. Listoffishspeciescollected.

Table2. Descriptive statisticsmean ±S.E.(range),foreachenvironmental variableandspecies richnessforeachhydroregions, 3-lettercodesindicatehydroregions (seetext).

HydroregionAltitudeCatchmentareaDfromthesourceSlopeWidthSpeciesrichness

CGA 152.8 ±9.9(34–285) 350.8 ±106.4(2–2810) 41.3 ±5.9(2.12–135) 2.67 ±0.22(0.50–6.25) 10.1 ±1.5(1.4–40.0) 8.6 ±0.5(4–17) GAR 251.2 ±23.0(8–715) 7481.1 ±1511.0(2–54051) 132.2 ±14.3(2.4–446.4) 1.99 ±0.24(0.50–7.14) 47.2 ±6.5(0.9–200.0) 15.6 ±1.3(2–32) ARI 291.6 ±16.7(120–545) 1052.0 ±142.1(6–4050) 57.0 ±5.8(6.0–156.7) 2.13 ±0.31(0.50–6.66) 29.7 ±3.0(1.7–75.0) 9.8 ±0.8(2–17) TAR 402.5 ±31.8(72–810) 1158.1 ±216.4(8–5155) 89.3 ±11.5(2–292) 2.79 ±0.30(0.50–7.14) 18.9 ±2.1(1–75) 13.6 ±0.9(1–26) AVE 284.2 ±23.9(60–836) 2327.8 ±359.9(3–15400) 117.7 ±11.5(3–364) 2.31 ±0.23(0.50–6.66) 44.2 ±4.8(1–200) 13.5 ±1.0(1–26) LOT 209.2 ±25.9(13–1000) 5023.1 ±468.1(60–10712) 205.6 ±17.1(20–480) 1.43 ±0.19(0.38–7.00) 56.9 ±4.8(2.25–125.00) 18.1 ±0.8(2–25)

Table 3. Cumulativespeciesrichnessoffishtesting hydroregions.Backwardsmodel. Only variableswithp0.05areinterpreted asstatistically significant. Forvariablesnotincluded inthemodelnoparameter estimate ispresented andtheFandpvaluescorrespond tothe valueswhenaddedtothefinalmodel.

EffectEstimate± standarderrorDFFP

14

12

10

8CGA

GAR

ARI

TAR

AVE

LOT

6

4

2

0

Figure 2. Summary ofGLMmodel forpredicting species richness usinghydroregion andenviron- mentalvariables.Meansarecorrectedfortheeffectsoffixedandrandomvariablesincludedinthefinal backwardsmodel.Variancebarsindicatestandarderrors(S.E.).

temperature may influence local fish species richness bygoverning population dynamics throughgrowthandfecundity(GILLETetal., 1995),i.e.byactingasaphysicochemicalhabi- tatfilter(sensuPOFF,1997)withrespecttospeciestraitssuchasmetabolism andenergetic demands.Slopeisalsoakeyvariableinstreamecosystems, asitstronglycontributestothe erosiveforceactingonsubstrateandcausingbedscourinagivenarea(COBBandFLANNA- GAN,1990; COBB etal., 1992), and subsequently impacts stream animals. According to REYJOLetal.(2003),weakerslopeswithinthepiedmontzoneinourstudyareaareasso- ciatedtohigherhabitatheterogeneity, thusfavouring higherspeciesrichness(e.g.,coexist- enceofbothsalmonidandcyprinidfish).

Nevertheless, ourresultstendtodemonstratethatfishspeciesrichnessdoesnothavethe potentialtodifferentiatehydroregions. Indeed,althoughtwocommunitiesfromdifferent hydroregions coulddifferintermsofspeciescomposition, theycouldhavesimilarnumber offishspecies.Althoughtheprinciplethatacatchment’s featurescontrolaquaticlifein streamsisusuallywellaccepted(ROSS,1963;HYNES, 1975),theliteratureoftenprovides contradictoryresults,probablydependingonboththefaunaandtheenvironmental features ofthestudyarea.Forexample,HUGUENY(1989)foundthatfishspeciesrichnessofariver wasrelatedtothesurfaceareaofitscatchmentinWestAfrica,whileHUGUESetal.(1987) suggestedthatfishassemblagesreflectcatchmentsandphysiographicprovinces.Inourstudy area,tworecentworksusedmacroinvertebrates toproducestreamclassifications.These classifications werebasedonassemblage compositions (CÉRÉGHINOetal.,2001)orspecies richness(CÉRÉGHINOetal.,2003),andprovidedtwodifferentbiogeographicmodelswhich mayhelptohighlightourownresults.CÉRÉGHINOetal.(2001)clearlysegregatedthreegeo- logicalareasonthebasisofassemblage compositions (PyreneesMountains inthesouthern partofthesystem,MassifCentralMountains intheeasternpart,andalluvialplaininnorth westernareas)thatencompasses thesixhydroregions weconsidered.Specifically, Côteaux deGascogne (whichshowedthelowestfishspeciesrichness) islocatedwithinthealluvial plain,i.e.,wheremacroinvertebratespeciesrichnesswasamongthe lowest.CÉRÉGHINOetal. (2003) thenshowed thatelevation andstream ordercontributed mosttothemodel inthe Garonnestreamsystem,whereaswatertemperature anddistancefromthesourcehada weakercontribution, i.e.,localinvertebrate speciesrichnesswasthelowestwhensitescom- binedlowelevation(<500m)withlowstreamorder(1and2).Concordantspatialpatterns inspeciesrichnessamongdifferenttaxamayresultfrom:1)randommechanisms,2)bio- ticinteractionsamongdifferenttaxa,3) commonenvironmentaldeterminants,or 4) spatial covarianceindifferentenvironmental factorsthatindependently accountfordiversityvaria- tionindifferenttaxa(GASTON,1996).Iflocalsystemsarecompared,itislikelythatahigh degree ofconcordance could begenerated through biotic factors (PASZKOWSKIandTONN,

2000).However,atbroaderspatialscalessuchashydroregions withintheGaronnestream system,congruentpatternsbetweenfishandinvertebratespeciesrichnessarealmostcer- tainlyaresultofsimilarresponsesbydifferenttaxatoenvironmental conditionsratherthan tobioticinteractions (HEINO,2002).Thisindicatesstreamscharacterizedbyloworderand lowelevation,suchasthoseintheCôteauxdeGascognehydroregion, arelikelytosupport lowerbiodiversity.

Duringthelastdecades, therehavebeenmanyattempts toproduce streamclassifications basedonaquaticcommunity features(e.g.,ILLIESandBOTOSENEANU, 1963;GIBONand STATZNER,1985;OMERNIK,1987;WHITTIERetal.,1988;TATE andHEINY,1995).Never- theless, fewtypological systems havehadmorethanlocalacceptance (PENNAK,1971). In thiscontext,ourresults(i.e.,altitudeandslopewerethemostsignificantvariableswhen explaininglocalfishspeciesrichness)supporttheideathatthemostuniversalclassification systemsbasically remainthosewhicharebasedonstreamsizeandlocationwithinastream system. Two reference works fundamentally highlighted this approach, i.e., the study by ILLIES(1961),whichrecognizedeightzoneswithinasingledrainagesystem,rangingfrom zoneI(springsor“eucrenon”)tozoneVIII(brackishzoneor“hypopotamon”),andtheRiver

Continuum Concept(VANNOTEetal.,1980),whichhadakeyinfluence onstreamecology duringthelast20years(RESHandKOBZINA,2003).Despitegeographicdifferencesinspe- ciesassemblages, suchmodelshavethebroadestcapabilityofdescribingspatialpatternsof communitydiversityintheRiverGaronnebasin.

5. Acknowledgements

Thisworkwaspartially funded bytheAgence del’EauAdour-Garonne andtheGis-Ecobag. We thankARTHUR COMPIN forhistechnicalassistance.

6. References

CAYROU,J.,A.COMPIN,N.GIANIandR.CÉRÉGHINO,2000:Associationsspécifiqueschezlesmacroin- vertébrésbenthiqueset leurutilisationpourla classificationdescoursd’eau.Casduréseauhydro- graphiqueAdour–Garonne(France).–Ann.Limnol.–Int.J.Lim.36:189–202.

CÉRÉGHINO,R.,J.L.GIRAUDEL andA.COMPIN,2001:Spatialanalysisofstreaminvertebratesdistribu- tionintheAdour-Garonne drainagebasin(France), usingKohonen SelfOrganizing Maps.–Ecol. Model.146:167–180.

CÉRÉGHINO,R.,Y.S.PARK,A.COMPINandS.LEK,2003:Predictingthespeciesrichnessofaquaticin- sectsinstreamsusingarestrictednumberofenvironmental variables.–J.N.Am.Benthol.Soc.22:

442–456.

COBB,D.G.andJ.F.FLANNAGAN,1990:TrichopteraandsubstratestabilityintheOchreRiver,Mani- toba.–Hydrobiologia 206:29–38.

COBB,D.G.,T.D.GALLOWAY andJ.F.FLANNAGAN,1992:Effectsofdischargeandsubstratestability ondensityandspeciescompositionofstreaminsects.–Can.J.Fish.Aquat.Sci.49:1788–1795.

CRAWLEY,M.J.,1993:Glimforecologists.Oxford:Blackwellscience.

CULP, J.M. and R.W. DAVIES, 1982: Analysis of longitudinal zonation and the river continuum concept of the Oldman – South Saskatchewan River system. – Can. J. Fish. Aquat. Sci. 39:

1258–1266.

DÉCAMPS,H.,M.FORTUNÉ,F.GAZELLEandG.PAUTOU,1988:Historicalinfluenceofmanontheripa- riandynamicsofafluviallandscape.–LandscapeEcol.1:163–173.

DELURY,D.B.,1947:Ontheestimationofbiologicalpopulations.–Biometrics3:145–167. DETHIER,M.andE.CASTELLA,2002:Atenyearssurveyoflongitudinal zonationandtemporalchan-

gesofmacrobenthiccommunitiesintheRhôneRiver,downstreamfromLakeGeneva(Switzerland).

–Ann.Limnol.–Int.J.Lim.38:151–162.

GASTON,K.J.,1996:Spatialcovariance inthespeciesrichnessofhighertaxa.Pages221–242.–In: M.E.HOCHBERG,J.CLOBERT andR.BARBAULT (eds.).Aspects ofthegenesisandmaintenance of biologicaldiversity.OxfordUniversityPress,Oxford,UK.

GIBON, F.M.and B.STATZNER, 1985: Longitudinal zonation oflotic insects intheBandama River

System(IvoryCoast).–Hydrobiologia 122:61–65.

GILLET,C.,J.P.DUBOISandS.BONNET,1995:Influenceoftemperatureandsizeoffemaleonthetiming ofspawningofperch(Percafluviatilis)inlakeGenevafrom1984to1993.–Environ.Biol.Fish.42:

355–363.

HAWKINS,C.P.andR.H.NORRIS,2000:Performanceofdifferentlandscapeclassifications foraquatic bioassessments: introduction.–J.N.Am.Benthol.Soc.19:367–369.

HAWKINS, C.P., R.H. NORRIS, J.N. HOGUE and J.W. FEMINELLA, 2000: Development and evalua- tion of predictive models for measuring the biological integrity of streams. – Ecol. Appl. 10:

1456–1477.

HEINO,J.,2002:Concordanceofspeciesrichnesspatternsamongmultiplefreshwatertaxa:a regional perspective.–Biodiv.Conserv.11:137–147.

HOLCÍK,J.,2003:ChangesinthefishfaunaandfisheriesintheSlovaksectionoftheDanubeRiver:a review.–Ann.Limnol.–IntJ.Lim39:177–195.

HUANG,S.L.andJ.J.FERNG,1990:Appliedlandclassificationforsurfacewaterqualitymanagement: I.Watershedclassification. –J.Environ.Manage.31:107–126.

HUET,M.,1949:Aperçudesrelationsentrelapenteetlespopulationspiscicolesdeseauxcourantes.– Rev.SuisseHydrol.11:332–351.

HUGHES, R.M., E.REXSTAD and C.E.BOND, 1987: The relathionship ofAquatic Ecoregions, River BasinsandPhysiographicProvincestotheIchtyogeographicRegionsofOregon.–Copeia2:423–432. HUGUENY,B.,1989:WestAfricanriversasbiogeographic islands:speciesrichnessoffishcommuni-

ties.–Oecologia79:236–243.

HYNES,H.B.N.,1975:Thestreamanditsvalley.–Verh.Internat.Verein.Limnol.19:1–15.

ILLIES,J.,1961:Versucheinerallgemeinen biozönotischen GliederungderFliessgewässer. –Int.Rev. Ges.Hydrobiol.46:205–213.

JOHNSON,R.K.,2000:Spatialcongruencebetweenecoregionsandlittoralmacroinvertebrate communi- ties.–J.N.Am.Benthol.Soc.19:475–486.

KRUK,A.andT.PENCZAK,2003:Impoundment impactonpopulations offacultative riverine fish.– Ann.Limnol.–Int.J.Lim.39:197–210.

LARSEN,D.P.,J.M.OMERNIK,R.M.HUGUES,C.M.ROHM,T.R.WHITTIER andA.J.KINNEY,1986: Thecorrespondencebetweenspatialpatternsin fishassemblagesin OhiostreamsandAquaticeco- regions.–Environ.Manage.10:815–828.

MASTRORILLO,S.,F.DAUBA andA.BELAUD,1996:Utilisationdesmicrohabitats parlevairon,legou- jonetlalochefranchedanstriosrivièresdusud-ouestdelaFrance.–Ann.Limnol.–Int.J.Lim.

32:185–195.

MASTRORILLO,S.,1997:Profilsécologiquesduvairon,delalochefrancheetdugoujonàl’échelledu microhabitatdanstroisrivièrespyrénéennes. PhDDissertation,UniversityofToulouse,106p.

MORIN, R. and R.J. NAIMAN, 1990: The relation of stream order to fish community dynamics in borealforestwatersheds.–Polsk.Arch.Hydrobiol.37:135–150.

OBERDORFF,T.,E.GUILBERT andJ.C.LUCCHETTA,1993:PatternsoffishspeciesrichnessintheSeine

Riverbasin,France.–Hydrobiologia 259:157–167.

OMERNIK,J.M.,1987:Ecoregions oftheconterminous UnitedStates.–Ann.Assoc.Am.Geogr.77:

118–125.

ORENDT,C.,2003:Aclassificationofsemi-naturalnorthernprealpineriverstretchesbasedonchirono- midcommunities. –Ann.Limnol.–Int.J.Lim.39:219–237.

OSWOOD, M.W., J.B. REYNOLDS, J.G. IRONS and A.M. MILNER, 2000: Distributions of freswater fishesinecoregionsandhydroregions ofAlaska.–J.N.Am.Benthol.Soc.19:405–418.

PASZKOWSKI,C.A.andW.M.TONN,2000:Communityconcordancebetweenthefishandaquaticbirds oflakesinnorthernAlberta:therelativeimportanceofenvironmentalandbioticfactors.–Freshwat. Biol.43:421–437.

PENNAK,R.W.,1971:Towardaclassification oflotichabitat.–Hydrobiologia 38:321–334.

POFF,N.L.,1997:Landscapefiltersandspeciestraits:towardsmechanisticunderstanding andpredic- tioninstreamecology.–J.N.Am.Benthol.Soc.16:391–409.

RESH,V.H.andN.G.KOBZINA,2003:Aperspectiveonthekeycitationsinfreshwaterbenthicscien- ce,andthestudiesthatinfluencedthem.–J.N.Am.Benthol.Soc.22:341–351.

REYJOL, Y., A. COMPIN, A. ALONSO-IBARRA and P. LIM, 2003: Longitudinal diversity patterns in streams:comparinginvertebratesandfishcommunities. –Arch.Hydrobiol.157:525–533.

ROHM,C.M.,J.W.GIESE andC.C.BENNETT,1987:Evaluationofanaquaticecoregionclassification ofstreamsinArkansas.–J.Freshwat.Ecol.4:127–140.

ROSS,H.H.,1963:Streamcommunitiesandterrestrialbiomes.–Arch.Hydrobiol.59:235–242. SANDIN, L. and R.K. JOHNSON, 2000: Ecoregions and benthic macroinvertebrate assemblages of

Swedishstreams.–J.N.Am.Benthol.Soc.19:462–474.

SASINSTITUT,2000:SAS-STAST software:Cary,NorthCarolina.

SEBER,G.A.F.andE.D.LECREN,1967:Estimatingpopulationsparametersfromcatcheslargetorela- tivepopulations.–J.Anim.Ecol.36:631–643.

SOULARD, A.,2000: Structure etrépartition descommunautés pisciaires àl’échelle duBassin dela

Garonne.Essaid’unebiotypologie. DEA,UniversityofToulouse,33p.

TATE,C.M.andJ.S.HEINY,1995:Theordination ofbenthicinvertebrate communities intheSouth

PlatteRiverBasininrelationtoenvironmental factors.–Freshwat.Biol.33:439–454.

VANNOTE,R.L.,G.W.MINSHALL,K.W.CUMMINS,J.R.SEDELL andC.E.CUSHING,1980:Theriver continuumconcept.–Can.J.Fish.Aquat.Sci.37:130–137.

VAN SICKLE,J.andR.M.HUGHES,2000:Classification strengthsofecoregions, catchments, andgeo- graphicclustersforaquaticvertebratesinOregon.–J.N.Am.Benthol.Soc.19:370–384.

WARREN,C.E.,1979:TowardClassificationandRationaleforWatershedManagementandStreamPro- tection.EPA600/3–79–059.U.S.Environmental ProtectionAgency,Corvallis,OR.

WHITTIER,T.R.,R.M.HUGHES andD.LARSEN,1988.Correspondence betweenecoregionsandspatial patternsinstreamecosystemsinOregon.–Can.J.Fish.Aquat.Sci.45:1264–1278.

WRIGHT,J.F.,D.W.SUTCLIFFEandM.T.FURSE(eds.),2000:Assessingthebiologicalqualityoffresh waters:RIVPACSandothertechniques.FreshwaterBiologicalAssociation,Ambleside,UK.

ManuscriptreceivedJuly31st,2003;revisedOctober13th,2003;acceptedOctober22nd,2003