You are on page 1of 12

Streamflow Regulation and Fish Community Structure

Author(s): Mark B. Bain, John T. Finn and Henry E. Booke


Reviewed work(s):
Source: Ecology, Vol. 69, No. 2 (Apr., 1988), pp. 382-392
Published by: Ecological Society of America
Stable URL: http://www.jstor.org/stable/1940436 .
Accessed: 16/02/2013 17:50

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Ecological Society of America is collaborating with JSTOR to digitize, preserve and extend access to Ecology.

http://www.jstor.org

This content downloaded on Sat, 16 Feb 2013 17:50:12 PM


All use subject to JSTOR Terms and Conditions
Ecology, 69(2), 1988, pp. 382-392
? 1988 by the Ecological Society of America

STREAMFLOW REGULATION AND FISH


COMMUNITY STRUCTURE'

MARK B. BAIN2 AND JOHN T. FINN


Department
ofForestry
and Wildlife
Management,University
ofMassachusetts,
Amherst,
Massachusetts
01003 USA

AND

HENRY E. BooKE
Massachusetts
Cooperative
Fishery
ResearchUnit,
3 University
ofMassachusetts,
Amherst, Massachusetts
01003 USA

Abstract. Many regulatedstreamsare characterizedby highlyvariable and unpredict-


able flowregimes.Since changes in streamflowdirectlymodifyphysicalhabitat,streams
withsuch highlyvariable flowsprovide highlyunstableaquatic habitats.We evaluated the
effectof artificialstreamflowfluctuationson streamfishcommunitiesby comparingfish
densities,in species and habitatgroups,betweentwo riversdiffering in daily flowregime:
one witha naturalflow,and one withhighlyregulatedflows.
We developed a simple model describingthe relationshipbetween available stream
habitatand its use by 15 species or size classes of fishin the naturalriver.Species and size
classes that used a specificset of microhabitatconditions were identifiedby comparing
habitatcharacteristicsforsamples withand withouteach typeof fish;forfishthat used a
particulartypeof microhabitat,we groupedspecies and size classes accordingto similarity
in microhabitatuse. Next, we categorizedstreamhabitatsamples in both the naturaland
regulatedriversinto groupson the basis of fishhabitatuse criteria.Fish densitiesforeach
fishand habitatgroupwere thenindividuallycompared betweenthe two rivers.
An abundant (>90% of all fish)and diverse (nine species) group of small-fishspecies
and size classes wererestrictedto microhabitatthatwas characterizedas shallow in depth,
slow in currentvelocity,and concentratedalong streammargins.This group of fishwas
reduced in abundance in the regulatedriverand absent at the studysite withthe greatest
flow fluctuation.Anotherfishgroup included species and size classes that used eithera
broad rangeof habitator a microhabitatthatwas deep, fast,or both,and was concentrated
in midstreamareas. The densityof fishin thisgroupwas higherin the regulatedriverand
peaked at the siteswiththegreatestfluctuationsin flow.Highlyvariable and unpredictable
flowregimesappear to be a high-frequency disturbancethateffects fishdifferentlydepending
on the way theyuse streamhabitatand acts to reduce communitycomplexity.
Keywords: community ecology;disturbance;environmental multi-
microhabitat;
variabilityfishes;
variateanalysis;streamflow
regulation;streamhabitat.

INTRODUCTION a changein streamflow(total streamdischarge)trans-


Flow modificationis one of the most widespread lates into a change in the water depth and velocity.
human disturbancesof stream environments(Fraser Also, as dischargechanges,streamsubstrateand cover
objectsbecome associated withdifferentcombinations
1972, Ward and Stanford 1983), and its effectsare
of waterdepthand velocity.Consequently,changesin
poorlyunderstood(Holden 1979, Sale 1985). The flow
streamflowcan be regarded as modificationsin the
regimesin manyregulatedstreamsand riversare high-
ly variable and unpredictable;typical seasonal flow physicalcomposition of the aquatic habitat. In addi-
variations may occur weekly or even daily, with no tion, changes in the patternof flow variability(flow
For anyspecificlocationin a stream, regime)can be viewed as an artificialdisturbanceof
regularperiodicity.
streamhabitatstability.
Zonation, or downstreamchangein fishcommunity
' Manuscript received 20 October 1986; revised 15 June
1987; accepted 17 July1987. compositionwithinriversystems,has been associated
2 Presentaddress: Alabama Cooperative Fish and Wildlife withgradientsof habitatchange (Kuehne 1962, Shel-
Research Unit, 331 Funchess Hall, Auburn University,Au- don 1968, Hynes 1970,Hawks 1975, Barilaet al. 198 1).
burn,Alabama 36849-4201 USA. In streamsof similar size, physicalhabitat character-
3Cooperators: United States Fish and Wildlife Service,
istics(waterdepth,currentvelocity,substrate)are im-
MassachusettsDivision of Fisheriesand Wildlife,Massachu-
settsDivision ofMarine Fisheries,and theUniversityofMas- portantfactorsinfluencingfishcommunitycomposi-
sachusetts. tion (Gorman and Karr 1978, Moreau and Legendre

This content downloaded on Sat, 16 Feb 2013 17:50:12 PM


All use subject to JSTOR Terms and Conditions
April 1988 FISH COMMUNITIES IN REGULATED STREAMS 383

1979, Savard and Moreau 1982, Schlosser 1982a, b, The same fishspecies occurredin bothofthetwo study
Moyle and Vondracek 1985). Studies of the role of streams or in the nearby section of the Connecticut
habitatstabilityin determiningfishcommunitystruc- River,withthreeexceptions:browntrout(Salmo trut-
ture have affirmedthe significanceof both seasonal ta) and rainbow trout(S. gairdneri)were stocked an-
(Kushlan 1976, Schlosser 1985) and long-term(Ho- nuallyonlyin the DeerfieldRiver,and parrofAtlantic
rowitz 1978) steamflowvariability.In general,stream salmon (S. salar) were stocked only in the Deerfield
fishcommunitystructure is stronglyinfluencedby both River in 1983. A detailed comparison of the physical
habitat composition and stability,where habitat sta- conditions,chemistry,and fishfauna of the two rivers
bilityis a functionof flowregime. and theirdrainagebasins was given by Bain (1985).
We compared the fishcommunityand habitat re- Althoughboth streamshad essentiallythe same an-
lationshipsin two rivers:one withartificialshort-term nual flowregime(bimodal; dischargepeaks in spring
flowfluctuationsand one with a natural flowregime. and fall)and responseto basin runoffevents,the study
The comparisonwas made on thebasis offishdensities reachescontrastedsharplyin dailyflowvariability.Flow
by species and habitatgroups.This approach enabled fluctuations on theDeerfieldRiver occurredfrequently
theidentification ofdifferencesamong streamfishden- (several times daily) and varied greatlyin amplitude
sitiesthataccounted forboth the directeffectsof flow (Fig. 1). They were caused by unpredictableand vari-
variability(i.e., habitat variability)and the way dif- able-volume water releases (a response to short-term
ferentfishused streamhabitat. Our work contributes electricalsupplyand demand in New England) froma
to researchon the effectsof artificialdisturbanceson series of fourhydroelectricdams (?21 m high) with
structureand functionof natural biological systems shallow,short(?2 km) impoundments.Flows fluctu-
(the "disturbancephenomena" of Pickettand White ate throughoutthe year except in springand briefly
1985). In one of our rivers,where disturbancesof a afterheavy rainfallor snowmelt,when riverdischarge
natural magnitudeoccurredon an artificialtemporal exceeds hydroelectricplant capacities. The flowfluc-
scale, we used community-levelstructuralcharacter- tuationsare typicallymost pronouncedduringthe dry
isticsto detectfishresponses. period of summerand early fall when no wateris re-
leased duringthe frequent3-h shutdownswhen the
STUDY AREAS
impoundmentsare being refilled.During these shut-
This studywas conductedon the lower segmentsof downs, dam leakage and tributariesmaintaina small
two fifth-order tributariesof the ConnecticutRiver. flowin the studyreach. The severityof flowfluctua-
The West River, in southernVermont,is an unregu- tions becomes attenuateddownstreamfrom the hy-
lated streamwitha mean annual dischargeof 29 m3/s droelectricfacilitiesbecause an abruptrise and fallat
and a basin area of 1092 km2at the studyreach (2.8- the downstreamdam (km 21.2) becomes a relatively
10.5 km upstreamfromthe mouth [42?52' N, 72033' gradual change at the end of the studyreach (Fig. 1).
W]). The WestRiverwas selectedas a "natural" stream Consequently,there is a gradientof decreasing flow
typical of Connecticut River tributariesin physical variabilityin the study reach. Physicochemicalcon-
structureand fishspecies composition.The Deerfield ditions do not vary substantiallyin response to flow
River, in westernMassachusetts,has an artificialflow changes (Bain 1985) since impoundmentshold little
regimewith a mean annual dischargeof 41 m3/sand waterand are regularlydrained and refilled.
a basin area of 1407 km2at the studyreach (13.2-20.2 The West River has a naturaldaily flowregimethat
km fromthe mouth [42034' N, 72?35' W]); short-term appears essentiallyconstantwhen compared withthat
flowregulationhas been extremeforat least 50 yr. in theDeerfieldRiver. In theabsence ofa runoff event,
The West and Deerfieldrivers drain adjacent wa- thereis almost no detectablechangein riverdischarge
tershedsof similar climate and geology (Brooks and over a 24-h period (Fig. 1). When runoffevents do
Deevey 1963). Althoughthe two studystreamsdiffer affectriverdischarge,the rate of streamflowchange is
in mean annual dischargeand drainagearea, theyare much more gradual than the artificialflowchanges in
similarin generalsize and appearance and fall in the the DeerfieldRiver.
medium-size river categoryof the river continuum
framework ofVannote et al. (1980). Both studyreaches METHODS
have a highgradient(3 m/km),primarilycoarse sub-
Field sampling
strate(bedrock, boulder, cobble), and littleinstream
cover such as logs and debris. Water in both streams Four studysites(81-231 m long)wereselectedalong
is highlyoxygenated,soft(10-27 mg/Ltotal hardness, each studyreachin each streamto representriverhab-
30-9 5,gSconductivity),and warm(maximumsummer itatin thevicinity.At each site,six permanenttransects
temperature27QC). Both studyreaches were immedi- were selected to proportionatelysample the general
ately upstream of a transitionfrom a high-gradient within-sitehabitat(e.g.,riffles,pool). On each transect,
streamto a low-gradientfloodplainriver(Connecticut fourevenlyspaced sample locations(23 M2) wereiden-
River) characterizedby low-velocitychannelswithfine tifiedbeforesamplingbegan:one adjacentto each bank,
substrate,backwaters,and abundant instreamcover. one-thirdacross thetransect,and two-thirdsacross the

This content downloaded on Sat, 16 Feb 2013 17:50:12 PM


All use subject to JSTOR Terms and Conditions
384 MARK B. BAIN ET AL. Ecology,Vol. 69, No. 2

8
WEST RIVER (km 20.2)

4_

01 I , l - I I

32 DEERFIELD RIVER (km 21.2)

E 12

2
32 _a
DEERFIELD RIVER(km14.8)

22 -

12-

21 A
2400 1200 2400 1200 2400 1200 2400

6 AUG 7 AUG 8 AUG

AUGUST 1983
FIG. 1. A representative3-d recordof streamdischargeforthetwostudyrivers.The WestRivercurve(toppanel)is
basedon data froma UnitedStatesGeologicalSurveystreamgagelocatedupstream fromthestudyreach.The WestRiver
was smallerat thislocationthanat thestudyreach,buttheflowpattern was thesame.The DeerfieldRivercurveswere
constructedfromhydroelectric generationrecordsofthedam upstreamfromthestudyreach(middlepanel)and data from
a streamgagein thelowerportion(bottompanel)ofthestudyreach.The largerwaterreleasein theDeerfield Riveron 8
August1983corresponded to a heavygenerationperiodthatoccurred
at irregular
intervals
duringmostweeks.WestRiver
dischargeremainednearlyconstantduringtheperiod,sincerainrunoff had no significant
effects
on eitherriver.River
kilometresin parenthesesindicatethedistancefromtherivermouthto thedischarge measurement point.

transect.Althoughthe sampling design was not ran- cornersand then leftundisturbedforat least 15 min.
domized, potential investigatorbias was reduced by In collectingfish,two personswearingwaders and car-
this presitingof samples. ryingvarious dip nets approached the frame from
The 24 sampling locations at each study site were downstreamto a pointat whichthe frameinteriorwas
sampled between midmorningand late afternoonon visible.A thirdpersonthenactivatedthepowersupply
days of typical seasonal flow conditions. Periods of froma remotelocation.All fishoriginating fromwithin
very large water releases in the DeerfieldRiver were the framewere collectedand the interiorof the frame
avoided to maintain samplingconsistency.The eight was thoroughlyinspectedat least twice. All fish >20
studysiteswere sampled once duringeach of fourpe- mm total lengthwere identifiedand measured.
riods: 6-22 October 1982 and 1983, 30 June-30 July When water exceeded wader depth (0.9 m) we vi-
1983, and 14-23 August1983, and thefourWestRiver sually counted fishin a 23-iM2 area. Wearing a face
sites were also sampled once between21 Julyand 15 mask and lyingon a surfmattress,one of us (M. B.
August 1982. Bain) positionedhimselfovera samplinglocationwhile
At each samplinglocation, we collected fishin the attachedto a steelcable suspendedacross theriver.He
23-iM2 area, eitherby using a propositionedarea elec- countedthefishbyspeciesand estimatedlengths,which
trofishing device (described and evaluated by Bain et werelateradjustedby usinga regressionequation (r2=
al. 1985a) or, in deep-waterareas, by snorkeling(11% 0.98, 28 pairsofknownand estimatedlengths).During
of total samples). The electrofishingdevice consisted all snorkeling,visibilitywas more thanadequate to see
of an electrodeframe(7.6 x 3.0 m) poweredby a 460- thestreambottomclearly(maximumwaterdepth,230
volt alternating-currentgenerator.The electrodeframe cm). Althoughwe did not compare electrofishing and
was positionedby manipulatingropes attachedto the snorkeling,otherinvestigatorshave shown that,when

This content downloaded on Sat, 16 Feb 2013 17:50:12 PM


All use subject to JSTOR Terms and Conditions
April 1988 FISH COMMUNITIES IN REGULATED STREAMS 385

visibilityis good, visual countsyieldresultsconsistent 1982, Finger 1982, Orth and Maughan 1982, Moyle
with those obtained by other sampling methods and Vondracek 1985), we divided some species into
(Northcoteand Wilkie 1963, Goldstein 1978, Sale and size classes by inspectingall possible bivariateplots of
Douglas 1981), especially when discrete areas are habitatvariables vs. fishlength.
searched (Keast and Harker 1977). First,we determinedwhichfishspeciesor size classes
Four habitat variables were measured at the center used a specificset of microhabitatconditions,usinga
ofeach 23-iM2samplinglocationafterthefishhad been multivariateanalysis of variance (MANOVA) test to
sampled. Waterdepthwas measuredto thecentimetre detectdifferences in habitatcompositionbetweentwo
with a weightedfiberglasstape measure. Surfacecur- groupsofsamples:thosewitha particularspecies(pres-
rent velocity was measured by timingthe float of a ent) and those withoutthatspecies (absent). However,
wood chip (38 x 19 x 19 mm) attached to a 1-m such a testposes an importantproblem; the presence
lengthof nylon twine. Surface currentvelocitymea- of a species in a sample indicatessuitable habitat,but
surementswereconverted(log/logregressionequation its absence does not necessarilyimplyunsuitablehab-
[r2 = 0.81], 119 pairs of measurementswitha surface itat. It seems reasonable to expect that not every 23-
wood chip and a Price AA currentmeter set at 60% m2 sample of suitable habitat will be occupied at the
depth) to mean velocity(velocityat 60% of depth),a time of sampling.Consequently,a species that uses a
formthathas oftenbeen reportedin otherstreamhab- specificmicrohabitatwill be found in a less variable
itat studies (Probst et al. 1984, Glova and Duncan set of conditions than that of the available habitat.
1985). Two streamsubstratevariables,coarsenessand While MANOVA is conservativewhen one group(ab-
heterogeneity, were quantifiedwith a 2-m lead-core sentsamples) has both a largesample size and disper-
rope divided into 20 10-cm sections(method of Bain sion (Green 1979), we checked all test resultsby dis-
et al. 1985b) by the same personthroughoutthe study. criminantanalysis group classification(as in Rice et
The 20-segmentrope was stretchedacross a sampling al. 1983) and logisticregression(Fienberg1980). Only
location (perpendicularto the current),and the dom- the MANOVA resultsare reportedbecause theywere
inant substratein each 10-cm section was recorded supportedby the alternativestatisticalproceduresin
visually as one of the followingtypes(modifiedfrom all tests.
Cummins 1962): 1 = smooth horizontalsurfacessuch Next we formedgroups of species and size classes
as bedrock,sand, or silt; 2 = gravel (2-16 mm diam- that used similar microhabitatsbased on mean posi-
eter);3 = pebble (17-64 mm); 4 = cobble (6 5-256 mm); tions (habitatspace centroids)in the available habitat
5 = boulder (>256 mm); 6 = irregular bedrock. The space. Correlationsamong the fourhabitat variables
mean of the 20 dominantsubstratevalues was used as were removed by principal components transforma-
an index of substratecoarseness. The standarddevia- tionofthehabitatdata because thistechniqueprovides
tion was used as an index of substrateheterogeneity. an objectiveformatforinterpreting species-habitatuse
patterns(Rotenberry and Wiens 1980,Carnesand Slade
Identificationoffishcommunityand 1982). Groups were identifiedby usinggroupaverage
habitatrelationships clusteranalysis of species centroidsin all fourtrans-
We grouped fishand habitat to produce a simple formeddimensions.
model describingthe relationshipbetween the fish Our finalstepin describinga fishcommunity-habitat
communityand available streamhabitat.The purpose model was to categorizehabitat on the basis of fish
of this dual categorizationapproach was to enable us grouppresenceor absence by directdiscriminantanal-
to make comparisons between the two riverson the ysis (Nie et al. 1975). Using West River data, we de-
basis of fishdensitieswithinhabitatcategories.To de- veloped a discriminantfunctionand then classified
velop the fish community-habitatmodel, we con- samples fromboth rivers.Althoughthis approach ap-
ducted three separate sequential analyses to answer pears circular,we wanted to compare quantitatively
threequestions: Which fishused a specificmicrohab- fishdensitiesbetweenriversand among sitesin a man-
itat? Can the various typesof fishbe grouped on the nerthataccountsfordifferences in habitatavailability
basis of similarityin microhabitatuse? Can the sam- and use ratherthan analyzing differencesby species
ples of streamhabitatbe groupedto reflectthe typeof along.
fishthat use the habitat?This approach is analogous
to categorizingstreamhabitatintopool and riffletypes Differencesbetweenriversand among sites
to analyze separatelythe abundance of pool and riffle Fish densities (number captured per 23 M2) were
fishes,except that we relied on statisticalresults to compared quantitativelybetween rivers and among
identifyfishand habitatcategories. sites, independentlyfor each fish-habitatcategory.
We used fishsampling and habitat data fromthe Densities were computed by studysite and sampling
West River to identifythe fishcommunity-habitat re- period foreach group of fishin each habitatcategory.
lationships,since it had a natural flow regime.Inas- Tests for differences between riversand among sites
much as microhabitatuse patternsrelatedto body size withinriverswere made by nonparametrictwo-way
have been reportedformany streamfishes(Bain et al. (riversand sites) layouttests(Kruskal-Wallis)because

This content downloaded on Sat, 16 Feb 2013 17:50:12 PM


All use subject to JSTOR Terms and Conditions
386 MARK B. BAIN ET AL. Ecology,Vol. 69, No. 2

therewere largedifferencesin groupvariances.A pos- WEST RIVER


teriorimultiple-rangetests,based on Kruskal-Wallis +4
+4~~~~~~~ DEEP
mean ranks,were used to identifydifferences among
sites withineach river. FINE I
RESULTS

During this study,we captureda total of 3039 fish


and coded names of Z
of 23 species (common, scientific, FAST
fishesare givenin theAppendix).Three species (small- 0
mouthbass, rockbass, whitesucker)weredivided into LRB

two lengthgroups(largeand small; size rangesin Ap- C.)~~~~F TD *+


SWS LMB
pendix)because theiruse ofhabitatwas relatedto body
size. Of the resulting26 species and size classes, 15
SLOW ~ ~ B *LND

were abundant enough to allow statisticalanalysis of Z S


habitatuse characteristics. P-.41
COARSE
Fish communityand habitatrelationships
C 4 PRINCIPA
SHALLOW
CMMSH N I+
A significantdifference(MANOVA, P < .05) was
foundin habitatcharacteristics(some combinationof
depth,velocity,and substrate)betweenfishpresentand -4
-4 +
fishabsent samples for 13 of the 15 fishspecies or size ~PRINCIPAL COMPONENT I
classes fromthe West River. These 13 species or size FIG. 2. Distributionof 13 microhabitatspecialists(habitat
classes werethereforeidentifiedas usingsome specific centroids)withinthe total available West River habitat (en-
set of the microhabitatconditionswithinthe available closed by heavy solid line). The extremesof the originalhab-
rangeof habitat conditions.We consideredthese fish itataxes primarilycomposingprincipalcomponentsI and II
are superimposedto assist in interpretation.The substrate
to be microhabitatspecialists.Because no differences
heterogeneity axis is not shown because it is the least impor-
(P > .10) in habitatcompositionwere foundbetween tantofthefouroriginalaxes and orientednearlyperpendicular
presentand absent samples forlargesmallmouthbass to the two-dimensionalplane illustratedhere.The nine types
or American eel, we identifiedthese fishas habitat offishidentifiedbyclusteranalysisas theshallow-slowhabitat
generalists.The analyses for bluegills and pumpkin- guild (SS GUILD) are enclosed by a lightsolid line. Because
in the clusteranalysis of centroidswe used substratehetero-
seeds werelimitedto thelate summerand fallsampling geneityas well as the three habitat variables shown, the
periodsand to the two sitesfarthestdownstreamsince relativeproximityofcentroidsis notfullyshownin thisfigure.
these fishwere presentin the lower West River only
duringthesetimes.All MANOVA testsincludedsam-
pling period as a factor(two-wayMANOVA) to test The fourfishspecies and size classes that were not
for seasonal differences.Because all seasonal and in- partof the SS GUILD (largerockbass, longnosedace,
teractiontermswerenonsignificant (P > .29), we pooled largemouthbass, largewhite sucker)used microhabi-
the data forall samplingperiods. tats otherthan the shallow, slow-watertype (Fig. 2).
Clusteranalysisoffishcentroidsforthe 13 specialist For example, largewhite suckersused a microhabitat
species and size classes indicatedthatnine used a sim- characterizedby verydeep and fairlyswiftwaterwith
ilartypeofmicrohabitat:small smallmouthbass, small relativelyfine substrate(i.e., at the heads of pools).
rock bass, bluegill,pumpkinseed,small whitesucker, American eel and smallmouthbass, the habitat gen-
blacknose dace, fallfish,mimic shiner,and tessellated eralists,werenotincludedin theclusteranalysisor Fig.
darter.To illustratetheclusteranalysisresults,we offer 2 because theyused habitatbroadlyand therefore over-
a two-dimensionalprincipalcomponentsplot (Fig. 2) lapped all otherspecies and size classes.
in which the available habitat is bounded, primary To group habitat,we classifiedthe 480 West River
originalaxes are superimposed,and fishcentroidsare samples by direct discriminantanalysis into habitat
plotted.The nine groupedfishused a restricted habitat used by SS GUILD fish(shallow and slow, SS HAB-
characterizedby very shallow and slow water, and ITAT) and habitatnot used by SS GUILD fish(deep
coarse substrate(i.e., the boulder- and cobble-strewn or fast,or both,DF HABITAT). We repeatedthisanal-
edge of the river).In addition, this group constituted ysis usinga stepwiseprocedureto identifythe habitat
an importantportion of the fishcommunitysince it variablesmostimportantin distinguishing betweenthe
was veryabundant(>90% ofall fishcaptured),diverse two groups. The stepwise entryorder,group means,
(9 of the 15 common species), and composed exclu- and discriminantfunctioncoefficientsindicated that
sivelyof small fish(see Appendix). The analysisto this waterdepthand currentvelocityweretheprimaryvari-
pointidentifiedan abundantand diversegroupofsmall ables distinguishingthe two typesof habitat(Table 1).
fishthat used shallow, slow microhabitat(hereinafter In summary,we simplifiedthefishcommunity-habitat
called the SS GUILD). relationshipsby placingboth fish(the SS GUILD and

This content downloaded on Sat, 16 Feb 2013 17:50:12 PM


All use subject to JSTOR Terms and Conditions
April 1988 FISH COMMUNITIES IN REGULATED STREAMS 387

TABLE 1. Means, discriminantfunctioncoefficients, and relative importanceof habitat variables for distinguishingWest
River habitatused (fishpresent)and not used (fishabsent) by SS GUILD fishes.

Group means Discriminant


function
Habitat variables* Guild present Guild absent coefficient Relative importance
Water depth (cm) 30.17 58.61 0.754 veryimportant
Currentvelocity(cm/s) 8.14 21.45 0.756 veryimportant
Substratecoarsenessindext 3.74 3.42 -0.371 less important
Substrateheterogeneityindex? 1.18 1.07 0.018 not important
* Habitat variables are listed in the orderselectedby stepwisediscriminantanalysis.
t Mean of 20 coded dominant substrateobservations.
? Standarddeviation of 20 coded dominant substrateobservations.

fishof otherspecies and size classes [non-SS GUILD]) observation)emptied into the downstreamend of a
and habitat (DF HABITAT and SS HABITAT) into pool partlyincludedin site 2. The highdensitiesof SS
two categories. GUILD fishat DeerfieldRiver site 2 may have been
The discriminantfunctiondeveloped withthe West largelydue to thisimmediatesourceofSS GUILD fish.
River data forclassifyinghabitat as SS HABITAT or For SS GUILD fishin DF HABITAT, there were
DF HABITAT was used to classifyDeerfield River greaterdensities(P = .039) in the West River than in
samples. Classificationresultswereinspectedby study the DeerfieldRiver, althoughthe densityof these fish
site and samplingperiod. In both rivers,slightlymore in thishabitatwas low in both rivers(Fig. 3B). There
samples were classifiedas SS HABITAT than as DF wereno differences (P ? .142) in densityof SS GUILD
HABITAT, althoughboth categorieswerealways well fishin DF HABITAT among siteson eitherriver.Since
representedat each site. the SS GUILD and DF HABITAT combinationis a
mismatch of fishand habitat type,we expected low
Comparisonsbetweenriversand among sites densitiesat least in the West River. However, the lack
For each studysite on each river,the densityof fish of significantlygreaterdensityof SS GUILD fishin the
was computed by type of fish (SS GUILD, non-SS DeerfieldRiver clearlyindicated that these fishwere
GUILD) and habitat(SS HABITAT, DF HABITAT). not shiftingto DF HABITAT. Such a shiftin habitat
Comparisons betweenriversand among sites on each use, had it been found,could have explained the re-
riverwere made independentlyforeach fishand hab- duced densityofSS GUILD fishin SS HABITAT iden-
itat combination.Samples (23 m2each) were counted tifiedearlier(Fig. 3A).
once foreach type of habitat,and each fishtype was The densityof otherspecies and size classes (non-
counted (numbersper 23-iM2 sample) separately. SS GUILD fish)in SS HABITAT (Fig. 3C) was greater
For SS GUILD fishin SS HABITAT (Fig. 3A), fish (P = .003) in the Deerfield River than in the West
densitywas greaterin theWest River thanin theDeer- River. There were no differences(P = .597) among
field River (P ' .001). There were differences(P = West River sites,althoughthereappeared to be mar-
.012) among West River sites; at site 4 (the farthest ginaldifferences (P = .091) amongDeerfieldRiversites.
upstream site) the densityof SS GUILD fishin SS In the DeerfieldRiver, the trendtoward highernon-
HABITAT was lowerthanat theotherthreesites.This SS GUILD densityin SS HABITAT at the sites with
resultwas largelyaccounted forby the movement of the greatestflow fluctuations(3 and 4; Fig. 3C) was
largenumbersof bluegillsand pumpkinseedsinto the primarilyattributableto American eels.
lower West River duringlate summerand fall.A sim- The densityofnon-SS GUILD fishin DF HABITAT
ilar migrationdid not occur in the Deerfield River, (Fig. 3D) was greater(P < .001) in the DeerfieldRiver
even thoughthese fishwere also presentin the lower thanin theWest River. There wereno differences (P =
sectionof thatriver.However, SS GUILD densityre- .105) among West River sites,althoughthereweredif-
mained higher(P = .001) in the West River than in ferences(P = .010) in theDeerfieldRiver. In theDeer-
the DeerfieldRiver even withthese species excluded. fieldRiver, the densitiesof non-SS GUILD fishwere
Thereweredifferences (P = .008) amongDeerfieldRiv- high at the most severelyflow-fluctuating upstream
er sites; the upstreamtwo sites with the greatestfluc- sites (3 and 4; Fig. 3D). The high densitiesat sites 3
tuationsin flow(sites 3 and 4; Fig. 3A) had fewerSS and 4 resultedprimarilyfroma combinationof rela-
GUILD fishin SS HABITAT than the downstream tivelylargenumbersof Americaneels, and largewhite
two sites. The SS GUILD fishwere rare at Deerfield suckers.
River site 3 and absent at site 4. In contrastto these In summary,the most notable resultswere compar-
sites,the densityof SS GUILD fish(primarilyfallfish) ativelyhighdensitiesof SS GUILD fishin SS HABI-
at site 2 in the DeerfieldRiver was similarto that in TAT in the West River and the eliminationof these
the West River (Fig. 3A). A large tributarywith high fishin the DeerfieldRiver at the site withthe greatest
densitiesof SS GUILD fish(M. B. Bain et al., personal fluctuationsin flow.The reduced densityand absence

This content downloaded on Sat, 16 Feb 2013 17:50:12 PM


All use subject to JSTOR Terms and Conditions
388 MARK B. BAIN ET AL. Ecology,Vol. 69, No. 2

of these fishin SS HABITAT in the DeerfieldRiver SS HABITAT DF HABITAT


could notbe accountedforbya shiftto DF HABITAT. (shallowandslow) (deepandfast)
12 5
For fishotherthan those of the SS GUILD, densities
A B
were greaterin the DeerfieldRiver than in the West
River regardlessof habitat type, and densities were
highestat the upstream sites with the greatestflow
fluctuations. 56
/ '

DISCUSSION

and streamhabitat
Fishcommunity
Althoughstreamfishcommunitystructurehas been o I A. .~ 0
1 2 3 4 1 2 3 4
summarizedon the basis of trophicgroups(Grossman
et al. 1982, Angermeier and Karr 1983), foraging modes 5 5
(Horowitz 1978), life historycharacteristics(Mahon C D
1984), and ecomorphology(Gatz 1979, 1981, Mahon
1984), our habitat-basedapproach provides a frame-
work for assessing the effectof artificialhabitat dis-
turbances.Habitat has been identifiedas the primary
basis on which many biological communitiesare or-
ganized (Schoener 1974), and numerous studies have
supportedthisgeneralizationforfishcommunities(e.g., 0
1 j

Gibbons and Gee 1972, Wernerand Hall 1976, Werner


et al. 1977, Tallman and Gee 1982, Schlosserand Toth 1 2 3 4 1 2 3 4
1984). Our statisticallyderived fishcommunity-hab- SITE SITE
itat model reflectsa simple relationshipbetween the FIG. 3. Medianfishdensities rangesfor
and interquartile
West River fishcommunityand the available stream fishandhabitattypebysiteand river.@-@ datafromthe
habitat.An abundant and diverse group of small-fish WestRiver,and0- - -O datafromtheDeerfield River.The
Y axis is on a largerscale fortheupperleftfigure (A), and
species and size classes (SS GUILD) were restrictedto theseverity fluctuationsin flowon theDeerfieldRiverin-
microhabitatcharacterizedprimarilyas relativelyshal- creasewithsitenumber, i.e.,in themoreupstream sites(X
low in depth,slow in currentvelocity,and concentrated axis).
along stream marginsin rifflesand pools. Fish that
were not part of the SS GUILD were relativelyscarce
and used eithera broad rangeof habitator some com- habitat, since midstreamareas generallyhave mod-
binationof deep or fastwatersthatoccurredprimarily erate velocities and depths; as stream size increases
in midstreamareas of riffles(shallow but fast),pools (medium-sized streamsand larger,orders >4), large
(slow but deep), or riffle-pooltransitions(deep and concentrationsof small fishmay become increasingly
fast). restrictedto streammarginsbecause midstreamareas
The fishcommunity-habitatmodel developed here are eithertoo deep or too fast,or both. Glova and
conflictswiththetraditionalview ofstreamsas a linear Duncan (1985) describedthispatternforvarious-sized
sequence of riffleand pool habitats. It emphasizes a channelsin a largebraided river:many small fishwere
bank-midstreamorientation,since shallow and slow evenlydistributedacross small channelriffles but were
habitatis concentratedalong the streamedge in both restrictedto stream margins in large channels. Al-
riffles and pools. In many studies(e.g., Sheldon 1968, thoughthe pool vs. riffleapproach has been usefulin
Gibbons and Gee 1972, Finger1982, Schlosser 1982a, many studies, it would have resulted in an entirely
b, Tallman and Gee 1982, Angermeierand Karr 1983, differentand less distinctpatternin our study.
Herbold 1984), fishuse of streamhabitatwas consid- Althoughour generalapproach differed fromthatin
ered withrespectto riffles (generallyshallow and fast), most otherstudies,many details of the fishcommu-
pools (deep and slow), and transitionzones (interme- nity-habitatmodel that we developed here are sup-
diate). Althoughthe importanceof a bank-midstream ported by the work of others. Like Sheldon (1968),
habitat orientationhas been noted for invertebrates Gorman and Karr (1978), and Moyle and Vondracek
(Cummins 1964, Peckarsky1983) it has not been di- (1985), we found water depth and current velocity to
rectlyidentifiedforfishesin high-gradient streams.The be the most importanthabitat variables affecting fish
difference in fishand habitat orientationbetweenthe distribution.Studies in Illinois (Schlosser 1982a, b),
presentstudy and those cited above may be due to New York (Finger 1982), and California(Moyle and
differences in streamsize or the relative size of sam- Baltz 1985) identifieda patternsimilar to that in the
plingunits. In headwaterstreams(orders 1-3 of Van- WestRiver:shallowand slowhabitatwas used bysmall,
note et al. [1980]), small fishuse most or all oftheriffle youngfishof several species, and deep areas were pri-

This content downloaded on Sat, 16 Feb 2013 17:50:12 PM


All use subject to JSTOR Terms and Conditions
April 1988 FISH COMMUNITIES IN REGULATED STREAMS 389

marilyinhabitedby larger,older fish.Also, our data that had the greatestfluctuationsin flow.This differ-
and analyses were sensitive enough to place fish in ence representsa major change in species, size class,
distinctcategories(SS GUILD and others) that were and numericalcompositionof a riverfishcommunity,
previouslyidentifiedas distinctonlyin termsofhabitat especiallysincethisgroupconstituted> 90% ofthefish
use: small and large white sucker (Finger 1982) and in the unregulatedriver.Some evidence suggestshow
small and large smallmouthbass (Probst et al. 1984). the fluctuating habitataffectsthis guild. Flow changes
The separategroupingof blacknose dace and longnose displace the shallow shorelinezones, forcingfishre-
dace, whichwereidentifiedby Gibbons and Gee (1972) strictedto these areas (the SS GUILD) to relocate to
as distinctlydifferent in microhabitatuse, was partic- maintaintheirspecificset of habitatconditions.Rapid
ularly notable because the longnose dace is a small decreases in flowpose a threatforany fishnot able to
species that uses fasterwater than that used by the move withtherecedingshoreline.On severaloccasions
abundantsmall SS GUILD fish.Consequently,thefish duringthe fieldworkin the DeerfieldRiver,we saw SS
community-habitatmodel developed here incorpo- GUILD fishstrandedand trappedin isolated puddles
rates many recognized stream fishcommunitychar- afterrapid decreasesin riverdischarge.Kroger(1973)
acteristics,even thoughwe did not relyon a rifflevs. reportedsimilarobservationsforsmall fishin a highly
pool habitat classificationbut instead used statistical fluctuatingsection of the Snake River, Idaho. Also,
methodsto identifygroupings. rapidflowincreasesmightexpose fishoftheSS GUILD
Althoughmost fishtend to stronglyprefera specific to increased predation,since shallow shorelineareas
typeof habitat(Hynes 1970, Gorman and Karr 1978, become accessible to larger piscivores as depth in-
Moreau and Legendre 1979, Moyle and Li 1979, Sa- creases, much as fluctuatingreservoirlevels increase
vard and Moreau 1982, Moyle and Vondracek 1985), thesusceptibility of shallow-waterfishto consumption
our resultssuggestedthattwo (large smallmouthbass, by piscivores (e.g., Heisey et al. 1980). Power et al.
American eel) were habitat generalists.In the West (1985) and Schlosser(1987) have clearlydocumented
River, the other 13 of 15 species or size classes used thatlargepiscivores(Micropterusspp.) forcesmall fish-
specifictypesof microhabitatthat could be quantita- es into shallow-waterrefugiain streams.If unstable
tivelyand effectively (a < .05) described.Consequent- shallow-waterhabitatsprovide less effective refugefor
ly, it cannot be assumed thatphysicalhabitat charac- small fish,then the functionalvalue of the habitat is
teristicsare importantin determiningthe distribution lostand a reductionin SS GUILD densitywould occur.
of all fishspecies in streams. In contrastto SS GUILD fish,the densityof other
species and size classes responded positively to in-
Effectsofflowfluctuations creased habitatvariability.As a group,these fishwere
Hydroelectricpowergenerationhas imposed a high- comparativelydense in theregulatedriverand densest
ly artificialflow regime on the Deerfield River. The at the siteswhereflowfluctuationswere greatest.This
frequency(up to severaltimesa day) and rate(minutes positiveresponsewas similarin SS HABITAT and DF
to hours,dependingon location of changein the Deer- HABITAT, and was a reflectionof both habitat spe-
fieldRiver) of fluctuationswere most uncharacteristic cialists, such as the large white suckers,and habitat
of a natural stream.Also, flowfluctuationswere un- generalists,such as the American eel and large small-
predictabledue to the complex mix of factors(such as mouth bass. Increased abundances of relativelyold,
electric demand, regional power supply, short-term large fishalso are associated with periods of frequent
nonhydroelectric capacity,upstreamreleases) thatde- naturalflowchanges(Schlosser 1982a). Increasedden-
terminepower generationschedules in the Deerfield sities of generalistfisheshave been associated with
River basin. Consequently,the aquatic habitat in the large flow changes (Harrell 1978) or other riverine
DeerfieldRiver was highlyunstable. Because natural modifications(Goldstein 1981). In theDeerfieldRiver,
environmentalinstabilityplays a criticalrole in shap- t~hehabitatof non-SS GUILD fishis changedless than
ing communitystructure(Wiens 1977, Connell 1978), thatofSS GUILD fish,fora givenflowchange,because
increased environmentalinstabilityshould non-SS GUILD fishuse eithermidstreamhabitats(deep
artificially
also affectcommunitystructure.Our analyses dem- or fastareas in riffles, pools, and transitionareas) or
onstratedclear differences in the fishcommunitiesbe- all types of habitat. However, we developed no evi-
tweenthe unregulatedWest River and the highlyfluc- dence to indicate which factorsaccount forincreased
tuating Deerfield River, and these differences densityof this group: food availability mightbe in-
correspondedqualitativelyto differences among sites creased as a resultof reduced consumptionof inver-
along the gradientof flowvariabilityin the Deerfield tebratesby the SS GUILD, increased availability of
River. invertebratesdue to flow fluctuations,or increased
The shallow-and slow-waterfishes(SS GUILD), an availability of the small longnose dace (a non-SS
abundant and diverse group of exclusivelysmall fish, GUILD fish)as the preyof piscivores.
were adversely affectedby the artificiallyhigh vari- The rivercontinuumconceptofVannoteet al. (1980)
abilityin flow.This group was reduced in abundance characterizes streamsof the size in the presentstudy
in the DeerfieldRiver and was absent at the studysite (fifthorder)as providingmoderatelyvariable environ-

This content downloaded on Sat, 16 Feb 2013 17:50:12 PM


All use subject to JSTOR Terms and Conditions
390 MARK B. BAIN ET AL. Ecology,Vol. 69, No. 2

ments.Connell's (1978) intermediatedisturbancehy- Massachusetts


Divisionof MarineFisheries,and theUni-
pothesis associated high community diversitywith versity
ofMassachusetts.
Thispaperis contribution
number
105oftheMassachusetts
CooperativeFisheryResearchUnit.
moderatelevels of environmentalvariability.Connell
and Keough (1985) add thatnaturaldisturbancesoccur LITERATURE CITED

on a rangeof spatial and temporalscales, and the re- Angermeier,P. L., and J. R. Karr. 1983. Fish communities
along environmentalgradients in a system of tropical
sponses of natural communitiesto them depend on
streams.EnvironmentalBiology of Fishes 9:117-135.
disturbancesize, intensity,and frequency.The artificial Bain, M. B. 1985. Fish communitystructurein riverswith
disturbancescaused by hydroelectric powergeneration naturaland modifieddailyflowregimes.Dissertation.Uni-
on the Deerfield River imposes a temporal scale of versityof Massachusetts,Amherst,Massachusetts,USA.
disturbancethat is entirelyoutside the natural range Bain, M. B., J. T. Finn, and H. E. Booke. 1985a. A quan-
titativemethodforsamplingriverinemicrohabitatsbyelec-
experiencedby streamfishesand beyond the capabil- trofishing. North American Journalof Fisheries Manage-
ities of many species to exploit. Ward and Stanford's ment5:489-493.
(1983) contentionthathydroelectric generationcauses Bain, M. B., J.T. Finn,and H. E. Booke. 1985b. Quantifying
excessive temporalvariabilitythat resultsin reduced streamsubstrateforhabitatanalysis studies.NorthAmer-
ican Journalof FisheriesManagement5:499-506.
communitydiversityof streambiota is supportedby
Bain, M. B., J. T. Finn, L. J. Gerardi,Jr.,M. R. Ross, and
our results;we documentedthe reductionand loss of W. P. Saunders,Jr. 1982. An evaluationofmethodologies
the most diverse and abundant fishcommunitycom- forassessingthe effectsof flowfluctuationson streamfish.
ponent(the SS GUILD). United States Fish and WildlifeService,FWS/OBS-82/63,
The coexistenceof some species and highcommu- Newton Corner,Massachusetts,USA.
Barila, T. Y., R. D. Williams, and J. R. Stauffer,Jr. 1981.
nitydiversityfrequentlydepend on habitat heteroge- The influenceof streamorderand selectedstreambed pa-
neitycaused by environmentaldisturbances(papers in rameterson fishdiversityin Raystown Branch, Susque-
Pickettand White 1985), but large-scaleor frequent hanna River drainage, Pennsylvania.Journalof Applied
exotic disturbancesmay impose habitat homogeneity Ecology 49:193-198.
Brooks, J. L., and E. S. Deevey, Jr. 1963. New England.
(Denslow 1985). Our resultssuggestthatthe extreme
Pages 117-162 in D. G. Frey,editor.Limnologyin North
flowvariabilityin DeerfieldRiver acts to impose func- America. Universityof Wisconsin Press, Madison, Wis-
tional habitat homogeneity.The reductionand elim- consin,USA.
ination of the SS GUILD under fluctuatinghabitat Carnes, B. A., and N. A. Slade. 1982. Some commentson
conditions indicates that this group was not able to niche analysisin canonical space. Ecology 63:888-893.
Connell, J. H. 1978. Diversityin tropical rain forestsand
effectively persistin its particularmicrohabitateven coral reefs.Science 199:1302-1310.
though that microhabitatalways physicallyexisted. Connell, J. H., and M. J. Keough. 1985. Disturbance and
Without the functionalavailability of shallow, slow, patch dynamics of subtidal marine animals on hard sub-
shorelineareas, the streamenvironmentbecomes one strata.Pages 125-151 in S. T. A. Pickettand P. S. White,
generaltype of usable habitat,which we found to be editors.The ecology of naturaldisturbanceand patch dy-
namics. Academic Press,New York, New York, USA.
dominated by the few habitat generalistsand those Cummins,K. W. 1962. An evaluation of some techniques
species usingmidstream-type habitat.Apparently,dis- forthe collectionand analysisof benthicsamples withspe-
turbancesof natural magnitudesthat occur at an ar- cial emphasison loticwaters.AmericanMidland Naturalist
tificiallyhighfrequencycan exceed the abilityof most 67:477-504.
1964. Factorslimitingthe microdistribution of lar-
species to exploit the variability.Our studyexpands vae of caddisflies,Pycnopsychelepida (Hagen) and P. gut-
on researchofnaturallydisturbedsystemsbyproviding tifer(Walker)in a Michiganstream.Ecological Monographs
resultson an extremecase of disturbancethatappears 34:271-295.
to reduce functionalhabitat heterogeneityand com- Denslow, J. S. 1985. Disturbance-mediatedcoexistenceof
species. Pages 307-323 in S. T. A. Pickettand P. S. White,
munitycomplexity. editors.The ecology of naturaldisturbanceand patch dy-
namics. Academic Press, New York, New York, USA.
ACKNOWLEDGMENTS
Fienberg,S. E. 1980. The analysis of cross-classifiedcate-
We wishto thankB. Behnke,C. D'Avanzo,P. Eschmeyer, goricaldata. MassachusettsInstituteof TechnologyPress,
K. LaGory,D. Miller,D. Orth,M. Ross,I. Schlosser,A. Zale, Cambridge,Massachusetts,USA.
and especially F. Rahelfortheirhelpfulcomments and sug- Finger,T. R. 1982. Fish community-habitatrelationsin a
gestions on themanuscript;K. Bugley, C. Cassidy,
J.Burnett, centralNew York stream.Journalof FreshwaterEcology
T. Clifford,S. Hadden,J.McMenemy, J.Nickolson,and R. 1:343-352.
Taylorfortheirassistancein thefield;A. Wentworth, who Fraser,J. C. 1972. Regulateddischargeand the streamen-
voluntarily providedhundreds ofhoursofhelpin thefield; vironment.Pages 263-286 in R. Oglesby, C. A. Carlson,
and personnel of theNew EnglandPowerCompany,Shel- and J. A. McCann, editors.River ecology and man. Aca-
burne,Massachusetts, Bostonand Maine Corporation, and demic Press,New York, New York, USA.
theUnitedStatesGeologicalSurvey(Bostonand Concord), Gatz, A. J.,Jr. 1979. Communityorganizationin fishesas
fordischarge data,information, and accessprivileges. indicatedby morphologicalfeatures.Ecology 60:711-718.
Financialsupportforthisprojectwas providedbygrants 1981. Morphologicallyinferred nichedifferentiation
fromtheNationalWildlifeFederationand theConnecticut in streamfishes.AmericanMidland Naturalist106:10-21.
RiverWatershed Council.Major funding was providedby Gibbons, J. R. H., and J. H. Gee. 1972. Ecological segre-
theMassachusetts CooperativeFishery ResearchUnit,which gation betweenlongnose dace (Genus Rhinichthys)in the
is jointlysponsoredby theUnitedStatesFishand Wildlife Mink River, Manitoba. Journalof the Fisheries Research
Service,Massachusetts Divisionof Fisheriesand Wildlife, Board of Canada 29:1245-1252.

This content downloaded on Sat, 16 Feb 2013 17:50:12 PM


All use subject to JSTOR Terms and Conditions
April1988 FISH COMMUNITIES IN REGULATED STREAMS 391

Glova, G. J.,and M. J. Duncan. 1985. Potentialeffectsof D. H. Bent. 1975. SPSS, statisticspackage forthe social
reducedflowson fishhabitatsin a largebraided river,New sciences. McGraw-Hill, New York, New York, USA.
Zealand. Transactions of the American Fisheries Society Northcote,T. G., and D. W. Wilke. 1963. Underwatercen-
114:165-181. sus of streamfishpopulations. Transactionsof the Amer-
Goldstein,R. M. 1978. Quantitativecomparisonof seining ican FisheriesSociety92:146-151.
and underwaterobservationfor streamfishsurveys.Pro- Orth, D. J., and 0. E. Maughan. 1982. Evaluation of the
gressiveFish-Culturist40:108-111. incrementalmethodologyforrecommendinginstreamflow
1981. Longitudinalsuccessionin impactassessment forfishes.Transactionsof the American Fisheries Society
ofriversystemfishcommunities.WaterResourcesBulletin 111:413-445.
17:75-81. Peckarsky,B. L. 1983. Biotic interactionsor abiotic limi-
Gorman,G. T., and J.R. Karr. 1978. Habitat structureand tations?A model of loticcommunitystructure.Pages 303-
streamfishcommunities.Ecology 59:507-515. 323 in T. D. Fontaine III and S. M. Bartell,editors. Dy-
Green,R. H. 1979. Samplingdesignand statisticalmethods namics of lotic ecosystems.Ann ArborScience Publishers,
forenvironmentalbiologists.John Wiley and Sons, New Ann Arbor,Michigan,USA.
York, New York, USA. Pickett,S. T. A., and P. S. White. 1985. The ecology of
Grossman,G. D., P. B. Moyle, and J. 0. Whitaker,Jr. 1982. naturaldisturbanceand patch dynamics.Academic Press,
Stochasticityin structuraland functionalcharacteristicsof New York, New York, USA.
an Indiana stream fishassemblage: a test of community Power, M. E., W. J. Matthews, and A. J. Stewart. 1985.
theory.American Naturalist120:423-454. Grazingminnows,piscivorousbass, and streamalgae: dy-
Harrell,H. L. 1978. Response of the Devil's River (Texas) namics of a stronginteraction.Ecology 66:1448-1456.
fishcommunityto flooding.Copeia 1978:60-68. Probst, W. E., C. F. Rabeni, W. G. Covington, and R. E.
Hawks, H. A. 1975. River zonationand classification.Pages Marteney. 1984. Resource use by stream-dwellingrock
312-374 inB. A. Whitton,editor.River ecology.University bass and smallmouthbass. Transactionsof the American
of CaliforniaPress, Los Angeles,California,USA. FisheriesSociety 113:283-294.
Heisey, P. G., D. Mathur,and N. C. Magnusson. 1980. Ac- Rice, J.,R. D. Ohmart,and B. W. Anderson. 1983. Habitat
celeratedgrowthof smallmouthbass in a pumped storage selectionattributesof an avian community:a discriminate
system.TransactionsoftheAmericanFisheriesSociety109: analysisinvestigation.Ecological Monographs53:263-290.
371-377. Rotenberry, J.T., and J.A. Wiens. 1980. Habitat structure,
Herbold, B. 1984. Structureof an Indiana streamfishas- patchiness,and avian communitiesin theNorthAmerican
sociation: choosingan appropriatemodel. American Nat- steppevegetation:a multivariateanalysis.Ecology61:1228-
uralist124:561-572. 1250.
Holden, P. B. 1979. Ecology of riverinefishesin regulated Sale, M. J. 1985. Aquatic ecosystemresponseto flowmod-
streamsystemswithemphasison theColorado River.Pages ification:an overview of the issues. Pages 25-31 in F. W.
57-74 inJ.V. Ward and J.A. Stanford,editors.The ecology Olson, R. G. White,and R. H. Hamre, editors.Proceedings
ofregulatedstreams.Plenum,New York, New York, USA. of the symposium on small hydropowerand fisheries.
Horowitz,R. J. 1978. Temporal variabilitypatternsand the AmericanFisheriesSociety,Bethesda, Maryland,USA.
distributionalpatternsof stream fish. Ecological Mono- Sale, P. F., and W. A. Douglas. 1981. Precisionand accuracy
graphs48:307-321. of visual census technique for fishassemblages on coral
Hynes,H. B. N. 1970. The ecologyof runningwaters.Uni- patch reefs.EnvironmentalBiology of Fishes 6:333-339.
versityof Toronto Press, Toronto, Canada. Savard, L., and G. Moreau. 1982. Etude des relationsentre
Keast, A., and J. Harker. 1977. Stripcounts as a means of les communautespiscicoles et les differents habitatsd'une
determiningdensities and habitat utilization patternsin rivierenordique:notiond'habitatoptimal.Canadian Jour-
lake fishes.EnvironmentalBiology of Fishes 1:181-188. nal of Zoology 60:3344-3352.
Kroger,R. L. 1973. Biological effectsof fluctuatingwater Schlosser,I. J. 1982a. Fish communitystructureand func-
levels in the Snake River, Grand Teton National Park, tion along two habitat gradientsin a headwater stream.
Wyoming.American Midland Naturalist89:478-481. Ecological Monographs52:395-414.
Kuehne, R. A. 1962. A classificationof streams,illustrated 1982b. Trophic structure, reproductivesuccess,and
by fishdistributionin an easternKentuckycreek.Ecology growthrate of fishesin a naturaland modifiedheadwater
43:608-614. stream.Canadian JournalofFisheriesand Aquatic Sciences
Kushlan, J. A. 1976. Environmentalstabilityand fishcom- 39:968-978.
munitydiversity.Ecology 57:821-825. . 1985. Flow regime,juvenile abundance, and the
Mahon, R. 1984. Divergentstructureof fishtaxocenes of assemblage structureof stream fishes.Ecology 66:1484-
northtemperatestreams.Canadian JournalofFisheriesand 1490.
Aquatic Sciences 41:330-350. 1987. The role of predationin age- and size-related
Moreau, G., and L. Legendre. 1979. Relation entrehabitat habitatuse by streamfishes.Ecology 68:651-659.
et peuplementsde poissons: essai de definitiond'une meth- Schlosser,I. J., and L. A. Toth. 1984. Niche relationships
ode numeriquepour des rivieresnordiques.Hydrobiologia and populationecologyofrainbow(Etheostomacaeruleum)
67:81-87. and fantail(E. flabellare) dartersin a temporallyvariable
Moyle, P. B., and D. M. Baltz. 1985. Microhabitatuse by environment.Oikos 42:229-238.
an assemblage of Californiastreamfishes:developing cri- Schoener,T. W. 1974. Resource partitioningin ecological
teriaforinstreamflowdeterminations.Transactionsof the communities.Science 185:27-39.
AmericanFisheriesSociety 114:695-704. Sheldon, A. L. 1968. Species diversityand longitudinal
Moyle, P. B., and H. W. Li. 1979. Communityecologyand succession in streamfishes.Ecology 49:193-198.
predator-prey relationsin warmwaterstreams.Pages 171- Tallman, R. F., and J. H. Gee. 1982. Intraspecificresource
180 inR. H. Stroudand H. Clepper,editors.Predator-prey partitioningin a headwaterstreamfish,the pearl dace Se-
systemsin fisheriesmanagement.Sport Fishing Institute, motilusmargarita(Cyprinidae).EnvironmentalBiologyof
Washington,D.C., USA. Fishes 7:243-249.
Moyle,P. B., and B. Vondracek. 1985. Persistenceand struc- Vannote,R. L., G. W. Minshall,K. W. Cummins,J.R. Sedell,
ture of the fishassemblage in a small California stream. and C. E. Cushing. 1980. The rivercontinuumconcept.
Ecology 66:1-13. Canadian Journalof Fisheries and Aquatic Sciences 37:
Nie, N. H., C. H. Hull, J. G. Jenkins,K. Steinbrenner,and 130-137.

This content downloaded on Sat, 16 Feb 2013 17:50:12 PM


All use subject to JSTOR Terms and Conditions
392 MARK B. BAIN ET AL. Ecology,Vol. 69, No. 2

Ward,J.V., and J.A. Stanford.1983. The intermediate-Werner, E. E., D. J.Hall, D. R. Laughlin,D. J.Wagner,L.
disturbance an explanation
hypothesis: forbioticdiversity A. Wilsmann, andF. C. Funk. 1977. Habitatpartitioning
patterns in loticecosystems.Pages347-356in T. D. Fon- in a freshwater fishcommunity. Journalof theFisheries
taineIII and S. M. Bartell,editors.Dynamicsofloticeco- ResearchBoardofCanada34:360-370.
systems. AnnArborScience,AnnArbor,Michigan, USA. Wiens,J.A. 1977. On competition and variableenviron-
Werner,E. E., andJ.D. Hall. 1976. Nicheshiftsin sunfish: ments.AmericanScientist 65:590-597.
experimental Science191:404-
evidenceand significance.
405.

APPENDIX
TABLE Al. Common name, abbreviationcode, scientificname, and rangeoftotallengthforspecies capturedin the Deerfield
River and West River, 1982-1983.

Code Total-lengthrange
Common name (used in Fig. 2) Scientificname (mm)
Common species
Americaneel Anquillarostrata 130-670
Blacknose dace BLKND Rhinichthysatratulus 21-54
Bluegill BG Lempomismacrochris 30-109
Fallfish FF Semotiluscorporalis 20-148
Largemouthbass LMB Micropterussalmoides 40-110
Longnose dace LND Rhinichthyscataractae 20-144
Mimic shiner MMSH Notropisvolucellus 43-61
Pumpkinseed PS Lepomisgibbosus 42-161
Rock bass Ambloplitesrupestris 30-202
Small, <100 mm SRB
Large, > 100 mm LRB
Smallmouthbass dolomieui
Micropterus 20-411
Small, <?100 mm SSMB
Large, >100 mm
Tessellated darter TD Etheostoma
olmstedi 26-98
White sucker Catostomus
commersoni 22-444
Small, < 150 mm SWS
Large, > 150 mm LWS
Uncommon species
Atlanticsalmon Salmo salar 124-192
Brown bullhead Ictalurusnebulosus 100-126
Brown trout Salmo trutta 189-356
Common shiner Notropiscornutus 73-78
Longnose sucker Catostomus catostomus 32-98
Rainbow trout Salmogairdneri 124-192
Sea lamprey Petromyzon marinus 62-111
Spottail shiner Notropishudsonius 43-63
Yellow bullhead Ictalurusnatalis 34
Yellow perch Percaflavescens 101-188
White perch Moroneamericana 49

This content downloaded on Sat, 16 Feb 2013 17:50:12 PM


All use subject to JSTOR Terms and Conditions

You might also like