Patterns ofrarefishandaquaticinsectsinasouthwesternFrench rivercatchmentinrelationtosimplephysicalvariables

F. Santoul,J. Figuerola,S. Mastrorillo andR. Ce´re´ghino

Santoul, F.,Figuerola, J.,Mastrorillo,S.andCe´re´ghino, R.2005.Patterns ofrarefish and aquatic insectsinasouthwestern French river catchment inrelation to simple physicalvariables. — Ecography 28:307—314.

Mostofourcurrent understandingofrarityhascomefromstudiesofterrestrial plants andanimals,whereasfreshwater habitats remainpoorlydocumentedunderthistopic. Here weconsidered thespatial distributionpatterns ofrarity at therivercatchment scale,forfivefreshwater taxa (fish,Ephemeroptera,Plecoptera, Trichoptera and Coleoptera)insouthwestern France. Thedatawerecollectedat554and155sampling sitesforfishand aquatic insects,respectively.General Linear Modelling wasusedto assessthe influence ofsome typological variables (elevation, stream order, distance fromsource,andreachslope)onlocalnumbers ofrarespecies(restrictedrange).The relativenumbersofrarespeciespertaxavariedfrom16% (Plecoptera) to59% (Trichoptera).GLM chieflyyieldedhighlysignificantcorrelationsbetweenrarity and distance from thesourceand/or elevation foralltaxa, showingthat numbers ofrare stream species increase towards downstream areas within the stream system. The spatialpatterns inrarity forthedifferentstudytaxawererather concordant,probably asaresultofsimilarresponsestoenvironmentalconditions. Byfocusingonintegrative variables, weemphasized the influence of river typology on the rarity of aquatic animals.Areaswhichcarryrarespeciesmayconcentrateanimportantfraction ofthe regionalbiodiversity.Ifend-usersneedgeographicmodels(i.e. maps)todesignriver managementframeworks, numericalpatterningis neededtoprovidetheoretical backgrounds:bypredicting what the rarity should belikeinagivenarea, wecan provideexplicitspatialschemes thatmaybeusefultotargetfurtherresearch,andto implement managementoptions.

F.Santoul(),S.Mastrorillo andR.Ce´re´ghino,Lab.d’Ecologie des Hydrosyste`mes,UMR 5177CNRS/UPS,Univ.Paul Sabatier, 118routedeNarbonne, F-31062 Toulouse Cedex 4, France. — J.Figuerola, Estacion Biologica deDon˜ana, AvenidadeMaria Luisas/n,PabellondelPeru,E-41013Sevilla,Spain.

Species richnessandthepresenceofrarespeciesare frequentlycitedcriteriaforsiteselectionby conserva- tionists (Prendergast et al. 1993, Myers et al. 2000). Local rarity may increase the likelihood that demo- graphicand/or environmentalstochasticity willwipeout populations, andarestrictedspatialdistribution(with individualsoccurringwithhighorlowdensities)implies that populationswillprobably experience adverse con- ditionssimultaneously (Gaston 1998).However,explain- ingrarity isoften adifficulttask (Pearson etal.1983,

MagurranandHenderson2003),asraritymaybedueto random settlement, ortorandom predationoremigra- tionorimmigration, ortocompetition,orinfacttoany physicalfactor that influenceaspecies’ distribution (Legalleetal.2005). Manydefinitionsofrarityexist (Rabinowitz 1981, Gaston 1994, Kunin and Gaston

1997).Atregionalorwatershedscales‘‘rare species’’ (restrictedrange)canbe consideredasthosespecies occurring at only a fewsites (Cao et al. 2001), and

‘‘rarity’’ canthen beconsidered asthenumber ofrare

speciesin agivenarea (Lennon et al. 2004).Thus, if raritypersecannotbedirectlyexplained,patterningand understanding the geographic variations of rarity re- mains avery importantpart of conservation biology (Chuetal.2003).

Mostofourcurrentunderstandingofrarityhascome

from studies ofterrestrial plants, birds, mammals and someinsects(ThomasandMallorie1985, Bergand Tjernberg 1996),whereasfreshwater habitats areunder- represented in published studies of rarity (Chapman

1999).Riversareincreasinglyaffectedbyanthropogenic disturbancesuchasflowregulation orpollution, result- ing in modifications oftheirphysicalandchemical conditions (Ward and Stanford 1979), disruption of natural dispersal pathways (Kruk and Penczack 2003), and, subsequently, in changes within animal commu- nities(DethierandCastella2002). Inmostcases,such alterations of river habitats lead to losses of taxa

(Brittain and Saltveit 1989),and spatial discontinuities

Compin and Ce´re´ghino 2003). Specifically, we thus sought tobring out explicitmodelswhichwould allow tobetterunderstand therelationships betweenriver typology, and the distribution of rarity. To this end, generallinearmodellingwasusedtoassesstheinfluence ofeachenvironmentalvariableonlocalnumbers ofrare species. Theresultsobtained forfishandinsectsare compared, inordertoidentifyimportant areasfor conservationwithinstreamsystems.

Methods

Studyarea

The River Garonne has its source in the Maladetta Glacier(Spain),itstotal lengthis525km,anditslopes from thesoutheast to thenorth-west, whereitreaches theAtlantic oceanthrough theGironde estuary(Fig.1).

in predictable downstream gradients (Ward and

Themeanannual dischargeamounts toca545m3

s—1.

Stanford 1983).For practical managementframeworks, explicitschemessuchasdistributionpatterns ofrare(or threatened)speciesistherefore needed to identify possible conservation areas within stream systems (Park etal.2003).InFrance, forexample, theLawon Water of January 1992 strengthened a governmental actionplanforthedelineation ofnatural zonesof ecological,faunistic andfloristicinterest. Thefirst objectiveofthisplanwas toidentifyareaswhich concentrate patrimonial values, i.e. containing rare species or endangered habitats.Similar objectives are part oftheRed Listindicator systemproposed bythe IUCN(thelattercombiningvalues concerningrarityof species with their time trends). For such purposes, severalquestionswere askedtothescientificexperts. Among thesequestions, twowereofparticularinterest:

1)within agivenregional system,whichareas contain rarespecies?and2)whichenvironmentalvariablesmay explainspatialvariations innumbers ofrarespecies?To address these concerns, distribution patterns of rarity mustbederivedfromenvironmentalconditions.

Theaimofourstudywastoassesstheinfluenceofa

limitedsetofenvironmentalvariables(elevation,stream order, distance from source, and reach slope) on the spatial distributionpatterns ofrarity attherivercatch- mentscale,fordifferentfreshwatertaxa.We examined whethersimple(i.e.easily mapped)environmentalvari- ablesarecapable ofpredicting whererare speciesexist, thusallowingforeasyuseofsuccessfulfinalmodelsby environmentalmanagers and policy makers concerned withpreservationofrarespeciesinrivers.Wefocusedon fishandonfourinsectorders (Ephemeroptera, Plecop- tera, Trichoptera and Coleoptera), these five groups beingcommonlyconsideredatthespecieslevel in freshwater studies, and being particularly sensitive to the impact ofhuman activities (Oberdorff etal.2001,

TheRiverGaronnestream systemdrains anareaofca

57000km2.ComparedwithotherFrench rivers(e.g.the

Seineriverand the Rhoˆneriver),the Garonneriveris lessdisturbed byindustrial pollution. Theclimateofthe region is influenced by oceanic processes, but this declines tothesoutheastwhereitundergoestheMedi- terraneaninfluencewithdrywindsandweaker pluvio- metry.

Data collection

Environmentalvariables

Eachsamplingsite (forfish oraquaticinsects)was characterised with four environmentalvariables: eleva- tion above sealevel(ma.s.l.),distance from thesource (km), stream order, and reach slope (per thousand). Theirdistributionis showninFig.2.Thesesimple variableswerechosenbecausetheyrelatethelocation of samplingsiteswithinthestreamsystem,theyareeasyto describeusingmaps,andtheiruse insuccessfulfinal modelscouldthereforereducetheeffortandcostofdata collectionforrivermanagementapplications.

Aquaticinsects

Wesampled155unstressedsitesrangingfrom10to2500 m a.s.l. Unstressed sites were defined as sites not subjectedtoanthropogenicimpactssuchasflowregula- tion, chemical pollution, or urban runoff (indexed by the FrenchWaterAgency:B garonne.fr/ ,seealsoCompin andCe´re´ghino 2003). Samplesweretakenfrom1988 to1998.Eachsitewas sampled at two periods during a same year, i.e. in summerandwinter.Allsampleswere takenfromthe various substratum types using a standard Surber

sampler (sampling area 0.1 m2, mesh size 0.3 mm).

Ephemeroptera, Plecoptera, Trichoptera,and Coleop-

Fig.1. TheGaronnestream system,andlocation ofthe samplingsitesforfish(blackdots) andEPTC (opencircles).

tera(EPTC)species wereidentifiedbyprofessional taxonomists. Wefocusedontheseinvertebrate taxa becauseEPTCarewell-knowntobesensitivetochanges inecosystemfeatures(ReshandJackson 1993);theyare thusassumedtobegooddescriptors oftheinfluenceof spatial changesinenvironmentalconditions. 283EPTC specieswereidentified, the detailed listofspecieswas giveninCe´re´ghino etal.(2001).

Fish

Weinvestigated 554leastimpacted orfairlyunstressed sites (see above) ranging from high mountain (2500 ma.s.l.)toplainorcoastal(10ma.s.l.)areas,wherewe recorded the composition of fish species assemblages. Thesesiteswhereevenlydistributed throughoutthe Garonnestream system.Site-specificdata forfishwere collectedbetween1980and2000.Allsitesweresampled

Fig.2. Distributionofeachenvironmentalvariableunder considerationforfish(n=554)andEPTC sites(n=155).

onceby electrofishing,duringlow-flowperiods,using standardized methods(two-passremovalsampling,De Lury1947,SeberandLeCren1967).Forty fishspecies wereidentified,amongwhich25werenativespecies(i.e.

15exoticspecies).ThedetailedlistwasgiveninSantoul etal.(2004, 2005). Inthisstudy,itshouldbenoticedthat exotic fish were not considered when selecting rare species.

Selectingrarespecies

Thedistributionofeachofthe283EPTC speciesinthe Garonne streamsystemwaspreviouslystudiedby Ce´re´ghinoetal.(2001), whoidentifiedthreespatial patterns: 1)localdistribution,i.e.speciesoccurring ina restricted geographic area, 2)longitudinal zonation, i.e. species occurring in different geographic areas, but withinacharacteristic altitudinalrange,and3)regional distribution,i.e. widespreadspecies. Similaranalyses recently conducted by Santoul et al. (2004, 2005) inourstudyareashowedthatmostfish followeda longitudinal zonation pattern (as defined above), whereasfewspecieshad alocaldistribution.Wethere- fore used these works to select rare species as those specieshavingalocaldistribution,i.e.havingarestricted rangesensuCaoetal.(2001)andLennon etal.(2004). Thedetailedlistofspeciesinvolvedinthisstudyisgiven inAppendix 1.

Data analyses

Thedependentvariableusedin ourstudy(rarespecies) corresponds tocount data. Theanalysisofthistypeof data is often problematic with usual ANOVA and standard regression methods due to the violation of theassumption ofnormallydistributederrorsofthe dependentvariable. However,GeneralLinearModelling (GLM) allowsamore versatile analysis ofcorrelation thanstandard regressionmethods, becausetheerror distribution ofthedependentvariableandthefunction linkingpredictors toitcanbeadjusted tothecharacter- isticsofthe data. For analysing rare species(Crawley

1993)wefittedmodelsusingaPoissondistributionanda loglinkfunction. Riverwasincludedasarandom factor in themodeltocontrolpseudoreplicationduetothe inclusion ofmore than one point from each river. To correct thepossible effectsofunder- oroverdispersion onstatisticaltests,devianceswerescaledwiththesquare root of the ratio deviance/degree of freedom (Anon.

2000).Data wereanalysed withtheGLIMMIXmacro forSAS8.2(Anon.2000), fittingamixedeffectsGeneral LinearModel(riveras arandomvariableandenviron- mental variables asfixedvariables). Main effectswere fitted using type III tests and a stepwise backwards removalprocedurewas usedtoobtainafinalmodel containing onlysignificantfactors.

Results

Among the283insectspecies,manyspecies(44%)were rare. Their percentage of occurrence (number of sites wherethespecieswasrecorded/numberofsampledsites) wasB5%.Thenumbers ofrarespeciesperinsectorder aregiveninFig.3.Rarity wasthehighestinColeoptera (56%)andTrichoptera(59%),andthelowestinPlecop- tera (16%).Ephemeropterawereintermediate,41%of the mayfly species being rare. Among the 25 native fishspecies,20%wererare(Fig.3),andalsooccurredin B5%ofthesampling sites.Insubsequent analyses,we focusedonlocalnumbers ofrarespecies.

Thefishmodelexplained50%ofthetotalvariancein numbers ofrarespecies,asestimatedbythedevianceof the final model (124.3) and that of the null model (250.7).Twotypological variableswerenegativelycorre- latedwithlocalnumbers ofrarefish(Table1):distance from the source (p=0.01) and elevation (pB0.0001). Conversely, stream order (pB0.0001) was positively correlated with numbers ofrare species.Finally, slope wasnotsignificantlycorrelated withrarity.

FortheEphemeropteramodel,onlydistancefromthe sourcewaspositivelycorrelatedwiththenumbersofrare species(p=0.01).Slope,elevationandstreamorderwere not significantly correlated with rarity of Ephemerop- tera.(Table1). InPlecoptera, nosignificantvariables allowedtobuildamodel.For TrichopteraandColeop- tera, elevation wasnegatively correlated (p=0.01 and pB0.0001,respectively)withthenumberofrarespecies. Moreover,slopewas positivelycorrelatedwithrarityin Coleoptera(p=0.04).

Insummary,GLMshowedthatthenumbersof rare stream specieswould tend to increase towards down- streamareaswithinthestreamsystem.

Discussion

Considering rarity through the number ofrare species rather thanintermsofspeciesassemblagessensustricto is likely to fit with a broader typological approach, because the resulting patterns are not expected to be region-specific(i.e.anymodelonlyreferringtoaregion- specific listofspeciesismorepronetohavelocal acceptance). Under this scope, and with the aim to

Fig.3. Numberofrarevs commonspeciesforeachtaxa(exotic fishspecieswerenotconsidered).

Table1. Modelsanalysingthedistributionpatterns ofFish(A), Ephemeroptera(B),Plecoptera (C),Trichoptera (D)and Coleoptera(E)atthestream systemscale.Backwards models. Only variables with pB0.05 are interpreted as statistically significant. For variables not included in the models no parameter estimate is presented and the F and p values correspond to the values when added to the final models. Deviance, and dispersion (f=deviance/degree offreedom) of thefinalmodelaregivenforeachanimalgroup.

wouldcreatemostofthespatialstructurein richness patterns of EPTC. In fish, local species richness in- creasedtowards downstream areas,asaresultofdown- stream additions ofspecies(Santoul etal.2004). Therefore, common and(toalesserextent)rarespecies wouldbothcontributetothespatialstructure inrichness patterns offish.

EffectEstimate9

standarderror

A—Fish

Intercept—0.90490.811

FDFp

The spatial patterns inrarity for the different study

taxa were rather concordant.This pattern may result from: 1) random mechanisms, 2) biotic interactions among different taxa, 3)common environmentaldeter-

Slope0.24 1.5100.62

Distancesource —1.45090.5826.19 1.5110.01

Elevation—0.01490.00243.71 1.511 B0.0001

Streamorder0.76590.18816.42 1.511 B0.0001

Deviance124.31

f0.69

B—Ephemeroptera

Intercept—4.19090.549

C—Plecoptera

Intercept

Slope
Distancesource
Elevation / 0.00 / 1.86 / 0.99
0.62 / 1.86 / 0.43
0.03 / 1.86 / 0.85
Streamorder / 0.05 / 1.86 / 0.82
D—Trichoptera
Intercept / 0.24990.315
Slope / 0.33 / 1.85 / 0.57
Distancesource / 0.31 / 1.85 / 0.58
Elevation / —0.000790.0002 / 2.56 / 1.86 / 0.01
Streamorder
Deviance / 112.62 / 0.29 / 1.85 / 0.59

f0.85

E

Deviance85.86

f0.68

address the first question of areas concentratingrare species,weprovided modelsoflongitudinal gradients in numbersofrarespecies. Raritywasrelatedtothe downstream location ofsamplingsiteswithinthestream system: it primarily increased with distance from the source,ordeclinedwithelevation.Inourstudyarea,two recent works described thespatial distributionpatterns offish(Santoul etal.2004,2005) andaquatic insects (Ce´re´ghino etal.2003)speciesrichness,thus providing modelswhich mayhelptohighlightourownresults. Ce´re´ghino et al. (2003) reported that EPTC richness peakedin theintermediatesectionof thedownstream continuum oftheGaronne streamsystem,i.e.atinter- mediatestreamorder(3rd and4th)andintermediate elevations (500—1200 m). Therefore, common species

minants, or 4)spatial covariance in different environ-

mentalfactorsthatindependentlyaccountfor diversity variation in different taxa (Gaston 1996). If local systemswerecompared, itislikelythat ahigh degree of concordance could be generated through biotic factors (Paszkowski and Tonn 2000). However, at broader spatial scales such as the Garonne stream system,congruentpatternsofraritybetweenfish and aquatic insects orders are almost certainly a result of similarresponsesby differenttaxatoenvironmental conditions rather than to biotic interactions (Heino

2002). Nevertheless,thecomparison oftheresultswe obtained for the various taxonomic groups allows to refine andmoderatethelongitudinalgradientmodelof rarity. Rare fish,E,P,T,andCoccurred inallsampled rivers.However, Plecoptera weremostlyconfinedtothe upper mountainoussections ofthe stream system(see alsoCayrou etal.2000).Their habitat range wasthus lowerthan inother taxa, withregards to the environ- mentalvariableswe considered.Consequently,ifthe longitudinal pattern is broadly acceptable, Plecoptera wouldnotberelevantorganisms forpatterning rarity withinlargewatersheds.ThenumbersofrareColeoptera speciesfollowedalongitudinal gradient, butinaddition, the significant relationship with slope suggested that local conditions (e.g.erosive forces generated through the combinationofslopewith other variables such as waterdepthandcurrentvelocity) haveagreaterim- portance whenexplainingrarity inthistaxa.Fishrarity wasnegatively correlated with both distance from the sourceandelevation.Althoughthese resultsseem contradictory, theysuggesttheimportance oflocal conditions, especiallywhenstreams taketheirsourceat lowelevations (i.e.inpiedmont areas), and thus carry rarespeciesatlowdistancefromthesource.

The second step ofsuch studies, i.e.identifying the

environmental variableswhichactuallyexplainspatial variations innumbers ofrarespeciesremainsadifficult task.Indeed,species’distributionisinfluencedbyalarge number ofenvironmentalfactors, suchasthegeological historyof thearea,environmental stability(Wardand Stanford 1979),ecosystemproductivity(Lavandier and De´camps1984), habitat heterogeneityandsuitability (GormanandKar1978),andcompetition andpredation (Pianka1978).Moreover, thesefactorsoperateatseveral

spatial andtemporal scales,e.g.geologichistory affects thebiogeographyofspeciesataregional scale,whereas physicalcharacteristics ofmicrohabitats mayinfluence localdistributions (Hastieetal.2000).Byfocusingon integrativevariables(e.g. elevation,streamorder),we emphasizedtheinfluenceofrivertypologyontherarity ofaquatic animals,withamarked concentrationofrare species inlargeandhigh-ordered streams.Thisscheme remainslikelytoprovideinsightstoboth managers and ecologists,becausetheunderlyinglocalconditionswhich are associated to such global river typologies are well known (see e.g. the River Continuum Concept by Vannote et al. 1980). For example, the literature supports theideathat elevation influencesthedistribu- tion ofstream speciesthrough water temperature (Vannote and Sweeney 1980, Newbold et al. 1994), because temperaturegoverns populationdynamics through growth and fecundity (Gillet et al. 1995),by acting as a physicochemical habitat filter (sensu Poff

1997)withrespecttospeciestraits suchasmetabolism andenergeticdemands.

Most species in biological communities are rare

(Lennon etal.2004),andthisistrueofaquatic animal communities(Marchantetal.1999).Rarityissometimes omitted in ecological studies, because rare species arebelievedtocontribute littletotheinterpretationof spatialandtemporal patterns ofbiodiversity (Caoetal.

2001),and/or becausetheymayadd noisetostatistical analyses(Cayrouetal.2000). Ontheotherhand,rare speciesare of special interest to both conservationists andenvironmentalmanagers (ReyBenayasetal.1999), whereasareaswhichcarryrarespeciesmayconcentrate animportantfraction oftheregional biodiversity (this study). Ifend-users needgeographic models(i.e.maps) todesignrivermanagement frameworks, numerical patterningisneededtoprovidetheoretical backgrounds (Whittier et al. 1988): by predicting what the rarity should belikeinagivenarea, wecanprovide explicit spatial schemes that may be useful to target further research, andtoimplement managementoptions.

Acknowledgements — This work was partially funded by the

French Water Agency(Agencedel’EauAdour-Garonne).

References

Anon. 2000. SAS/STAT software: user’s guide. — SAS Inst, Cary,NC.

Berg,A˚ .and Tjernberg, M.1996.Common and rare Swedish

vertebrates — distributionandhabitat preferences. — Biodiv.

Conserv. 5:101—128.

Brittain, J.E.andSaltveit,S.J.1989.Areviewoftheeffectof

riverregulation onmayflies(Ephemeroptera).— Regulated

Rivers:Res.Manage. 3:191—204.

Cao,Y.,Larsen,D.P.andThorne, R.St-J.2001. Rarespecies in

multivariate analysis for bioassessment: some considera- tions. — J.North Am.Benthol. Soc.20:144—153.

Cayrou, J.etal.2000. Associationsspe´cifiqueschezles macroinverte´bre´s benthiquesetleurutilisationpourla classificationdescoursd’eau.Casdure´seauhydrographique Adour — Garonne (France). — Ann. Limnol. — Int. J. Limnol. 36:189—202.

Ce´re´ghino,R.,Giraudel, J.L.andCompin,A.2001. Spatial analysisofstream invertebratesdistributionintheAdour-

Garonne drainage basin (France), using Kohonen Self

Organizing Maps. — Ecol.Modell. 146:167—180.

Ce´re´ghino, R. et al. 2003.Predicting the speciesrichness of

aquatic insects in streams using a limited number of

environmentalvariables. — J.North Am.Benthol. Soc.22:

442—456.

Chapman, M. G. 1999. Are there adequate data to assess

how well theories of rarity apply to marine invertebra-

tes?— Biodiv.Conserv. 8:1295—1318.

Chu,C.,Minns,C.K.andMandrak,N.E.2003.Comparative

regional assessment of factors impacting freshwater fish

biodiversityinCanada.— Can.J.Fish.Aquat. Sci.60: 624—

634.

Compin, A. and Ce´re´ghino, R. 2003. Sensitivity of aquatic

insects speciesrichness to disturbancein the Adour-Gar-

onnestreamsystem(France). — Ecol.Indicators3:135—142. Crawley,M.J.1993.Glimforecologists. — Blackwell.

DeLury, D.B.1947.Ontheestimation ofbiological popula-

tions. — Biometrics3:145—167.

Dethier, M. and Castella, E. 2002. A ten years survey of

longitudinal zonation and temporal changes of macro-

benthic communities in the Rhoˆne River, downstream

fromLakeGeneva (Switzerland). — Ann.Limnol. — Int. J.

Limnol. 38:151—162.

Gaston,K.J.1994.Rarity. — ChapmanandHall.

Gaston, K.J.1996.Spatialcovarianceinthespeciesrichnessof

higher taxa. — In: Hochberg, M. E., Clobert, J. and

Barbault, R.(eds),Aspectsofthegenesisandmaintenance

ofbiologicaldiversity.OxfordUniv.Press,pp.221—242. Gaston,K.J.1998.Rarity asdouble jeopardy. — Nature 394:

229—230.

Gillet, C., Dubois, J. P. and Bonnet, S. 1995. Influence of

temperatureandsizeoffemaleonthetimingofspawningof

perch(Perca fluviatilis)inlakeGeneva from1984to1993.

— Environ. Biol.Fish.42:355—363.

Gorman,O. T. and Kar, J. R. 1978. Habitatstructure and

stream fishcommunities. — Ecology59:507—515.

Hastie, L. C., Boon, P. J. and Young, M. R. 2000.Physical

microhabitat requirements of freshwater pearl mussels,

Margaritiferamargaritifera(L.). — Hydrobiologia429:59—

71.

Heino,J.2002.Concordanceofspeciesrichnesspatterns among

multiple freshwater taxa: aregional perspective. — Biodiv. Conserv. 11:137—147.

Kruk, A. and Penczak, T. 2003. Impoundment impact on

populations of facultative riverine fish. — Ann. Limnol.

— Int.J.Limnol. 39:197—210.

Kunin, W.E. and Gaston,K. J.1997.The biology ofrarity.

Causes and consequences of rare-common differen-

ces.— ChapmanandHall.

Lavandier, P.andDe´camps, H.1984.Estaragne. — In:Whitton,

B.A.(ed.),EcologyofEuropeanrivers.Blackwell,pp.237—

264.

Legalle,M.etal.2005.Factors influencingthespatialdistribu-

tion patterns of the bullhead (Cottus gobio L., Pisces

Cottidae): amulti-scalestudy. — Biodiv.Conserv., inpress. Lennon, J.etal.2004.Contributionofrarityandcommonness

topatterns ofspeciesrichness. — Ecol.Lett.7:81—87. Magurran,A.E.and Henderson,P.A.2003.Explaining the

excessofrarespeciesinnatural speciesabundancedistribu- tions. — Nature 422:714—716.

Marchant,R. etal. 1999.Classification ofmacroinvertebrate

communities across drainage basins inVictoria, Australia:

consequences of sampling on a broad spatial scale for

predictivemodelling. — FreshwaterBiol.41:1—16.

Myers, N. etal. 2000.Biodiversity hotspots for conservation priorities. — Nature 403:853—858.

Newbold, J.D., Sweeney,B.W.and Vannote, R. L.1994.A

modelforseasonalsynchronyinstreammayflies.— J.North

Am.Benthol.Soc.13:3—18.

Oberdorff, T., Hugueny, B.and Vigneron, T.2001.Isassem-

blagevariabilityrelatedtoenvironmental variability?An answerforriverinefish.— Oikos93:419—428.

Park,Y.S.etal.2003.Applicationsofartificialneuralnetworks

forpatterningandpredicting aquatic insectspeciesrichness inrunning waters. — Ecol.Modell. 160:265—280.

Paszkowski, C. A. and Tonn, W. M. 2000. Community

concordance between the fish and aquatic birds of lakesinnorthern Alberta:therelativeimportance of environmental and biotic factors. — FreshwaterBiol. 43:

421—437.

Pearson,T.H.,Gray,J.S. andJohannessen,P.J.1983.Objective

selectionofsensitivespeciesindicativeofpollution-induced

change inbenthic communities. 2.Data analyses. — Mar. Ecol.Prog.Ser.12:237—255.

Pianka, E.R. 1978.Evolutionaryecology, 2nd ed. — Harper

andRow.

Poff,N.L.1997.Landscape filtersand speciestraits: towards

mechanistic understandingand prediction in stream ecol- ogy. — J.North Am.Benthol. Soc.16:391—409.

Prendergast,J.R.etal.1993.Rare species,thecoincidenceof

diversityhotspotsandconservationstrategies. — Nature365:

335—337.

Rabinowitz,D. 1981.Sevenforms ofrarity. — In: Synge,H.

(ed.), The biological aspects of rare plant conservation. Wiley,pp.205—217.

Resh, V. H. and Jackson, J. K. 1993. Rapid assessment

approachestobiomonitoring usingbenthicmacroinverte- brates. — In: Rosenberg, D. M. and Resh, V. H. (eds), Freshwaterbiomonitoringand benthic macroinvertebrates.

ChapmanandHall,pp.195—223.

ReyBenayas,J.M.etal.1999.Commonness andrarity:theory andapplicationofanewmodeltoMediterraneanmontane grasslands. — Conserv. Ecol.3:5.

Santoul, F. et al. 2004. Environmental factors influencing localfish speciesrichnessanddifferencesbetweenhydro- regionsinsouth-western France. — Int.Rev.Hydrobiol.89:

79—87.

Santoul,F.etal.2005.Spatialpatterns ofthebiologicaltraitsof

freshwater fishcommunities inS.W.France. — J.FishBiol.,

inpress.

Seber,G.A.F.andLeCren,E.D.1967.Estimatingpopulations

parametersfrom catcheslargetorelativepopulations. — J. Anim.Ecol.36:631—643.

Thomas, C. D. and Mallorie, H. C. 1985. Rarity, species

richness and conservation:butterflies ofthe Atlas Moun- tainsinMorrocco.— Biol.Conserv. 33:95—117.

Vannote, R. L. and Sweeney, B. W. 1980. Geographic

analysis of thermal equilibria: a conceptual model for evaluating the effect of natural and modified thermal

regimes on aquatic insect communities. — Am. Nat. 115:

667—695.

Vannote, R.L.etal.1980.Therivercontinuum concept. — Can.

J.Fish.Aquat. Sci.37:130—137.

Ward, J. V. and Stanford, J. A. 1979. Ecological factors

controlling stream zoobenthoswith emphasis on thermal

modification of regulated streams. — In: Ward, J. V.and

Stanford, J. A. (eds), The ecology of regulated streams.

PlenumPress,pp.35—55.

Ward, J. V. and Stanford, J. A. 1983. The intermediate

disturbancehypothesis: anexplainationforbioticdiversity

patterns in lotic ecosystems. — In: Fountain, T. D. and Bartell, S. M (eds), Dynamics of lotic ecosystems. Ann Arbor Sciences,AnnArbor, MI,pp.347—356.

Whittier,T.R.,Hughes,R.M.andLarsen,D.1988. Corre- spondence between ecoregions and spatial patterns in stream ecosystems inOregon. — Can. J.Fish. Aquat. Sci.

45:1264—1278.

SubjectEditor: AndersKlemetsen.

Appendix 1. List of the rare species involved in this study.

Fish

Barbusmeridionalis(Risso,1826) Rhodeussericeus(Pallas,1776) Salaria fluviatilis(Asso,1801)

Gasterosteusaculeatus Linnaeus, 1758

Pungitiuspungitius(Linnaeus, 1758)

Ephemeroptera

AcentrellasinaicaBogoescu,1931

Alainitesalbinatii Sartori Thomas, 1989

Alainitesmuticus(Linnaeus, 1758)

BaetispavidusGrandi,1949

Caenisluctuosa (Bu¨rmeister,1839)

CaenispusillaNavas, 1913

Choroterpespicteti(Eaton, 1871)Cloeondipterum(Linnaeus, 1761)Ecdyonurusaurantiacus (Bu¨rmeister,1839)Ecdyonurusinsignis(Eaton, 1870)Electrogenalateralis (Curtis, 1834)Ephemeralineata Eaton,1870