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
Original
Articles
Defining
quantitative
stream
disturbance
gradients
and
the
additive
role
of
habitat
variation
to
explain
macroinvertebrate
taxa
richness
Raphael
Ligeiro
a,∗,
Robert
M.
Hughes
b,
Philip
R.
Kaufmann
c,
Diego
R.
Macedo
a,d,
Kele
R.
Firmiano
a,
Wander
R.
Ferreira
a,
Déborah
Oliveira
a,
Adriano
S.
Melo
e,
Marcos
Callisto
aaLaboratóriodeEcologiadeBentos,DepartamentodeBiologiaGeral,InstitutodeCiênciasBiológicas,UniversidadeFederaldeMinasGerais,Av.AntônioCarlos6627,CP486,CEP
30161-970BeloHorizonte,MinasGerais,Brazil
bDepartmentofFisheries&Wildlife,OregonStateUniversityandAmnisOpesInstitute,NashHall,97331-4501Corvallis,OR,USA
cU.S.EnvironmentalProtectionAgency,OfficeofResearch&Development,NationalHealth&EnvironmentalEffectsLab.,WesternEcologyDivision,200SW35Street,97333Corvallis,
OR,USA
dInstitutoBrasileirodeGeografiaeEstatística,RuaOliveira523,CEP30310-150BeloHorizonte,MinasGerais,Brazil
eDepartamentodeEcologia,InstitutodeCiênciasBiológicas,UniversidadeFederaldeGoiás,CampusII,SamambaiaCP131,CEP74001-970Goiânia,Goiás,Brazil
a
r
t
i
c
l
e
i
n
f
o
Articlehistory:Received14June2012
Receivedinrevisedform6September2012 Accepted9September2012
Keywords:
Referenceconditionapproach Localdisturbances Catchmentdisturbances Disturbanceindices Streamhabitats EPTassemblages
a
b
s
t
r
a
c
t
Moststudiesdealingwiththeuseofecologicalindicatorsandotherappliedecologicalresearchrelyon somedefinitionorconceptofwhatconstitutesleast-,intermediate-andmost-disturbedcondition. Cur-rently,mostrigorousmethodologiesdesignedtodefinethoseconditionsaresuitedtolargespatialextents (nations,ecoregions)andmanysites(hundredstothousands).Theobjectiveofthisstudywastodescribe amethodologytoquantitativelydefineadisturbancegradientfor40sitesineachoftwosmall south-easternBrazilriverbasins.Theassessmentofanthropogenicdisturbanceexperiencedbyeachsitewas basedsolelyonmeasurementsstrictlyrelatedtotheintensityandextentofanthropogenicpressures.We calculatedtwoindices:oneconcernedsite-scalepressuresandtheothercatchment-scalepressures.We combinedthosetwoindicesintoasingleintegrateddisturbanceindex(IDI)becausedisturbances oper-atingatbothscalesaffectstreambiota.Thelocal-andcatchment-scaledisturbanceindiceswereweakly correlatedinthetwobasins(r=0.21and0.35)andbothsignificantly(p<0.05)reducedsiteEPT(insect ordersEphemeroptera,Plecoptera,Trichoptera)richness.TheIDIalsoperformedwellinexplainingEPT richnessinthebasinthatpresentedthestrongerdisturbancegradient(R2=0.39,p<0.001).Natural habi-tatvariabilitywasassessedasasecondsourceofvariationinEPTrichness.Streamsizeandmicrohabitats werethekeyhabitatcharacteristicsnotrelatedtodisturbancesthatenhancedtheexplanationofEPT richnessoverthatattributedtotheIDI.InbothbasinstheIDIplushabitatmetricstogetherexplained around50%ofEPTrichnessvariation.Inthebasinwiththeweakerdisturbancegradient,naturalhabitat explainedmorevariationinEPTrichnessthandidtheIDI,aresultthathasimplicationsforbiomonitoring studies.Weconcludethatquantitativelydefineddisturbancegradientsofferareliableandcomprehensive characterizationofanthropogenicpressurethatintegratesdatafromdifferentspatialscales.
©2012ElsevierLtd.Allrightsreserved.
1. Introduction
Thedevelopmentandmaintenanceofhumansocietiesrelyon
the conservationof freshwater resources and of theecological
Abbreviations:LDI,localdisturbanceindex;CDI,catchmentdisturbanceindex; IDI,integrateddisturbanceindex.
∗ Correspondingauthor.Tel.:+553134094597;fax:+553134092569. E-mailaddresses:ligeirobio@gmail.com(R.Ligeiro),
Hughes.Bob@epamail.epa.gov(R.M.Hughes),Kaufmann.Phil@epamail.epa.gov (P.R.Kaufmann),rodriguesmacedo@gmail.com(D.R.Macedo),
kelerocha@gmail.com(K.R.Firmiano),ferreirawr@gmail.com(W.R.Ferreira), deborah.ufmg@gmail.com(D.Oliveira),asm.adrimelo@gmail.com(A.S.Melo), mcallisto13@gmail.com(M.Callisto).
servicesthat streams and riversprovide (Karr,1999).
Monitor-ingthe“ecosystem health”ofstreams(sensuNorrisandThoms,
1999)isafundamentalstepforconsciousandeffective
manage-mentofcatchments(Boulton,1999).Currently,biomonitoringis
consideredoneofthemostefficientwaystoassessstream
con-dition(Marchantetal.,2006).Macroinvertebrateassemblagesare
responsivetoenvironmentalconditionandthusintegrate
phys-ical,chemicalandbiologicalaspectsofecosystems.Accordingly,
theyareconsideredgoodbiologicalindicatorsofstream
ecolog-icalcondition(KarrandChu, 1999;Bonada etal.,2006;Hughes
andPeck,2008)andareextensivelyusedinmultimetricindices
(MMIs)forsuchpurposes(Reynoldsonetal.,1997;Barbouretal.,
1999;Klemmetal.,2003;Heringetal.,2006;Whittieretal.,2007a).
TheEPTassemblages(insectordersEphemeroptera,Plecopteraand
1470-160X/$–seefrontmatter©2012ElsevierLtd.Allrightsreserved. http://dx.doi.org/10.1016/j.ecolind.2012.09.004
Trichoptera),particularly,haveproveneffectiveecological
indica-torsofhumandisturbances(RosenbergandResh,1993;Stoddard
etal.,2008).
Agoalofmanybiomonitoringapproachesistoreporthowtest
sitesdeviatefromthe“undisturbed”(natural)conditioninterms
ofthestructureand/orcompositionoftheassemblagesthey
sup-port.Thisistypicallyaccomplishedbydesignating“referencesites”,
thatis,sites minimallyaffectedby humanactivitiesand whose
biological,physicalandchemicalfeaturesserveasreference
con-ditionfornaturallevelsofpatternsandprocesses(Hughesetal.,
1986;Stoddardetal.,2006;Hawkinsetal.,2010).Asetof
refer-encesitesshouldbespecificforaparticulartypology(e.g.,altitude,
streamsize,andpredominantsubstrate)andgeographicdomain
(biomeandecoregion)becausetheseareimportantnaturaldrivers
ofstreamcharacteristics,includingtheirbiota(Hughesetal.,1986,
1990;Gerritsenetal.,2000;Waiteetal.,2000;Sánchez-Montoya
etal.,2007).Thisframeworkhasbeenestablishedasthe“reference
conditionapproach”(RCA)(Baileyetal.,2004).Inmostcasesitisnot
practicaltoseeksitesthathavetrulyundisturbed/minimally
dis-turbedconditionsbecause(1)humanmodificationsarewidespread
inmostlandscapesworldwide, and (2) manyplaceshave been
modifiedforhundreds(oreventhousands)ofyears(Stoddardetal.,
2006;Whittieretal.,2007b; Herlihyet al.,2008).Instead,sites
inleast-disturbedcondition,i.e.,thebestsetofsitesavailablein
acontinuousgradientofdisturbance,aretypicallyusedto
repre-sent“reference”conditions(Reynoldsonetal.,1997;Stoddardetal.,
2006;YatesandBailey,2010).
Itisexplicitlystatedin theRCA thatthereferencecondition
shouldbechosenbasedstrictlyoncriteriaconcerningtheminimal
exposureofthesitestohumandisturbances(Baileyetal.,2004).
Althoughhumandisturbancesaffectstreambiologicalandhabitat
attributes(Maddock,1999),referencesiteselectionshouldnotbe
basedoneitherbecauseitisdifficulttodistinguishbetweeneffects
fromhumandisturbanceandnaturalvariation(DovciakandPerry,
2002;Morenoetal.,2006).Infact,akeyaspectoftheRCAisthat
naturalvariabilityisintrinsicinecosystemsandthatthisvariability
mustbeaccountedforbyusingmodelstounderstandtheeffects
ofhumandisturbanceonassemblagestructureoffish(Oberdorff
etal.,2002;Tejerina-Garroetal.,2006;Pontetal.,2006,2009)and
macroinvertebrates(Clarkeetal.,2003;Baileyetal.,2004;Hawkins
etal.,2010;Moyaetal.,2011).
A multitude of stressors have been identified and used as
criteriafor determining referencesites.As geographic
informa-tionsystem(GIS) technologyhasbecomeoperationallysimpler
andwidely available(Kinget al.,2005), disturbances identified
at the catchment scale have been used for defining potential
reference areas (Collier et al., 2007; Wang et al., 2008).
How-ever,humanmodificationsactingatboth large(catchment)and
local(streamchannelandriparianzone)scalesshouldbe
inves-tigated becausepressures or stressors operating at both scales
canimpairthestreambiota (Bryceetal., 1999;Whittieret al.,
2007b;Hughesetal.,2010).
Increasingly,methodsfordefiningandselectingreferencesites
areappliedtolargespatialextents(wholeecoregions,states,and
countries),commonlyinvolvinghundredsorthousandsofsites.The
EnvironmentalProtectionAgencyoftheUnitedStatesofAmerica
(US-EPA),initsnationalWadeableStreamAssessment(WSA)
pro-gram,screenedaseriesofphysicalhabitatandwaterqualitydata,
settingthresholdsfortheselectionofleast-disturbedsitesin
differ-entecoregions(Herlihyetal.,2008).Thesame“filtering”approach
wasemployedinregionalassessmentsmadebythesameagency
(Klemmetal.,2003;Whittieretal.,2007b).Inasimilarapproach,
alargesetofcriteriaofhumandisturbancesoperatingatbothlocal
andregionalspatialscaleswereusedtoselectleast-and
most-disturbedsitesonEuropeanstreams(Nijboeretal.,2004;Pontetal.,
2006;Sánchez-Montoyaetal.,2009).
However,methodologiesemployedatlargespatialextentsmay
beinappropriateforstudiesdealingwithmorerestrictedspatial
extentsandfarfewersites.First,formostecosystemslocatedin
lessstudiedregionsoftheworld,suchasintropicaldeveloping
countries,thereisnoreliableinformationaboutthephysicaland
chemicalthresholdsthatindicatesubstantialdisturbance(Boyero
etal.,2009).Second,theapplicationofrigidfilterstoasmallnumber
ofsitesislikelytoselecttoofewsites,ornoneatall.EveninEurope,
whenhundredsofsitesfrom4countrieswereanalyzed,formany
streamtypesitwasnotpossibletofindanysinglesitethatfulfilled
allthecriteriaproposedforEuropeanreferenceconditions(Nijboer
etal.,2004).Nevertheless,manymonitoringinitiativesareapplied
atmorerestrictedgeographicareas(smalltomedium-sizedbasins
orsub-basins)andfarfewersites(dozensatbest)(Baptistaetal.,
2007;Morenoetal.,2009;Oliveiraetal.,2011;Surianoetal.,2011).
Toourknowledge,nosystematicmethodologyhasbeenproposed
toclearlydefinedisturbanceconditionsinthosesituations.
Whenworkingwithfewsites,insteadoftryingtoallocatesites
into‘boxed’categoriesfromtheonsetoftheproject(e.g.,least-,
intermediate-,andmost-disturbedsites),theuseofacontinuous
disturbancegradientcanbemoreadvantageousforclassifyingthe
sitesincludedinthestudy,enablingthedefinitionaposterioriof
theleast-disturbedsitesandthemost-disturbedsites.This
con-trastisnecessaryforthedevelopmentofMMIs(e.g.,Stoddardetal.,
2008;Oliveiraetal.,2011).Forinstance,predictivemodelsarefirst
concernedwithdescribingassemblagecompositioninreference
conditions(Reynoldsonetal.,1995),i.e.,the“goodtail”ofa
disturb-ancegradient.Inadditiontobiomonitoringstudies,anyapplied
ecologicalresearchconcernedwithchangesinpatternsand
pro-cessesassociatedwiththeintensityofhumanmodificationswill
benefitfromtheuseofadisturbancegradient.
Inthisstudywepresentamethodologytoquantitativelydefine
disturbancegradientsintwobasinssampledwitharelativelysmall
numberofsites(40each),eachbasinincludingarangeofsitesfrom
relativelyundisturbedtogreatlyaltered.Tothisend,weworked
withtwohypotheses.(1)Disturbancestakingplaceatbothlocal
(streamsites)andcatchmentspatialscalesreducetheEPT
assem-blagerichnessofthesites.(2)TheproportionofvariationinEPT
richnessassociatedwithnaturalvariabilityamongsitehabitatswill
begreaterinthebasinwiththeweakeranthropogenicdisturbance
gradient.
2. Methods
2.1. Studyarea
WesampledstreamsintwobasinsoftheCerradobiomeinthe
stateofMinasGerais,southeasternBrazil:UpperAraguaribasin(in
theParanáriverbasin)andtheUpperSãoFranciscobasin(inthe
SãoFranciscoriverbasin)(Fig.1).Bothstudyareaswere
demar-catedupstreamofthefirstbigreservoirofeachbasin(NovaPonte
andTrêsMariasreservoirs,respectively).TheCerradoisthe
second-mostextensivebiomeoftheNeotropics(Wantzen,2003),originally
covering20%ofBrazilianterritory,andoneoftheterrestrial
biodi-versity“hotspots”oftheplanet(Myersetal.,2000).Itisalsoone
ofthemostthreatenedduetoever-expandingpastureand
agricul-turalactivities(Wantzenetal.,2006).TheCerradoclimatehastwo
welldefinedannualseasons:adryseasonfromOctobertoMarch,
andawetseasonfromApriltoSeptember,with1200–1800mm
ofprecipitationperyear.Thevegetationistypicallysavannah-like,
withdenserforestformationsalongwatercoursesandwetareas.
Mostpeoplelivinginthestudyareasdwellonfarmsandinsmall
towns(upto20,000inhabitants),althoughafewsmallcities(up
to80,000inhabitants)arepresent.TheUpperAraguarihasawell
Fig.1.Locationofthebasinsandstreamsitessampled.(A)UpperAraguaribasinand(B)UpperSãoFranciscobasin.
soy,coffee,corn,andsugarcane culture.Irrigatedagricultureis
lesscommonintheUpperSãoFrancisco,wherepastureandsmall
familyfarmspredominate.
2.2. Siteselection
Forty“wadeable”streamsites(thatcanbetraversedbya
per-sonwading)rangingfrom1stto3rdorder(sensuStrahler,1957)
wereselectedon1:100,000scalemapsineachbasinandsampled
duringthedryseason.Thesiteselectionwasperformedthrougha
probability-baseddesignasdescribedinOlsenandPeck(2008),the
sameprocedureusedbytheUS-EPAintheEnvironmental
Moni-toringandAssessmentProgramWesternPilotStudy(EMAP-West,
Stoddardetal.,2005)anditsnationalWadeableStream
Assess-ment(WSA,Paulsenetal.,2008).Inthisapproach,amastersample
frame(MS)isfirstestablishedusingadigitizeddrainagesystem
map(1:100,000scale),andthenthesamplesitesareselectedviaa
hierarchical,spatiallyweightedcriteria(StevensandOlsen,2003).
Thisprocedureassuresabalancedselectionofsitesacrosstherange
ofstreamordersandgeographiclocation.TheUpperAraguarisites
weresampledinSeptember2009andtheUpperSãoFranciscosites
weresampledinAugust/September2010.
2.3. Sitehabitatmeasurements
ThefieldphysicalhabitatwasmeasuredasdescribedinPeck
etal.(2006).Thesitelengthsweresetat40timestheirmeanwetted
width,andaminimumof150m.Giventheirnarrowwidths,most
siteswere150mlong.Ineachsite,11equidistantcross-sectional
transectsweremarked,defining10sectionsofthesamelength.
In each transect and along thesections, a largeset of
mea-surementswererecorded,includingsitemorphology(e.g.,slope,
sinuosity,wettedandbankfullwidth,depth,andincisionheight),
habitatcharacteristics(e.g.,substratesizeandembeddedness,flow
type,and largewood),riparianstructure(e.g.,mid-channeland
marginshading,treeand herbaceouscoverdensity)andhuman
disturbanceinthechannelandriparianzone(e.g.,presenceof
pas-ture,crops,pipes,andtrash).Habitatmetricswerethencalculated
followingKaufmannetal.(1999).
Thefollowingphysicalandchemicalcharacteristicsofthewater
columnwerealsomeasuredinthefieldforeachsite:pH,
electri-calconductivity,andtotaldissolvedsolids(TDS).Watersamples
were collectedfor furtheranalysis in the laboratory, including
dissolved oxygen, turbidity, total alkalinity, total nitrogen, and
totalphosphorus.ThoseanalyseswereconductedfollowingAPHA
(1998).
Thesitenutrientconcentrationsofbothbasinswereextremely
low and not indicative of anthropogenic sources. In the Upper
Araguari,thevalues were0.06±0.01mg/L(mean±SD) fortotal
nitrogenand0.03±0.01mg/Lfortotalphosphorus.The
concen-trationsintheUpperSãoFranciscowere0.08±0.06mg/Lfortotal
nitrogenand0.02±0.01mg/Lfortotalphosphorus.
2.4. Macroinvertebratesamplingandlaboratoryprocessing
ThebiologicalsamplingalsofollowedtheprotocolofPecketal.
(2006)and Hughes and Peck (2008).Elevensample units were
takenperstreamsite,onepertransect,generatingonecomposite
sampleforeachsite.Eachsampleunitwascollectedthroughuseof
aD-net(30cmmouthwidth,500mmesh),effectivelysampling
1m2ofstreambottomareasampledpersite.Thesampleunitswere
obtainedbyfollowingasystematiczigzagpatternalongthesites
toavoidbiasinhabitatselection.Immediatelyaftercollection,the
compositesampleswereplacedinindividualplasticbucketsand
preservedwith10%formalin.
Inthelaboratory,themacroinvertebratesweresortedbyeye,
and theEPTindividualswereidentified togenusundera100×
Fig.2. Conceptualmodelofthedisturbanceplanewhoseaxesrepresenttheamount ofdisturbanceobservedatthelocalscale(in-streamandriparianzone)andatthe catchmentscale.Theidealleast-disturbedsiteswouldbethoselocatedclosestto theoriginoftheaxes,withfewdisturbancesobservedatbothscales.Theideal most-disturbedsiteswouldbethoselocatedintheoppositecorneroftheplane.Asingle measurementofdisturbancecanbetheEuclideandistance(calculatedthroughthe Pythagoreantheorem)betweenthelocationofthesiteintheplaneandtheorigin oftheaxes(seetheexampleinthefigurewithsite“A”).Forthispurposetheaxes valuesshouldbestandardizedatthesamescale.
keys(Pérez,1988;FernándezandDomínguez,2001;Mugnaietal., 2010).
2.5. Dataanalyses
2.5.1. Calculationofthedisturbancegradient
Todescribethetotalexposureofthesitestohumanpressures,
wedevelopedtwoseparateindices:onereflectingdisturbancesat
thesitescaleandonereflectingdisturbancesatthecatchmentscale,
bothhavingtheirorigins(0values)representingtheabsenceof
evi-denceofdisturbances.Ineachindex,thehigherthesitevalue,the
greatertheintensityofhumanmodificationsobservedforthatsite,
i.e.,thegreaterthedeviationfromthepristineconditionatthat
spa-tialscale.Thus,wepositionedeachsiteina‘disturbancebi-plane’
constructedwiththetwodisturbanceindicesasaxes.The‘ideal’
referencesitesshouldbethoselackingevidenceofhuman
mod-ificationsatboth near/in-streamand catchmentscales(concept
of minimally disturbed condition; Stoddard et al., 2006).
Typi-cally,however,referencesitesarethosewiththeleastdisturbances
amongthesitesavailable(conceptofleast-disturbedcondition;
Stoddardetal.,2006).Throughthisconceptualmodel,the
least-andmost-disturbedsitesinapoolofsitescanbevisualized
accord-ingtotheirpositionsinthedisturbanceplane,theleast-disturbed
sitesbeingclosertotheorigin(lowerleftcorneroftheplane)and
themost-disturbedsitesbeingfarthestfromtheorigin(upperright
corneroftheplane)(Fig.2).
Forquantifyingthelocaldisturbanceindex(LDI)weusedthe
metricW1hall,calculatedasdescribedinKaufmannetal.(1999),
ameasurecommonlyusedintheUS-EPAstreamassessments.This
metricsummarizestheamountofevidenceobservedin-channel
andintheriparianzonefor11typesofdisturbances(buildings,
channelrevetment,pavement,roads,pipes,trashandlandfill,parks
andlawns,rowcropagriculture,pasture,loggingandmining)along
theeleventransectsdemarkedatthestreamsite.Thevaluesare
weightedaccordingtotheproximityoftheobservationfromthe
streamchannel(Kaufmannetal.,1999).
Weassessedwatershedlandusesforeachsitethroughuseof
manualimageinterpretation.Watershedswereextractedfromthe
terrainmodelfromtheShuttleRadarTopographicMission–SRTM
(USGS,2005).Wemanuallyinterpretedhighresolution
multispec-tralimagesinconjunctionwiththeLandsatTMsensorusingSpring
software(Camaraetal.,1996).Thehigh-resolutionimages
pro-videdinformationabouttheshapeandtextureoftheelements,and
theLandsatimagesshowedspectralresponsefordifferenttargets.
Ourmappingidentifiedthreehuman-influencedlanduses
(pas-ture,agriculture,andurban).Thecatchmentpercentagesofeach
landusewereestimatedforeachsite.
Thecatchmentdisturbanceindex(CDI)wasbasedonthehuman
landusesinthecatchmentsandwascalculatedfollowing
Rawer-Jostetal.(2004),accordingtotheformula:
catchmentdisturbanceindex (CDI)=4×%urbanareas
+2×%agriculturalareas+%pastureareas
We evaluated the collinearity betweenlocal and catchment
humandisturbancesineachbasinthroughuseofPearson
corre-lationsbetweentheLDIandtheCDIvaluesofthesites.
Becausethelocalandthecatchmentdisturbanceindicesdonot
sharethesamenumericalscale,bothwereseparatelystandardized
toprovideasimilarscaleinvalues.Thistransformationwas
neces-sarytoreliablycalculateanintegrateddisturbanceindexforeach
site,basedonboththelocalandcatchmentindices(seebelow).The
valuesofeachindexweredividedby75%ofthemaximumvalue
thateachcantheoreticallyachieve.Wedidnotusethemaximum
valuesofeachindexforthesestandardizationsbecausethose
val-uesarerarelyachieved.Dividingbythemaximumvalueswould
shrinkgreatly andunnecessarily thevaluesin thestandardized
indices,shiftingnearlyallthesitesveryclosetotheoriginofthe
disturbanceplane.
TheCDIvaluespotentiallyrangefrom0(nolanduseinthe
catch-ment)to400(entirecatchmentoccupiedbyurbanareas).Sothe
valuesofthisindexweredividedby300.TheLDIvalues(W1hall
metric)potentiallyrangefrom0(noevidenceofanytypeof
dis-turbanceinthechannelorriparianzone)to16.5(all11typesof
disturbancesobservedinsidethestreamchannelinalltransects).
Butthistheoreticaluppervalueishighlyunlikelybecauseofspatial
limitationsandnegativecolinearitiesamongthetypesof
disturb-ance(listedabove).TheempiricalmaximumvalueoftheW1hall
metricisaround7(Kaufmannetal.,1999), sothevaluesofthis
indexweredividedby5.
Tosummarizethedisturbancesmeasuredatbothscalesina
singleindexwecalculatedforeachsiteanintegrateddisturbance
index(IDI).It wasmeasuredastheEuclidiandistance between
thepositionofthesiteinthedisturbanceplane(axes
standard-ized)totheoriginoftheplane(Fig.2).Thiswasperformedthrough
applicationofthePythagoreantheorem:
integrateddisturbanceindex(IDI)=
LDI52
+CDI3002
1/2
ThehighertheIDIofasite,themorethatsitedeviatesfromthe
‘origin’,i.e.,fromthe‘ideal’referenceconditionofnodisturbance
insidethestreamchannel,intheriparianzone,orinthecatchment.
Thus,wedefinedthedisturbancegradientsimplyastheascending
ordinationoftheIDI’sinapoolofsites.Thesteeperthedisturbance
gradientinapoolofsites,thegreaterthedifferenceinecological
conditionbetweentheleast-andmost-disturbedsitesinthepool.
2.5.2. EPTrichnessassociationswiththedisturbanceindices
To evaluate how EPT assemblages responded tothe degree
Table1
CandidatesitehabitatmetricsforexplainingEPTrichnessvariabilityinbothstudiedbasins.
Metricname Metriccode Notsignificantlycorrelated Notstronglycorrelated
Withdisturbances(p>0.05) Amongeachother(r<0.6)
UpperAraguari UpperSãoFrancisco UpperAraguari UpperSãoFrancisco
Meanwidth xwidth *
Meandepth xdepth *
Meanslope xslope
Meanbankfullwidth XBKF W
Meanwidth× meandepth XWXD * * * *
Mean(width/depth) xwdrat * *
Meandepth×meanslope xdepthxslope
Bankfull(width/depth) BKFWDrat * * * *
Meanresidualpoolarea rp100 *
Meanwatervolume/m2 v1w msq
Ripariancanopy(>5mhigh)presence xpcan
Ripariancanopy(>5mhigh)cover XC *
Totalripariancover(allvegetationlayers) xcmg * * * *
Totalriparianwoodycover xcmgw
Meancanopydensity(mid-stream) xcdenmid
Naturalcoverinthestream(all) xfcnat * *
Naturalcoverprovidedbylargewood xfclwd * *
Percentageoffastwater pctfast * * * *
Percentageoffines(siltandclay) pctfn *
Percentageofsand+fines pct sfgf *
Percentageofcobble pctcb * *
Percentageofcoarsesubstrate(>16mm) pctbigr *
Logofmeansubstratediameter lsubdmm * *
Meansubstrateembeddedness xembed
Logofrelativebedstability LRBS
pH pH * *
Conductivity(S/cm) Cond
Totaldissolvedsolids(g/L) TDS * *
Turbity(NTU) Turb * *
Dissolvedoxygen(mg/L) DO * *
Alkalinity(mequiv./L) Alk
conductedmultiplelinearregressionsbetweenEPTrichnessand
thestandardizedLDIandCDIofthesitesforeachbasin.Wealso
regressedEPTrichnessagainsttheIDItoevaluateitsperformance
relativetoEPTrichnessvariability.
2.5.3. Contributionofnaturalvariabilityofsitehabitat
characteristicstoexplainingthevariationofEPTrichness
Throughthefollowingmethodology,weevaluatedhowmuch
naturalphysicalhabitatvariabilityaddedtotheexplanationofEPT
richnessprovidedbythedisturbancegradientalone.Theprocess
wasperformedseparatelyforeachbasin(Fig.3).
Westartedwithasetof31habitatmetricscalculatedfromthe
rawfielddata(Table1).Withthesemetricsweaimedto
repre-sentkeyaspectsofthehabitatsofthesites,suchasmorphology
(e.g.,mean wetted and bankfull width,mean depth, and mean
slope),ripariancondition(e.g.,riparianvegetationextentandmean
canopy cover),habitatheterogeneity(e.g.,% fastwater,% large
substrates,%finesubstrates,andmeansubstrateembeddedness)
andwaterquality(e.g.,dissolvedoxygen,pH,andalkalinity).We
obtainedPearsoncorrelationsbetweenthosemetricsandallthe
disturbancedescriptorswehadavailable:the3landuses
percent-ages,the11typesoflocalsitedisturbances,theLDI,theCDIand
theIDI.Allmetricssignificantlycorrelated(p<0.05)withanyof
thedisturbancedescriptorsweredisregardedforthenextstepof
theanalysis.Inthisway,wefilteredallthehabitatmetricsthat
couldbeaffectedbyhumandisturbancesofanykind;theremaining
metricswereconsideredassourcesofnaturalvariationinthesites.
Next,a Pearsonproduct–momentcorrelationmatrix was
calcu-latedwithallthemetricsnotcorrelatedwithdisturbanceevidence.
Theredundantmetrics(r>0.6)wereremovedandthechoiceofthe
metricstoberetainedwasbasedonecologicalrationale.
Amongthe31initialhabitatmetrics,manywerenot
signifi-cantlycorrelatedwithhumandisturbances(9intheUpperAraguari
and15intheUpperSãoFrancisco,Table1).Inbothbasins,someof
theremainingmetricswereremovedbecauseofhighcolinearities
(r>0.6).IntheUpperAraguari,meandepthandmeanresidualpool
areawereremovedandmeanwettedwidth×meanthalwegdepth
waskept,becausewebelievethelattermetricbestsummarized
thestreamchannelsize.IntheUpperSãoFrancisco,meanwetted
widthwasremovedandmeanwettedwidth×meanthalwegdepth
waskept(samereasonasabove)andripariancanopycoverwas
removedandtotalripariancover,amoreembracingmetric,was
kept.Percentageofcoarsesubstrates(>16mm),percentageoffines
(<0.06mm: silt and clay), percentage of sand+fines (<2.0mm),
andlogofthegeometricmeansubstratediameterhadhigh
cor-relations. We chose toinclude log of mean substratediameter
becauseitbestrepresentedthepredominantsubstratesizesofthe
sites.
We used the reduced set of habitat metrics to perform a
hierarchicalmultipleregression,forcingtheentranceofthe
inte-grateddisturbanceindex(IDI)inthefirstblockandallowing,in
the secondblock, a best-subsetsmultiple regression procedure
searchforthecombinationsofhabitatmetricsthatbestexplained
the remaining variability in EPT richness. The R2 values were
considered ascriteria for the selection ofthe bestmodels. We
restricted the number of predictor variables in the final
mod-elstoa total of 4 (10%of40 sites)toavoid model over-fitting
(Harrell, 2001; Tabachnick and Fidell, 2007). Thus, in addition
totheIDIinthefirstblock,threehabitatmetricswereallowed
toenterinthesecondblock.Hierarchicalmultipleregressionis
an efficientwaytoisolatethe contributionofsomefactor in a
regressionmodelbecauseresidualregressionscanleadtobiased
Fig.3.Summaryofthemethodologicaldesignusedtoteststatisticallyhowsitehabitatmetricsnotsubjectedtohumandisturbancesenhancedtheexplanationgivenbythe integrateddisturbanceindex(IDI)toEPTrichnessineachbasin.
(A) (B)
Upper Araguari Upper São Francisco
Basin 0 1 2 3 4 5 Loc a l D is tur ba nc e I n de x ( L D I)
Upper Araguari Upper São Francisco
Basin 0 50 100 150 200 250 300 C a tc h m en t D ist u rb a n c e I n d ex ( C D I)
Fig.4. Distribution(mediansandquartiles)ofthevaluesofthe(A)localdisturbanceindexandofthe(B)catchmentdisturbanceindexineachstudiedbasin.
2002).OnlypredictorvariableswithindividualF-values>1were
allowedinthefinalmodels.Thestatisticalsignificanceofthe
hier-archicalmultipleregressions(block1vsblock1+block2)were
tested through analysis of variance (ANOVA’s). In this way we
tested whether thehabitat metrics contributed significantly to
the explanation of EPT richness derived from theIDI for each
basin.
3. Results
3.1. Localandcatchmentdisturbanceindicesandthedisturbance
plane
ThetwobasinshadsimilarpatternsinmostLDIvalues(Fig.4A),
althoughtheUpperAraguaribasinhadafewhighervalues,
result-ing from urban sites. On the other hand, the patterns of CDI
valuesvariedconsiderablybetweenthebasins,theUpperAraguari
had higher CDI values than the Upper São Francisco(Fig. 4B).
IntheUpperSãoFrancisco,onlyonesitehad aCDIvalue>100.
Thedifferingpatternsareexplainedbythelandusepatternsin
both basins (Fig. 5A–C). In the Upper Araguari we observed a
higherproportionofagricultureinthecatchments,whereasinthe
UpperSãoFranciscopasturepredominated.Proportionsofurban
areas were low in both basins, most catchments having none.
ThePearson correlations betweentheLDI andCDI scoreswere
weak(r=0.21intheUpperAraguariandr=0.35intheUpperSão
Francisco).
In both basins few sites were located close to the
ori-gin on the disturbance plane (Fig. 6), but because of higher
CDI values, more Upper Araguari sites were located farther
from the origin. This distribution pattern is summarized by
thedifferent slopes of the disturbancegradients of thebasins,
showing the IDI values in ascending order (Fig. 7). In the
Upper Araguari we observed a much wider range in site
IDI values (i.e., more sites nearer and farther from the
ori-gin), indicating a much stronger disturbance gradient in that
basin.
3.2. DescriptionoftheEPTassemblages
A total of 5463 EPTindividuals (61 genera) were identified
in Upper Araguari sites, and 15,133 EPT individuals (65
gen-era)wereidentifiedinUpperSãoFranciscosites.Inbothbasins
Ephemeroptera comprised the majority of the EPT genera (30
in theUpper Araguari and 35 in theUpper São Francisco) and
numberoforganisms(3291intheUpperAraguariand12,529in
theUpperSãoFrancisco).IntheUpperAraguari,themost
abun-dantgenerawereSmicridea(Trichoptera),andtheEphemeroptera
Thraulodes, Traverhyphes and Tricorythopsis. Those four genera
represented 43% of theEPT individuals collected in the Upper
Araguari.IntheUpperSãoFrancisco,themostabundantgenera
wereCallibaetis,Cloeodes,Americabaetis,CaenisandTraverhyphes,
allEphemeroptera.Thosefivegenerarepresented54%oftheEPT
individualscollectedintheUpperSãoFrancisco.Around25%ofthe
taxaidentifiedintheUpperAraguari,and20%ofthetaxa
iden-tified in the Upper São Francisco,can beconsidered rare taxa,
withjust5orfewerindividualsidentifiedacrossallsitesofeach
basin.
3.3. EPTrichnessversusdisturbanceindices
The variation of EPT richness explained by theLDI and CDI
togetherwasmuchhigherintheUpperAraguari(R2=0.40)than
intheUpperSãoFrancisco(R2=0.18)(Table2).Inbothbasins,EPT
richnesswassignificantlyrelatedtotheCDI,butonlyintheUpper
AraguarididtheLDIcontributesignificantlytoexplainEPTrichness
variation(Table2).TheslopebetweenLDIandEPTrichnessinthe
UpperSãoFranciscoapproachedzero(Table2).Asexpected,all
sig-nificantrelationshipswerenegative.IntheUpperAraguari,theIDI
explainedamoderateamountofEPTrichness(Simplelinear
regres-sion,R2=0.39,F
(1,38)=24.6,p<0.001;Fig.8A),nearlythesameas
thecombinedexplanationsgivenbytheLDIandCDIinthe
multi-pleregression.IntheUpperSãoFrancisco,theIDIexplainedpoorly,
butsignificantly,EPTrichnessvariation(simplelinearregression,
R2=0.11,F
(1,38)=4.55,p=0.039;Fig.8B).
Table2
MultipleregressionresultsforeachbasinwithEPTrichnessastheresponsevariableandthelocaldisturbanceindex(LDI)andthecatchmentdisturbanceindex(CDI)as predictorvariables.
F-Value(2,37) p-Value R-Square Beta Std.err.ofbeta t(37) p-Value
UpperAraguari 12.5 <0.001 0.403
Intercept 9.806 <0.001
CDI −0.450 0.130 −3.467 0.001
LDI −0.364 0.130 −2.802 0.008
UpperSãoFrancisco 4.005 0.027 0.178
Intercept 8.820 <0.001
CDI −0.424 0.159 −2.671 0.011
0 20 40 60 80 100 Percentage of pasture 0 20 40 60 80 100 Percentage of agriculture (A) (B) (C) Basin 0 20 40 60 80 100
Percentage of urban areas
UpperAraguari UpperSão Francisco
Fig.5.Distribution(mediansandquartiles)ofthepercentagesof(A)pasture,(B) rowcropagriculturaland(C)urbanareasinthecatchmentsofthesitessampledin eachbasin.
3.4. Contributionofnaturalvariabilityofhabitatcharacteristics
inexplainingEPTrichness
Thehierarchicalregressionsinformedhowtheexplanations(R2
values)givenbytheIDIstoEPTrichnessvariationswereincreased
bytheadditionofhabitatmetricsnotrelated tohuman
distur-bances.In theUpper Araguari,theincrementwaslow and just
marginallysignificant(Table3).Inthatbasin,theR2valueincreased
from0.39to0.49,anincreaseof0.1.Ontheotherhand,intheUpper
SãoFranciscotheincrementwasmuchgreater,theR2valuerising
from0.11to0.50,anincrementof0.39.Theamountofexplanation
given bythe combinedmodels (IDI+habitatmetrics not
corre-latedwithdisturbance)weresimilarinbothbasins,withR2values
around0.5,meaningthatthefinalmodelsexplainedonlyabouthalf
thevariation.
Thecombined models generated frombest-subsetsmultiple
regressionshad, inadditiontotheIDI,2 habitatmetricsinthe
UpperAraguariand3habitatmetricsintheUpperSãoFrancisco
(allwithF-values>1,Table3).Inboth basins,asite sizemetric
(meanwidth×meandepth)wasimportantinexplainingEPT
rich-nessvariation.IntheUpperAraguari,anothermorphologicmetric
(bankfullwidth/depth)wasincorporatedinthemodel,whereasin
theUpperSãoFrancisco,microhabitatmetrics(percentfastflows
andlogofmeansubstratediameter)wereincluded.
4. Discussion
4.1. Premisesforcomparisonsbetweensites
Ithasbeenlongrecognizedthatsomegeographic(e.g.,
ecore-gions) and non-geographic features (e.g.,typologies) of stream
sitesexercise astronginfluence onthecomposition and
struc-tureoftheirmacroinvertebrateassemblages(Hughes,1985,1995;
Gerritsenetal.,2000).Accordingly,itisimportantfortheassigned
referencesitesandthetestsitesofastudytosharethesekey
biolog-icaldrivers,allowingreliablecomparisonsbetweenthem(Herlihy
etal.,2008).InthewordsofGerritsenetal.(2000)itisimportant
to“putlikewithlike”.
Gradualchangesinthehabitattemplate,intheavailablefood
resources,andinthebiologicalassemblagesnaturallyoccuralong
the longitudinal gradient of lotic ecosystems (from spring to
mouth),resultingmainlyfromdownstreamchangesintheir
mor-phologicaldimensions,catchmentareasanddischarges(Vannote
etal.,1980;Poole,2002;Hughesetal.,2011).Wereducedsuch
sourcesofvariationbyselectingstreamswithsimilar
morphologi-caldimensions.Allsitescanbeclassifiedassmallstreams,closeto
theheadwaters.
There is no geographic classification formally designed for
Brazil that is comparable in detail to the ecoregion
classifica-tionsoftheUSA(Omernik,1995)orEurope(e.g.,Gustafssonand
Ahlén,1996).However,thebasinsstudiedareinthesamebiome
0.0 0.2 0.4 0.6 0.8 1.0
LDI (standardized) 0.0 0.2 0.4 0.6 0.8 1.0
CDI (standardized) UpperAraguari sites
UpperSão Francisco sites
Fig.6.Distributionofthesitesofeachbasininthedisturbanceplane,withUpperAraguarisitesrepresentedbyopencircles()andUpperSãoFranciscositesrepresented byfilledboxes( ).Theaxesofthelocaldisturbanceindex(LDI)andthecatchmentdisturbanceindex(CDI)werestandardizedatthesamescale(relativepositionsofthe sitesoneachaxis,andintheplane,wereretained).
0 5 10 15 20 25 30 35 40 Ranking of the streams (from least- to
most-disturbed) 0.0 0.2 0.4 0.6 0.8 1.0 1.2
Integrated Disturbance Index (IDI)
UpperAraguari sites UpperSão Francisco sites
Fig.7.Disturbancegradientsinbothbasins,representedbyascendingsitevaluesoftheintegrateddisturbanceindex(IDI).UpperAraguarisitesarerepresentedbyopen circles()andUpperSãoFranciscositesarerepresentedbyfilledboxes( ).
(Cerrado)andinthesamegeneralterrestrialandaquaticecoregions
outlinedbyOlsonetal.(2001),meaningthatthesitessharesimilar
climatic,edaphic,vegetation,geologicalandbiogeographic
condi-tions(Olsonetal.,2001;Wantzen,2003).Moreover,thebasinswere
analyzedseparately,andtheirindividualareasaremuchsmaller
thanthoseoftheUSlevelIVecoregions,themostdetailedlevel
oftheirclassification.Thus, althoughlackinganofficialdetailed
classification,weconsiderallthesitesinthesameecoregion.
4.2. Theroleofthedisturbancesmeasuredatlocalandcatchment
spatialscales
Asstatedintheclassicalviewofstreamimpairment,human
disturbancesoperatingatmultiplescalescanalterpatternsand
processesofthenaturalhabitat,ultimatelyleadingtomodifications
orimpairmentofbiologicalassemblages(Karr,1999;Norrisand
Thoms,1999;Bryceetal.,1999;FeldandHering,2007).However,
theexactmechanisticpathwaysamongtheoriginsofimpairment,
thehabitatmodifications,andthebiologicalresponsesarenotwell
knowninmostcases(BedfordandPreston,1988;Karr,1991).For
this reason,rather thansearching forall theindividualsources
ofimpairment,itisimportanttodevelopagroupofdisturbance
metricsthatcanserveasgeneralindicatorsofthetotalpressureto
whichanecosystemmaybesubjected(Boulton,1999).
Disturbancesinthechannelorriparianzonecanimpairthe
habi-tatsandthebiota(Bryceetal.,1999;DeathandJoy,2004;Kaufmann
andHughes,2006).Becausecatchmentsdrivethestreamfeatures
inalmosteveryaspect(Hynes,1975;Wiens,2002),humanland
usesarealsousuallylinkedwiththeecologicalconditionofstreams
(Bryce et al., 1999; Allan,2004; Wang et al., 2008).Non-point
sources in catchments commonly contribute excess sediments,
nutrientsandpollutantstostreamsandrivers(AllanandCastillo,
2007;Allan,2004).Humanactivitiesinthecatchmentalso
influ-encetheconditionofstreamriparianzones(VanSickleetal.,2004;
Sponselleretal.,2001;Miserendinoetal.,2011).Theorderingof
“disturbancepotential”usedinthisstudy(urbanareashavingmore
weightthanrowcropagriculture,whichinturnhasmoreweight
thanpasture),aswellastheuseofthewholecatchmentareaas
the“buffer”toestimatecatchmenthumanpressures,are
corrob-oratedbymanypreviousstudies(Sponselleretal.,2001;Mebane
etal.,2003;Wangetal.,2008;Guckeretal.,2009;Trautweinetal.,
2011).Inourstudy,disturbancesmeasuredatlocalandcatchment
spatialscalesbothreduced EPTrichness,corroboratingourfirst
hypothesis.InagreementwithKailetal.(2012),catchment
distur-banceshadagreatereffectthanlocaldisturbancesinthesebasins.
Thelatterwerenotevensignificantlyrelatedtomacroinvertebrate
richnessintheUpperSãoFranciscosites.
Localdisturbancewasnotcorrelatedwithcatchment
disturb-ance.Thislackofassociationmeansthatcatchmentlanduseswere
notdrivingnearorin-streammodifications,andwhatisobserved
at onescale candiffer fromwhat isobserved at theother.For
instance,inourstudyweobservedcatchmentshighlydominated
byrowcropagriculturebutwithundisturbedriparianvegetation
and streamchannels.Conversely,we alsohadcatchments with
mostly natural land coverbut stream channelsaltered by
live-stock.Scenarioslikethesearelikelytohappenelsewhere(Nijboer
etal.,2004).Consequently,relyingonjustonescaletodescribethe
levelofhumanpressureatasitecanleadtomisleading
interpre-tationsofbiologicalresponses(Bryceetal.,1999;FeldandHering,
2007).
Table3
Hierarchicalmultipleregressionresultscontrastingthesignificanceofthedifferencesbetweentheregressionmodelsineachbasin.Thefirstmodels(block1)consistedof simpleregressionswithEPTrichnessastheresponsevariableandtheintegrateddisturbanceindex(IDI)asthepredictorvariable.Thesecondmodels(block1+block2) includedaspredictorvariablesthehabitatmetricsselectedbythebestsubsetsprocedureasthosewhich,togetherwiththeIDI,betterexplainedEPTrichness.Habitatmetric codesaredefinedinTable1.
Basin F-Value p-Value R-Square Metrics’meanbetavalues ANOVAtestforhierarchical
regressionanalysis[block1vs (block1+block2)] F-Value p-Value UpperAraguari Model1 24.6 <0.001 0.393 IDI 3.194 0.053 −0.627
Model2 11.28 <0.001 0.484 IDI−0.523 XWXD0.227 BKF0.299WDrat
UpperSãoFrancisco
Model1 4.548 0.04 0.107 IDI
9.144 <0.001 −0.327
(A)
(B)
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Integrated Disturbance Index (IDI)
0 5 10 15 20 25 30 35 40 EPT r ic h n e s s 0 5 10 15 20 25 30 35 40 E P T r ic hne s s
Fig.8. Linearregressionsbetweentheintegrateddisturbanceindex(IDI)valuesand theEPTrichnessofthesitesof(A)theUpperAraguaribasin,representedbyopen circles(),and(B)theUpperSãoFranciscobasin,representedbyfilledboxes( ).
Theintegrateddisturbanceindex(IDI)proved tobea useful
andaccurateunivariatedescriptorofthetotalityofdisturbances
measuredatdifferentspatialscales.Itexplainedthevariabilityin
EPTrichnessbetterthanseparatelocalandcatchmentindices,and
almostaswellaswhenthosetwoindiceswereseparatelyincluded
inmultipleregression.Theexistenceofasingleindexto
summa-rizetheoverallecologicalcondition,althoughneverperfect,isa
quickand practical way todescribetheconditionof individual
sitesand therelative conditionof a site in comparisonto
oth-ers(Bryceet al.,1999; Wanget al.,2008).Thisis necessaryto
setdisturbancethresholdsand topresent tosocietyand
stake-holdersanobjectiveandsimplemeasurementofsiteconditions
(Hughesand Peck,2008).Therange anddistributionofIDI
val-uesacrossarepresentativepoolofsitescanindicatethestrength
ofthedisturbancegradientinaregion.Thegreatertherangeand
evennessofthedistributionofsitesacrossthatrange,thegreater
thestrengthofthedisturbancegradient(shownintheascending
ordinationsofFig.7),andthegreatertheexpecteddifferencesin
ecologicalconditionbetweentheleast- andthemost-disturbed
sites.
4.3. Theroleofnaturalhabitatvariation
Theimportanceofnaturalstreamhabitatvariation hasbeen
long recognized in stream ecology (Karr and Dudley, 1981;
Allan and Castillo, 2007). Metrics related to hydromorphology
(percentage of fast flows, mean wetted width×mean thalweg
depth, bankfull width/depth, log of geometric mean substrate
diameter), which were not related to human disturbances in
thesebasins,helpedexplainEPTrichnessvariability,apartfrom
theeffects thatcouldbeattributedsolely tohumaninfluences.
Thosefactorsarecommonlyreportedasimportantfor
structur-ing stream macroinvertebrate assemblages (Schmera and Er ˝os,
2004; Brooks et al., 2005; LeCraw and Mackereth, 2010).
Con-sistent with our second hypothesis, the relative and absolute
contributionof thenaturalhabitatwasmuchmorepronounced
intheUpperSãoFranciscobasin,whichhadaweakerdisturbance
gradient.
Oneconclusionemergingfromourresultsisthatifthe
anthro-pogenicdisturbancegradientisnotstrong,thedeleteriouseffect
ofhumanactivitiesonassemblagerichnesswillbemostlyeclipsed
byvariationassociatedwithstreamhabitatnaturalvariability.In
otherwords,thedisturbance“signal”willbeburiedbyhabitat
vari-ation“noise”(ParsonsandNorris,1996;GerthandHerlihy,2006).
Ascan beobserved in theUpper São FranciscoBasin(Fig.8B),
sitesthatwereslightlymoreperturbedfrequentlyhadhigherEPT
richness thanothers thatwereslightly less perturbed.Many of
thesedivergences in relation towhat would be expectedfrom
thedisturbance-onlymodelwereprobablydrivenbydifferences
instreamhydromorphology.IntheUpperAraguaribasin,which
hadastrongerdisturbancegradient,thosesituationsalsooccurred,
butlessfrequently(Fig.8A).Asecondconclusionisthattheeffort
tocontrolbroad-scaledriversofbiological assemblagesthrough
useof ecoregionsandstreamtypologiesdoesnoteliminatethe
necessitytoaccountforlocalhabitatvariabilitywhencomparing
sites(Hughesetal.,1986;Waiteetal.,2000;Pintoetal.,2009).
Althoughweaimedtostandardizethestreamsizes,asizemetric
(meanwidth×meandepth)stillexplainedsignificantdifferences
inEPTrichness.Inaddition,evenneighboringsitesmayhavehighly
dissimilarhabitatsandbiologicalassemblages(Downesetal.,2000;
FinnandPoff,2005;Ligeiroetal.,2010),sothatecoregion
standard-izationalsoisnotenough.
TheamountofEPTrichnessvariabilityexplainedwassimilarin
bothbasins(around50%).Thisvaluecanbeconsideredhigh,given:
(1)theintrinsiccomplexityandunpredictabilityofstream
ecosys-temsandthedifficultyofobtaininggoodmodelsofthem(Harris
andHeathwaite,2011),(2)thesourcesofvariationnotaccounted
forinthisstudy,suchaslegacyeffects(Allan,2004)and
condi-tionsatupstreamreaches(KailandHering,2009)oratneighboring
sites(Sandersonetal.,2005),and(3)theintrinsicunpredictability
(“noise”)relatedtoseasonalandsamplingvariability(Kaufmann
etal.,1999;KaufmannandHughes,2006).Weemphasizethatthe
streamhabitatcontributiontorichnessexplanationwasanalyzed
inaveryconservativeway.Toreliablydeterminethedegreethat
naturalhabitatvariabilitycanaddexplanationatvaryinglevelsof
disturbancestrength,wedealtonlywiththehabitatmetricsnot
significantlycorrelatedwithanyofthedisturbancemeasurements
wehadavailable.Inthisregard,weevendiscardedmetrics
sig-nificantlybutweaklycorrelatedtodisturbance(e.g.,r<0.4).Thus,
webelievethathabitatvariabilityhasagreaterroleinstructuring
macroinvertebrateassemblagesthanshowninourresults,because
thoserejectedhabitatmetricsthatwererelatedtohuman
distur-banceswerealsodrivenbynaturalvariabilitytosomedegree(King
etal.,2005).
4.4. Importanceoftheconstructionofadisturbancegradient
Theexplicit,quantitativedeterminationofadisturbance
gra-dientismoreadvantageousthanasetofdisturbancecategories
because distinct separations in ecological conditions shouldbe
rareinanygroupofsites(Whittieretal.,2007b;Herlihyetal.,
most-disturbed, or impaired. Depending on the intensity and
extentofhumaninfluencesinthelandscape,sometimesitis
nec-essarytorelaxthestringencyoftheacceptancethresholdsinorder
tofindleast-disturbedconditions(Stoddardetal.,2006;Whittier
etal.,2007b;Herlihyetal.,2008).Soitisimportanttorecognizethe
relativityoftermslike“least”,or“most”,whendescribingecological
condition(Stoddardetal.,2006).Absolute,“boxed”designations,
althoughcomfortableandoperationallyeasiertohandle,canlead
tomisunderstandingsor erroneouscomparisons amongstudies
simplybecausethetrueecologicalconditionsofthesitesalongthe
disturbancegradientcontinuumwerenotexplicitlystated.
Oftenthedesignations ofreferenceandmost-disturbedsites
aremadepriortosampling(Baileyetal.,2004).GISdataand
tech-niqueshavebeenwidelyappliedwhenscreeningforreferencesites
(Collieretal.,2007;YatesandBailey,2010)andfield
reconnais-sanceisstronglyrecommended (Hugheset al.,1986;Yatesand
Bailey,2010).Yet,eveninthosecasesweencourageresearchersto
quantitativelyre-assessthedisturbancegradientafterfield
samp-lingtocheckthevalidityofanypreviousclassificationsandthe
exactquantitativedifferenceintheconditionsbetweenthe
“refer-ence”and“test”sites.
4.5. Thebenefits,scopeandfurtherpossibilitiesoftheproposed
methodology
Thedisturbanceplaneconceivedinthiswork,visually
describ-ingtheintensityofhumandisturbancesatbothlocalandcatchment
scales,establishedaneasyandintuitivewaytodescribethetotal
amountofpressureatsites.Thedisturbanceplanefacilitates
com-parisonsofsiteconditionsinamorestraightforwardandspecific
manner,quantitativelypositioningeachsitealong adisturbance
continuum,ratherthanassigninglabelstothesites.When
neces-sary,labelssuchas“minimally-”,“least-”and“most-disturbed”can
beassignedtositesbasedonquantitativedataversussubjective
decisions.Objectivecriteriaandquantitativeapproachestoselect
referencesiteshavebeenprovenmoreefficientforselectingthe
“best”sites(Whittieretal.,2007b),andthesamemaybetruefor
selectingthe“worst”ones.
Becauseonlydirectobservationsofhumanactivitieswereused
todescribeanthropogenicpressure,furthercharacterizationofthe
chemical and physical habitatofthe least-and most-disturbed
sitescanbemadewithoutincurringanyconceptualcircularity.As
addressedbefore,metricslikedissolvednutrientconcentrations,
ripariancoverandsedimentsizes,althoughcommonlyassociated
withhumanmodifications, are also subjectto natural
variabil-ity(Kinget al.,2005;Miserendino etal.,2011).Forexample,in
this studyno landuse measurement orlocal modificationwas
correlated withnutrientconcentrations (total phosphorousand
totalnitrogen).Lownutrientconcentrationsarecommonin
Cer-rado streams because of naturally oligotrophic soils (Wantzen,
2003).In theUpper São Francisco, noevidence of disturbance
wascorrelatedwithsubstratesizesandriparianvegetationcover
(Table1).So,inaccordwithBaileyetal.(2004),naturalpatterns,
notresearchers’opinions,shouldbeusedtocharacterizereference
conditionattributes.
Theproposedmethodologywaswellsuitedfordescribingthe
disturbancegradientofthe40siteswestudiedineachbasin.When
necessary,sitesfromdifferentregionscanbeincorporatedinthe
samedisturbanceplane(asshowninFig.6).Webelievethatthis
methodologyis alsoapplicabletolargerdatasets,although
fur-therresearchisneededtoconfirmthisassumptionandtocompare
outputsgeneratedthroughotherapproaches.
Dependingonresearcherpreferencesandtheamountofdata
available,localand/orcatchmentdisturbanceindicescanbe
cal-culated in different ways, perhaps using different disturbance
measurements. For instance, other commonly used metrics to
characterizehumanpressureincludehumanpopulationdensity,
livestockdensity,number ofdwellingsand roaddensity(Wang
etal.,2008;Brownetal.,2009).Ifonedesiresfurtherchangesinthis
methodology,moredisturbanceaxescanbeaddedtothemodel,
perhapsrepresentingfactorsconsideredkeystressorsinparticular
studies (e.g.,dams and toxic substances).This willgenerate
n-dimensionaldisturbancepolygons,ratherthanthebi-dimensional
disturbanceplanepresentedinthiswork.Althoughsuch
refine-mentserodethesimplicityandvisualappealofthemodel,they
couldimprovetheaccuracyoftheintegrateddisturbance
quantifi-cationsofthesites(Danzetal.,2007).
Inourstudy,theIDIwasareliableunivariatemeasurementof
sitedisturbancestatus.TheIDIisalsoagoodtoolfordescribing
thedisturbancegradientstrengthinapoolofsites,viatherange
anddistributionofitsvalues.So,ratherthanastandardizedand
rigidmethodology,weofferaflexibleandadaptiveframeworkfor
characterizingandquantifyingdisturbanceinmanysituations.
5. Conclusions
Weshowedthroughourresultsthatareliableand
comprehen-sivecharacterizationofhumanpressuresonstreamsreliesonthe
useofdifferenttoolsandshouldintegratedatafromdifferent
spa-tialscales.Inourstudy,localandcatchmentdisturbanceswerenot
correlated,andbothindependentlyaffectedsiteEPTassemblages.
Theproposedmethodologyquantifiedthehumanpressureonsites
withoutresortingtonaturallyvaryinghabitatmetrics.We
demon-stratedthatthestrengthofthedisturbancegradientinfluencedthe
degreetowhichnaturalhabitatvariabilityexplainedEPTrichness
variation,afindingthathasimportantimplicationsfor
biomoni-toringstudies.Thus,theuseofquantitativedisturbancegradients
isessentialforefficientuseofecologicalindicatorsandweadvise
researcherstodefinequantitativelythedisturbancestatusoftheir
studysites.Inthisstudywepresentedaframeworkfordoingso.
Acknowledgements
Wereceivedfundingandsupportforthisresearchfrom
CEMIG-ProgramaPeixeVivo,CAPES,CNPq,FAPEMIGandFulbrightBrasil.
WethankTonyOlsen,MarcWeber,andPhilLarsenoftheCorvallis
EPALaboratory(Oregon,USA)forassistanceindevelopingspatial
samplingdesignsandselectingwadeablestreamsamplesitesin
ourBrazilianbasins.WethankAmandaNahlikandBobOzretich,
fromUS-EPA,forfurthereditorialchanges.JulianaFranc¸aandAna
PaulaEllerwereresponsibleforwaterqualityanalyses.CarlosB.M.
Alveshelpedwithgenerallogisticsandfieldwork.Colleaguesfrom
theLaboratóriodeEcologiadeBentos(UFMG),CEFET-MG,Federal
UniversityofLavras (UFLA)andPontificalCatholicUniversityof
MinasGerais(PUC-MG)assistedwithfieldcollections.Discussions
heldduringthe“WorkshoponEcologicalAssessment:the
Foun-dationforEvaluatingBiologicalPatterns”(October3–7,2011,US
EPAWesternEcologyDivision,Corvallis,OR,USA)addedconcepts
andideasforthisstudy.Themanuscriptwaswrittenwhilethefirst
authorwasaguestresearcherattheUSEPACorvallisLaboratory.
References
Allan,J.D.,2004.Landscapeandriverscapes:theinfluenceoflanduseonriver ecosys-tems.Annu.Rev.Ecol.Evol.Syst.35,257–284.
Allan,J.D.,Castillo,M.M.,2007.StreamEcology:StructureandFunctionofRunning Waters,2nded.ChapmanandHall,NewYork,NY.
AmericanPublicHealthAssociation(APHA),1998.In:Clesceri,L.S.,Greenberg,A.E., Eaton,A.D.(Eds.),StandardMethodsfortheExaminationofWaterand Waste-water.,20thed.AmericanPublicHealthAssociation,Washington,DC. Bailey,R.C.,Norris,R.H.,Reynoldson,T.B.,2004.BioassessmentofFreshwater
Ecosys-temsusingtheReferenceConditionApproach.KluwerAcademicPublishers, Boston.
Baptista,D.F.,Buss,D.F.,Egler,M.,Giovanelli,A.,Silveira,M.P.,Nessimian,J.,2007. Amultimetricindexbasedonbenthicmacroinvertebratesforevaluationof AtlanticForeststreamsatRiodeJaneiroState,Brazil.Hydrobiologia575,83–94. Barbour,M.T.,Gerritsen, J.,Snyder,B.D., Stribling,J.B., 1999. Rapid Bioassess-mentProtocolsforuseinStreamsandWadeableRivers:Periphyton,Benthic MacroinvertebratesandFish,2nded.EPA841-B-99-002,OfficeofWater,US EnvironmentalProtectionAgency,Washington,DC,p.339.
Bedford,B.,Preston,E.,1988.Developingthescientificbasisforassessing cumu-lativeeffectsofwetlandlossanddegradationonlandscapefunctions:status, perspectivesandprospects.Environ.Manage.12,751–771.
Bonada,N.,Prat,N.,Resh,V.H.,Statzner,B.,2006.Developmentsinaquaticinsect biomonitoring:acomparativeanalysisofrecentapproaches.Annu.Rev. Ento-mol.51,495–523.
Boulton,A.J.,1999.Anoverviewofriverhealthassessment:philosophies,practice, problemsandprognosis.Freshw.Biol.41,469–479.
Boyero,L.,Ramirez,A.,Dudgeon,D.,Pearson,R.G.,2009.Aretropicalstreamsreally different?J.N.Am.Benthol.Soc.28,397–403.
Brooks,A.J.,Haeusler,T.,Reinfelds,I.,Williams,S.,2005.Hydraulicmicrohabitats andthedistributionofmacroinvertebrateassemblagesinriffles.Freshw.Biol. 50,331–344.
Brown,L.R.,Cuffney,T.F.,Coles,J.F.,Fitzpatrick,F.,McMahon,G.,Steuer,J.,Bell,A.H., May,J.T.,2009.UrbanstreamsacrosstheUSA:lessonslearnedfromstudiesin 9metropolitanareas.J.N.Am.Benthol.Soc.28,1051–1069.
Bryce,S.A.,Larsen,D.P.,Hughes,R.M.,Kaufmann,P.R.,1999.Assessingrelativerisks toaquaticecosystems:aMid-Appalachiancasestudy.J.Am.WaterRes.Ass.35, 23–36.
Camara,G.,Cartaxo,R.,Souza,M.,Freitas,U.M.,Garrido,J.,1996.SPRING:integrating remotesensingandGISbyobject-orienteddatamodelling.J.Comput.Graph.20, 395–403.
Clarke,R.T.,Wright,J.F.,Furse,M.T.,2003.RIVPACSmodels forpredictingthe expectedmacroinvertebratefaunaandassessingtheecologicalqualityofrivers. Ecol.Model.160,219–233.
Collier,K.J.,Haigh,A.,Kelly,J.,2007.CouplingGISandmultivariateapproachesto referencesiteselectionforwadeablestreammonitoring.Environ.Monit.Assess. 127,29–45.
Danz,N.P.,Niemi,G.J.,Regal,R.R.,Hollenhorst,T.,Johnson,L.B.,Hanowski,J.,Axler, R.,Ciborowski,J.J.H.,Hrabik,T.,Brady,V.J.,Kelly,J.R.,Brazner,J.C.,Howe,R.W., Host,G.E.,2007.IntegratedmeasuresofanthropogenicstressintheU.S.Great Lakesbasin.Environ.Manage.39,631–647.
Death,R.G.,Joy,M.K.,2004.Invertebratecommunitystructureinstreamsofthe Manawatu–Wanganuiregion,NewZealand:therolesofcatchmentversusreach scaleinfluences.Freshw.Biol.49,982–997.
Dovciak,A.,Perry,J.A.,2002.Insearchofeffectivescalesforstreammanagement: doesagroecoregion,watershed,ortheirintersectionbestexplainthevariance instreammacroinvertebratecommunities?Environ.Manage.30,365–377. Downes,B.J.,Hindell,J.S.,Bond,N.R.,2000.What’sinasite?Variationinlotic
macroinvertebratedensityanddiversityinaspatiallyreplicatedexperiment. Aust.J.Ecol.25,128–139.
Feld,C.K.,Hering,D.,2007.Communitystructureorfunction:effectsof environmen-talstressonbenthicmacroinvertebratesatdifferentspatialscales.Freshw.Biol. 52,1380–1399.
Fernández,H.R.,Domínguez,E.,2001.Guiaparaladeterminacióndelosartrópodos bentônicossudamericanos.UniversidadNacionaldeTucumán,Tucumán. Finn,D.S.,Poff,N.L.,2005.Variabilityandconvergenceinbenthiccommunitiesalong
thelongitudinalgradientsoffourphysicallysimilarRockyMountainstreams. Freshw.Biol.50,243–261.
Freckleton,R.P.,2002.Onthemisuseofresidualsinecology:regressionofresiduals vs.multipleregression.J.Anim.Ecol.71,542–545.
Gerth,W.J.,Herlihy,A.T.,2006.Effectofsamplingdifferenthabitattypesinregional macroinvertebratebioassessmentsurveys.J.N.Am.Benthol.Soc.25,501–512. Gucker,B.,Boechat,I.G.,Giani,A.,2009.Impactsofagriculturallanduseonecosystem structureandwhole-streammetabolismoftropicalCerradostreams.Freshw. Biol.54,2069–2085.
Gustafsson,L.,Ahlén,I.,1996.GeographyofPlantsandAnimals.Almqvistand Wik-sellInternational,Stockholm.
Gerritsen,J.,Barbour,M.T.,King,K.,2000.Apples,oranges,andecoregions:on deter-miningpatterninaquaticassemblages.J.N.Am.Benthol.Soc.19,487–496. HarrellJr.,F.E.,2001.RegressionModelingStrategies,withApplicationstoLinear
Models,LogisticRegression,andSurvivalAnalysis.Springer-Verlag,NewYork. Harris, G.P., Heathwaite, A.L., 2011. Why is achieving good ecologi-cal outcomes in rivers so difficult? Freshw. Biol, http://dx.doi.org/10. 1111/j.1365-2427.2011.02640.x.
Hawkins,C.P.,Olson,J.R.,Hill,R.A.,2010.Thereferencecondition:predicting bench-marksforecologicalandwater-qualityassessments.J.N.Am.Benthol.Soc.29, 312–343.
Hering,D.,Feld,C.K.,Moog,O.,Ofenbock,T.,2006.Cookbookforthedevelopment ofamultimetricindexforbiologicalconditionofaquaticecosystems: expe-riencesfromtheEuropeanAQEMandSTARprojectsandrelatedinitiatives. Hydrobiologia566,311–342.
Herlihy,A.T.,Paulsen,S.G.,VanSickle,J.,Stoddard,J.L.,Hawkins,C.P.,Yuan,L.L.,2008. Strivingforconsistencyinanationalassessment:thechallengesofapplyinga reference-conditionapproachatacontinentalscale.J.N.Am.Benthol.Soc.27, 860–877.
Hughes,R.M.,1985.Useofwatershedcharacteristicstoselectcontrolstreamsfor estimatingeffectsofmetalminingwastesonextensivelydisturbedstreams. Environ.Manage.9,253–262.
Hughes,R.M.,Larsen,D.P.,Omernik,J.M.,1986.Regionalreferencesites:amethod forassessingstreampotentials.Environ.Manage.10,629–635.
Hughes,R.M.,Whittier,T.R.,Rohm,C.M.,Larsen,D.P.,1990.Aregionalframework forestablishingrecoverycriteria.Environ.Manage.14,673–683.
Hughes,R.M.,1995.Definingacceptablebiologicalstatusbycomparingwith ref-erenceconditions.In:Davis,W.,Simon,T.(Eds.),BiologicalAssessmentand Criteria:ToolsforWaterResourcePlanningandDecisionMaking.Lewis Pub-lishers,BocaRaton,FL.
Hughes,R.M.,Peck,D.V.,2008.Acquiringdataforlargeaquaticresourcesurveys: theartofcompromiseamongscience,logistics,andreality.J.N.Am.Benthol. Soc.27,837–859.
Hughes,R.M.,Herlihy, A.T.,Kaufmann,P.R.,2010.Anevaluationofqualitative indexesofphysicalhabitatappliedtoagriculturalstreamsintenU.S.states. J.Am.WaterRes.Assoc.46,792–806.
Hughes,R.M.,Kaufmann,P.R.,Weber,M.H.,2011.Nationalandregionalcomparisons betweenStrahlerorderandstreamsize.J.N.Am.Benthol.Soc.30,103–121. Hynes,H.B.N.,1975.Thestreamanditsvalley.Verh.Int.Verein.Theor.Ang.Limnol.
19,1–15.
Kail,J.,Hering,D.,2009.Theinfluenceofadjacentstreamreachesonthelocal ecologicalstatusofCentralEuropeanmountainstreams.RiverRes.Appl.25, 537–550.
Kail,J.,Arleb,J.,Jähnig,S.C.,2012.Limitingfactorsandthresholdsfor macroinverte-brateassemblagesinEuropeanrivers:empiricalevidencefromthreedatasets onwaterquality,catchmenturbanization,andriverrestoration.Ecol.Indic.18, 63–72.
Karr,J.R.,Dudley,D.R.,1981.Ecologicalperspectiveonwaterqualitygoals.Environ. Manage.5,55–68.
Karr,J.R.,1991.Biologicalintegrity:along-neglectedaspectofwaterresource man-agement.Ecol.Appl.1,66–84.
Karr,J.R.,1999.Definingandmeasuringriverhealth.Freshw.Biol.41,221–234. Karr,J.R.,Chu,E.W.,1999.RestoringLifeinRunningWaters:BetterBiological
Mon-itoring.IslandPress,Washington,DC.
Kaufmann,P.R.,Levine,P.,Robison,E.G.,Seeliger,C.,Peck,D.V.,1999.Quantifying PhysicalHabitatinWadeableStreams.EPA/620/R-99/003.U.S.Environmental ProtectionAgency,Washington,DC.
Kaufmann,P.R.,Hughes,R.M.,2006.Geomorphicandanthropogenicinfluenceson fishandamphibiansinPacificNorthwestcoastalstreams.In:Hughes,R.M., Wang,L.,Seelbach,P.W.(Eds.),LandscapeInfluencesonStreamHabitatand Bio-logicalAssemblages,Symposium48.AmericanFisheriesSociety,pp.429–455. King,R.S.,Baker,M.E.,Whigham,D.F.,Weller,D.E.,Jordan,T.E.,Kazyak,P.F.,Hurd,
M.K.,2005.Spatialconsiderationsforlinkingwatershedlandcovertoecological indicatorsinstreams.Ecol.Appl.15,137–152.
Klemm,D.J.,Blocksom,K.A.,Fulk,F.A.,Herlihy,A.T.,Hughes,R.M.,Kaufmann,P.R., Peck,D.V.,Stoddard,J.L.,Thoeny,W.T.,2003.Developmentandevaluationof amacroinvertebratebioticintegrityindex(MBII)forregionallyassessing Mid-AtlanticHighlandsstreams.Environ.Manage.31,656–669.
LeCraw,R.M.,Mackereth,R.,2010.Sourcesofsmall-scalevariationinheadwater streaminvertebratecommunities.Freshw.Biol.55,1219–1233.
Ligeiro,R.,Melo,A.S.,Callisto,M.,2010.Spatialscaleandthediversityof macroin-vertebratesinaneotropicalcatchment.Freshw.Biol.55,424–435.
Maddock,I.,1999.Theimportanceofphysicalhabitatassessmentforevaluating riverhealth.Freshw.Biol.41,373–391.
Marchant,R.,Norris,R.H.,Milligan,A.,2006.Evaluationandapplicationofmethods forbiologicalassessmentofstreams:summaryofpapers.Hydrobiologia572, 1–7.
Mebane,C.A.,Maret,T.R.,Hughes,R.M.,2003.Anindexofbiologicalintegrity(IBI) forPacificNorthwestrivers.Trans.Am.Fish.Soc.132,239–261.
Miserendino,M.L.,Casaux,R.,Archangelsky,M.,DiPrinzio,C.Y.,Brand,C.,Kutschker, A.M.,2011.Assessingland-useeffectsonwaterquality,in-streamhabitat, ripar-ianecosystemsandbiodiversityinPatagoniannorthweststreams.Sci.Total Environ.409,612–624.
Moreno,J.L.,Navarro,C.,LasHeras,J.D.,2006.Abioticecotypesinsouth-central Spanish rivers: reference conditions and pollution. Environ. Pollut. 143, 388–396.
Moreno,P.,Franc¸a,J.S.,Ferreira,W.R.,Paz,A.D.,Monteiro,I.,Callisto,M.,2009.Use oftheBEASTmodelforbiomonitoringwaterqualityinaneotropicalbasin. Hydrobiologia630,231–242.
Moya,N.,Hughes,R.M.,Dominguez,E.,Gibon,F.M.,Goita,E.,Oberdorff,T.,2011. Macroinvertebrate-basedmultimetricpredictive models formeasuring the bioticconditionofBolivianstreams.Ecol.Indic.11,840–847.
Mugnai,R.,Nessimian,J.L.,Baptista,D.F.,2010.ManualdeIdentificac¸ãode Macroin-vertebradosAquáticosdoEstadodoRiodeJaneiro.TechnicalBooks,Riode Janeiro.
Myers,N.,Mittermeier,R.A.,Mittermeier,C.G.,Fonseca,G.A.B.,Kent,J.,2000. Biodi-versityhotspotsforconservationpriorities.Nature403,853–858.
Nijboer,R.C.,Johnson,R.K.,Verdonschot,P.F.M.,Sommerhäuser,M.,Buffagni,A., 2004.EstablishingreferenceconditionsforEuropeanstreams.Hydrobiologia 516,93–107.
Norris,R.H.,Thoms,M.C.,1999.Whatisriverhealth?Freshw.Biol.41,197–211. Oberdorff,T.,Pont,D.,Hugueny,B.,Porcher,J.P.,2002.Developmentandvalidation
ofafish-basedindex(FBI)fortheassessmentof“riverhealth”inFrance.Freshw. Biol.47,1720–1735.
Oliveira,R.B.S.,Mugnai,R.,Castro,C.M.,Baptista,D.F.,Hughes,R.M.,2011.Towards arapidbioassessmentprotocolforwadeablestreamsinBrazil:development ofamultimetricindexbasedonbenthicmacroinvertebrates.Ecol.Indic.11, 1584–1593.
Olsen,A.R.,Peck,D.V.,2008.SurveydesignandextentestimatesfortheWadeable StreamsAssessment.J.N.Am.Benthol.Soc.27,822–836.
Olson,D.M., Dinerstein,E.,Wikramanayake, E.D.,Burgess,N.D., Powell,G.V.N., Underwood,E.C.,D’Amico,J.A.,Itoua,I.,Strand,H.E.,Morrison,J.C.,Loucks,C.J., Allnutt,T.F.,Ricketts,T.H.,Kura,Y.,Lamoreux,J.F.,Wettengel,W.W.,Hedao,P., Kassem,K.R.,2001.Terrestrialecoregionsoftheworld:anewmapoflifeon Earth.Bioscience51,933–938.
Omernik,J.M.,1995.Ecoregions:aspatialframeworkforenvironmental manage-ment.In:Davis,W.,Simon,T.P.(Eds.),BiologicalAssessmentandCriteria:Tools forWaterResourcePlanningandDecisionMaking.LewisPublishing,BocaRaton, FL.
Parsons,M.,Norris,R.H.,1996.Theeffectofhabitatspecificsamplingonbiological assessmentofwaterqualityusingapredictivemodel.Freshw.Biol.36,419–434. Paulsen,S.G.,Mayio,A.,Peck,D.V.,Stoddard,J.L.,Tarquinio,E.,Holdsworth,S.M., VanSickle,J.,Yuan,L.L.,Hawkins,C.P.,Herlihy,A.T.,Kaufmann,P.R.,Barbour, M.T.,Larsen,D.P.,Olsen,A.R.,2008.ConditionofstreamecosystemsintheUS: anoverviewofthefirstnationalassessment.J.N.Am.Benthol.Soc.27,812–821. Peck,D.V.,Herlihy,A.T.,Hill,B.H.,Hughes,R.M.,Kaufmann,P.R.,Klemm,D.J., Lazor-chak,J.M.,McCormick,F.H.,Peterson,S.A.,Ringold,P.L.,Magee,T.,Cappaert,M.R., 2006.EnvironmentalMonitoringandAssessmentProgram– SurfaceWaters WesternPilotStudy:FieldOperations ManualforWadeableStreams.EPA 600/R-06/003.U.S.EnvironmentalProtectionAgency,OfficeofResearchand Development,Washington,DC.
Pérez,G.R.,1988.Guíaparaelestudiodelosmacroinvertebradosacuáticosdel DepartamentodeAntioquia.EditorialPresenciaLtda.,Bogotá,Colombia. Pinto,B.C.T.,Araújo,F.G.,Rodriguez,V.D.,Hughes,R.M.,2009.Localandecoregion
effectsonfishassemblagestructureintributariesoftheRioParaíbadoSul,Brazil. Freshw.Biol.54,2600–2615.
Pont,D.,Hugueny,B.,Beier,U.,Goffaux,D.,Melcher,A.,Noble,R.,Rogers,C.,Roset, N.,Schmutz,S.,2006.Assessingriverbioticconditionatthecontinentalscale:a Europeanapproachusingfunctionalmetricsandfishassemblages.J.Appl.Ecol. 43,70–80.
Pont,D.,Hughes,R.M.,Whittier,T.R.,Schmutz,S.,2009.Apredictiveindexofbiotic integritymodelforaquatic-vertebrateassemblagesofwesternU.S.streams. Trans.Am.Fish.Soc.138,292–305.
Poole,G.C.,2002.Fluviallandscapeecology:addressinguniquenesswithintheriver discontinuum.Freshw.Biol.47,641–660.
Rawer-Jost,C.,Zenker,A.,Böhmer,J.,2004.ReferenceconditionsofGermanstream typesanalysedandrevisedwithmacroinvertebrate fauna.Limnologica34, 390–397.
Reynoldson,T.B.,Bailey,R.C.,Day,K.E.,Norris,R.H.,1995.Biologicalguidelinesfor freshwatersedimentbasedonBEnthicAssessmentofSedimenT(theBEAST) usingamultivariateapproachforpredictingbiologicalstate.Aust.J.Ecol.20, 198–219.
Reynoldson,T.B.,Norris,R.H.,Resh,V.H.,Day,K.E.,Rosenberg,D.M.,1997.The ref-erencecondition:acomparisonofmultimetricandmultivariateapproachesto assesswater-qualityimpairmentusingbenthicmacroinvertebrates.J.N.Am. Benthol.Soc.16,833–852.
Rosenberg,D.M.,Resh,V.H.,1993.FreshwaterBiomonitoringandBenthic Macroin-vertebrates.ChapmanandHall,NewYork.
Sánchez-Montoya,M.M.,Puntí,T.,Suárez,M.L.,Vidal-Abarca,M.R.,Rieradevall,M., Poquet,J.M.,Zamora-Munõz,C.,Robles,S.,Álvarez,M.,Alba-Tercedor,J.,Toro, M.,Pujante,A.,Munné,A.,Prat,N.,2007.Concordancebetweenecotypesand macroinvertebrateassemblagesinMediterraneanstreams.Freshw.Biol.52, 2240–2255.
Sánchez-Montoya,M.M.,Vidal-Abarca,M.R.,Puntí,T.,Poquet,J.M.,Prat,N., Riera-devall,M.,Alba-Tercedor,J.,Zamora-Munõz,C.,Toro,M.,Robles,S.,Álvarez,M., Suárez,M.L.,2009.DefiningcriteriatoselectreferencesitesinMediterranean streams.Hydrobiologia619,39–54.
Sanderson,R.A.,Eyre,M.D.,Rushton,S.P.,2005.Theinfluenceofstreaminvertebrate compositionatneighbouringsitesonlocalassemblagecomposition.Freshw. Biol.50,221–231.
Schmera,D.,Er ˝os,T.,2004.Effectofriverbedmorphology,streamorderandseason onthestructuralandfunctionalattributesofcaddisflyassemblages.Ann.Limnol. Int.J.Limnol.40,193–200.
Sponseller,R.A.,Benfield,E.F.,Valett,H.M.,2001.Relationshipsbetweenlanduse, spatialscaleandstreammacroinvertebratecommunities.Freshw.Biol.46, 1409–1424.
Stevens,D.L.,Olsen,A.R.,2003.Varianceestimationforspatiallybalancedsamples ofenvironmentalresources.Environmetrics14,593–610.
Stoddard,J.L.,Peck,D.V.,Olsen,A.R.,Paulsen,S.G.,VanSickle,J.,Herlihy,A.T., Kauf-mann,P.R.,Hughes,R.M.,Whittier,T.R.,Lomnicky,G.,Larsen,D.P.,Peterson,S.A., Ringold,P.L.,2005.AnEcologicalAssessmentofWesternStreamsandRivers. U.S.EnvironmentalProtectionAgency,OregonStateUniversity,andDynamic Corporation,Corvallis,OR.
Stoddard,J.,Larsen,D.P.,Hawkins,C.P.,Johnson,R.K.,Norris,R.H.,2006.Setting expectationsfortheecologicalconditionofstreams:theconceptofreference condition.Ecol.Appl.16,1267–1276.
Stoddard,J.L.,Herlihy,A.T.,Peck,D.V.,Hughes,R.M.,Whittier,T.R.,Tarquinio,E., 2008.Aprocessforcreatingmulti-metricindicesforlarge-scaleaquaticsurveys. J.N.Am.Benthol.Soc.27,878–891.
Strahler,A.N.,1957.Quantitativeanalysisofwatershedgeomorphology.Trans.Am. Geophys.Union38,913–920.
Suriano,M.T.,Fonseca-Gessner,A.A.,Roque,F.O.,Froehlich,C.G.,2011.Choiceof macroinvertebratemetricstoevaluatestreamconditionsinAtlanticforest, Brazil.Environ.Monit.Assoc.175,87–101.
Tabachnick,B.G.,Fidell,L.S.,2007.UsingMultivariateStatistics,5thed.Pearson/Allyn &Bacon,Boston.
Tejerina-Garro,F.L.,deMérona,B.,Oberdorff,T.,Hugueny,B.,2006.Afish-based indexoflargeriverqualityforFrenchGuiana(SouthAmerica):methodand preliminaryresults.Aquat.LivingResour.19,31–46.
Trautwein,C.,Schinegger,R.,Schmutz,S.,2011.Cumulativeeffectsoflanduse onfishmetricsindifferenttypesofrunningwatersinAustria.Aquat.Sci.74, 329–341.
USGS(UnitedStatesGeologicalSurvey),2005.ShuttleRadarTopographyMission– SRTM.,http://www.srtm.usgs.gov
VanSickle,J.,Baker,J.,Herlihy,A.,Bayley,P.,Gregory,S.,Haggerty,P.,Ashkenas, L.,Li,J.,2004.Projectingthebiologicalconditionofstreamsunderalternative scenariosofhumanlanduse.Ecol.Appl.14,368–380.
Vannote,R.L.,Minshall,G.W.,Cummins,K.W.,Sedell,J.R.,Cushing,C.E.,1980.The rivercontinuumconcept.Can.J.Fish.Aquat.Sci.37,130–137.
Waite,I.R.,Herlihy,A.T.,Larsen,D.P.,Klemm,D.J.,2000.Comparingstrengthsof geo-graphicandnongeographicclassificationsofstreambenthicmacroinvertebrates intheMid-Atlantic.J.N.Am.Benthol.Soc.19,429–441.
Wang,L.,Brenden,T.,Seelbach,P.,Cooper,A.,Allan,D.,ClarkJr.,R.,Wiley,M.,2008. Landscapebasedidentificationofhumandisturbancegradientsandreference conditionsforMichiganstreams.Environ.Monit.Assess.141,1–17.
Wantzen,K.M.,2003.Cerradostreams–characteristicsofathreatened freshwa-terecosystemtypeonthetertiaryshieldsofSouthAmerica.Amazoniana17, 485–502.
Wantzen,K.M.,Siqueira,A.,NunesDaCunha,C.,Sá,M.F.P.,2006.Stream–valley systemsoftheBraziliancerrado:impactassessmentandconservationscheme. Aquat.Cons.:Mar.Freshw.Ecosyst.16,713–732.
Whittier,T.R.,Hughes,R.M.,Stoddard,J.L.,Lomnicky,G.A.,Peck,D.V.,Herlihy,A.T., 2007a.Astructuredapproachfordevelopingindicesofbioticintegrity:three examplesfromwesternUSAstreamsandrivers.Trans.Am.Fish.Soc.136, 718–735.
Whittier,T.R.,Stoddard,J.L.,Larsen,D.P.,Herlihy,A.T.,2007b.Selectingreference sitesforstreambiologicalassessments:bestprofessionaljudgmentorobjective criteria.J.N.Am.Benthol.Soc.26,349–360.
Wiens,J.A.,2002.Riverinelandscapes:takinglandscapeecologyintothewater. Freshw.Biol.47,501–515.
Yates,A.G.,Bailey,R.C.,2010.Selectingobjectivelydefinedreferencesitesforstream bioassessmentprograms.Environ.Monit.Assess.170,129–140.