The Natural Flow RegimeA paradigm for river conservation and restoration

N. LeRoy Poff, J. David Allan, Mark B. Bain, James R. Karr, Karen L. Prestegaard,Brian D. Richter, Richard E. Sparks, and Julie C. Stromberg

H umans have long been fasci-nated by the dynamism offree-flowing waters. Yet we

have expended great effort to tame

irivers for transportation, water sup-ply, flood control, agriculture, and

;f

power generation. It is now recog-nized that harnessing of streams andrivers comes at great cost: Many

. rivers no longer support socially val-ued native species or sustain healthyecosystems that provide importantgoods and services (Naiman et al.

; ‘1995, NRC 1992).

N. LeRoy Poff is an assistant professorin the Department of Biology, ColoradoState University, Fort Collins, CO 80523-11878 and formerly senior scientist atTrout Unlimited, Arlington, VA 22209.J. David Allan is a professor at the Schoolof Natural Resources & Environment,IJniversity of Michigan, Ann Arbor, MI

I

4.8109-l 115. Mark B. Bain is a researchscientist and associate professor at theNew York Cooperative Fish & Wildlife

tResearch Unit of the Department of

iNatural Resources, Cornell University,Ithaca, NY 14853-3001. James R. Karris a professor in the departments of Fish-eries and Zoology, Box 357980, Univer-sity of Washington, Seattle, WA 98195-7980. Karen L. Prestegaard is an associateprofessor in the Department of Geology,University of Maryland, College Park, MD20742. Brian D. Richter is national hy-drologist in the Biohydrology Program,The Nature Conservancy, Hayden, CO81639. Richard E. Sparks is director ofthe River Research Laboratories at theIllinois Natural History Survey, Havana,IL 62644. Julie C. Stromberg is an asso-ciate professor in the Department ofPlant Biology, Arizona State University,Tempe, AZ 85281. 0 1997 AmericanInstitute of Biological Sciences.

13‘, December 1991

The ecological integrity

of river ecosystemsdepends on their natural

dynamic character

The extensive ecological degrada-tion and loss of biological diversityresulting from river exploitation iseliciting widespread concern for con-servation and restoration of healthyriver ecosystems among scientists andthe lay public alike (Allan and Flecker1993, Hughes and Noss 1992, Karret al. 1985, TNC 1996, Williams etal. 1996). Extirpation of species, clo-sures of fisheries, groundwater deple-tion, declines in water quality andavailability, and more frequent andintense flooding are increasingly rec-ognized as consequences of currentriver management and developmentpolicies (Abramovitz 1996, Collieret al. 1996, Naiman et al. 1995). Thebroad social support in the UnitedStates for the Endangered SpeciesAct, the recognition of the intrinsicvalue of noncommercial native spe-cies, and the proliferation of water-shed councils and riverwatch teamsare evidence of society’s interest inmaintaining the ecological integrityand self-sustaining productivity offree-flowing river systems.

Society’s ability to maintain andrestore the integrity of river ecosys-tems requires that conservation andm a n a g e m e n t a c t i o n s b e f i r m l ygrounded in scientific understand-

ing. However, current managementapproaches often fail to recognizethe fundamental scientific principlethat the integrity of flowing watersystems depends largely on their natu-ral dynamic character; as a result,these methods frequently prevent suc-cessful river conservation or restora-tion. Streamflow quantity and tim-ing are critical components of watersupply, water quality, and the eco-logical integrity of river systems. In-deed, streamflow, which is stronglycorrelated with many critical physi-cochemical characteristics of rivers,such as water temperature, channelgeomorphology, and habitat diver-sity, can be considered a “mastervariable” that limits the distributionand abundance of riverine species(Power et al. 1995, Resh et al. 1988)and regulates the ecological integrityof flowing water systems (Figure 1).Until recently, however, the impor-tance of natural streamflow variabil-ity in maintaining healthy aquaticecosystems has been virtually ignoredin a management context.

Historically, the “protection” ofriver ecosystems has been limited inscope, emphasizing water quality andonly one aspect of water quantity:minimum flow. Water resourcesmanagement has also suffered fromthe often incongruent perspectivesand fragmented responsibili ty ofagencies (for example, the US ArmyCorps of Engineers and Bureau ofReclamation are responsible for wa-ter supply and flood control, the USEnvironmental Protection Agencyand state environmental agencies forwater quality, and the US Fish &

769

Figure 1. Flow regimeis ofcentral importancein sustaining the eco-logical integrity of flow-ing water systems. Thefive components of theflow regime-magni-tude, frequency, dura-tion, timing, and rateof change-influenceintegrity both directlyand indirectly, throughtheir effects on otherprimary regulators ofintegrity. Modificationof flow thus has cas-cading effects on theecological integrity ofrivers. After Karr 1991.

Flow RegimeMagnitudeFrequencyDurationTimingRate of Change

Ecological Integrity

Wildlife Service for water-dependentspecies of sporting, commercial, orconservation value), making it diffi-cult, if not impossible, to manage theentire river ecosystem (Karr 1991).However, environmental dynamismis now recognized as central to sus-taining and conserving native spe-cies diversity and ecological integ-rity in rivers and other ecosystems(Helling and Meffe 1996, Hughes1994, Pickett et al. 1992, Stanford etal. 1996), and coordinated actionsare therefore necessary to protectand restore a river’s natural flowvariability.

In this article, we synthesize exist-ing scientific knowledge to argue thatthe natural flow regime plays a criticalrole in sustaining native biodiversityand ecosystem integrity in rivers.Decades of observation of the effectsof human alteration of natural flowregimes have resulted in a well-grounded scientific perspective onwhy altering hydrologic variabilityin rivers is ecologically harmful (e.g.,Arthington et al. 1991, Castleberryet al. 1996, Hill et al. 1991, Johnsonet al. 1976, Richter et al. 1997, Sparks1995, Stanford et al. 1996,Toth 1995,Tyus 1990). Current pressing demandson water use and the continuing alter-ation of watersheds require scientiststo help develop management proto-cols that can accommodate economicuses while protecting ecosystem func-tions. For humans to continue to relyon river ecosystems for sustainablefood production, power production,waste assimilation, and flood con-trol, a new, holistic, ecoIogicaI per-

770

spective on water management isneeded to guide society’s interac-tions with rivers.

The natural flow regime

The natural flow of a river varies ontime scales of hours, days, seasons,years, and longer. Many years ofobservation from a streamflow gaugeare generally needed to describe thecharacteristic pattern of a river’s flowquantity, timing, and variability-that is, its natural flow regime. Com-ponents of a natural flow regime canbe characterized using various timeseries (e.g., Fourier and wavelet) andprobability analyses of, for example,extremely high or low flows, or ofthe entire range of flows expressedas average daily discharge (Dunneand Leopold 1978). In watershedslacking long-term streamflow data,analyses can be extended statisti-cally from gauged streams in thesame geographic area. The frequencyof large-magnitude floods can be es-timated by paleohydrologic studiesof debris left by floods and by studiesof historical damage to living trees(Hupp and Osterkamp 1985, Knox1972). These historical techniques canbe used to extend existing hydrologicrecords or to provide estimates offlood flows for ungauged sites.

River flow regimes show regionalpatterns that are determined largelyby river size and by geographic varia-tion in climate, geology, topogra-phy, and vegetative cover. For ex-ample, some streams in regions withlittle seasonality in precipitation ex-

hibit relatively stable hydrographsdue to high groundwater inputs (Fig-ure 2a), whereas other streams canfluctuate greatly at virtually any timeof year (Figure 2b). In regions withseasonal precipitation, some streamsare dominated by snowmelt, result-ing in pronounced, predictable run-off patterns (Figure 2c), and otherslack snow accumulation and exhibitmore variable runoff patterns duringthe rainy season, with peaks occur-ring after each substantial stormevent (Figure 2d).

Five critical components of theflow regime regulate ecological pro-cesses in river ecosystems: the mag-nitude, frequency, duration, timing,and rate of change of hydrologicconditions (Poff and Ward 1989,Richter et al. 1996, Walker et al.1995). These components can be usedto characterize the entire range offlows and specific hydrologic phe-nomena, such as floods or low flows,that are critical to the integrity ofriver ecosystems. Furthermore, bydefining flow regimes in these terms,the ecological consequences of par-ticular human activities that modifyone or more components of the flowregime can be considered explicitly.

l The magnitude of discharge’ at anygiven time interval is simply theamount of water moving past a fixedlocation per unit time. Magnitudecan refer either to absolute or torelative discharge (e.g., the amountof water that inundates a floodplain).Maximum and minimum magnitudesof flow vary with climate and water-shed size both within and amongriver systems.l The frequency of occurrence refersto how often a flow above a givenmagnitude recurs over some speci-fied time interval. Frequency of oc-currence is inversely related to flowmagnitude. For example, a loo-yearflood is equaled or exceeded on aver-age once every 100 years (i.e., achance of 0.01 of occurring in anygiven year). The average (median)

‘Discharge (also known as streamflow, flow,or flow rate) is always expressed in dimen-sions of volume per time. However, a greatvariety of units are used to describe flow,depending on custom and purpose ofcharac-terization: Flows can be expressed in near-instantaneous terms (e.g., ft3/s and m’/s) orover long time intervals (e.g., acre-ft/yr).

BioScience Vol. 47 No. 11

flow is determined from a data seriesof discharges defined over a specifictime interval., and it has a frequencyof occurrence of 0.5 (a 50% prob-ability).*The &r&on is the period of timeassociated with a specific flow condi-tion. Duration can be defined relativeto a particular flow event (e.g., aflood-plain may be inundated for a specificnumber of days by a ten-year flood),or it can be a defined as a compositeexpressed over a specified time period(e.g., the number of days in a yearwhen flow exceeds some value).*The timing, or predictability, offlows of defined magnitude refers tothe regularity with which they occur.This regularity can be defined for-mally or informally and with refer-ence to different time scales (Poff1996). For example, annual peak flowsmay occur with low seasonal predict-abthty (Figure 2b) or with high sea-sonal predictability (Figure 2~).*The rate of (change, or flashiness,refers to how quickly flow changesfrom one magnitude to another. Atthe extremes, “‘flashy” streams haverapid rates of change (Figure 2b),whereas “stable” streams have slowrates of change (Figure 2a).

Hydrologic processes and the flowregime. All river flow derives ulti-mately from precipitation, but in anygiven time and place a river’s flow isderived from some combination ofsurface water, soil water, and ground-water. Climate, geology, topogra-phy, soils, and vegetation help todetermine both the supply of waterand the pathways by which precipi-tation reaches the channel. The wa-ter movement pathways depicted inFigure 3a illustrate why rivers indifferent settings have different flowregimes and why flow is variable invirtually all rivers. Collectively, over-land and shallow subsurface flowpathways create hydrograph peaks,which are the river’s response tostorm events. By contrast, deepergroundwater pathways are respon-sible for baseflow, the form of deliv-ery during periods of little rainfall.

Variability in intensity, timing,and duration of precipitation (as rainor as snow) and in the effects ofterrain, soil texture, and plant evapo-transpiration on the hydrologic cyclecombine to create local and regional

December 1997

a'1 h C- ,

Figure 2. Flow histories based on long-term, daily mean discharge records. Thesehistories show within- and among-year variation for (a) Augusta Creek, MI, (b)Satilla River, GA, (c) upper Colorado River, CO, and (d) South Fork of theMcKenzie River, OR. Each water year begins on October 1 and ends on September30. Adapted from Doff and Ward 1990.

flow patterns. For example, highflows due to rainstorms may occurover periods of hours (for permeablesoils) or even minutes (for imperme-able soils), whereas snow will meltover a period of days or weeks, whichslowly builds the peak snowmeltflood. As one proceeds downstreamwithin a watershed, river flow reflectsthe sum of flow generation and rout-ing processes operating in multiplesmall tributary watersheds. The traveltime of flow down the river system,combined with nonsynchronous tribu-tary inputs and larger downstreamchannel and floodplain storage ca-pacities, act to attenuate and todampen flow peaks. Consequently,annual hydrographs in large streamstypically show peaks created by wide-spread storms or snowmelt eventsand broad seasonal influences thataffect many tributaries together(Dunne and Leopold 1978).

The natural flow regime organizesand defines river ecosystems. In riv-ers, the physical structure of the en-vironment and, thus, of the habitat,is defined largely by physical pro-cesses, especially the movement ofwater and sediment within the chan-nel and between the channel and flood-plain. To understand the biodiversity,production, and sustainabili ty ofriver ecosystems, it is necessary toappreciate the central organizing roleplayed by a dynamically varyingphysical environment.

The physical habitat of a riverincludes sediment size and heteroge-neity, channel and floodplain mor-phology, and other geomorphic fea-tures. These features form as theavailable sediment, woody debris,and other transportablematerials aremoved and deposited by flow. Thus,habitat conditions associated withchannels and floodplains vary among

771

i

Figure 3. Stream valley cross-sections at various locations in a watershed illustrate basicprinciples about natural pathways of water moving downhill and human influences onhydrology. Runoff, which occurs when precipitation exceeds losses due to evaporationand plant transpiration, can be divided into four components (a): overland flow (1) occurswhen precipitation exceeds the infiltration capacity of the soil; shallow subsurfacestormflow (2) represents water that infiltrates the soil but is routed relatively quickly tothe stream channel; saturated overland flow (3) occurs where the water table is close tothe surface, such as adjacent to the stream channel, upstream of first-order tributaries,and in soils saturated by prior precipitation; and groundwater flow (4) representsrelatively deep and slow pathways of water movement and provides water to the streamchannel even during periods of little or no precipitation. Collectively, overland andshallow subsurface flow pathways create the peaks in the hydrograph that are a river’sresponse to storm events, whereas deeper groundwater pathways are responsible forbaseflow. Urbanized (b) and agricultural (c) land uses increase surface flow by increasingthe extent of impermeable surfaces, reducing vegetation cover, and installing drainagesystems. Relative to the unaltered state, channels often are scoured to greater depth byunnaturally high flood crests and water tables are lowered, causing baseflow to drop.Side-channels, wetlands, and episodically flooded lowlands comprise the diverse flood-plain habitats of unmodified river ecosystems (d). Levees or flood walls (e) constructedalong the banks retain flood waters in the main channel and lead to a loss of floodplainhabitat diversity and function. Dams impede the downstream movement of water and cangreatly modify a river’s flow regime, depending on whether they are operated for storage(e) or as “run-of-river,” such as for navigation (f).

rivers in accordance with both flow 1960). In many streams and riverscharacteristics and the type and theavailability of transportable materials.

with a small range of flood flows,bankfull flow can build and main-

Within a river, different habitatfeatures are created and maintained

tain the active floodplain through

by a w ide r ange o f f l ows . Fo r ex -s t r e a m m i g r a t i o n ( L e o p o l d e t a l .

ample, many channel and floodplain1964) . However , the concept of a

features, such as river bars and riffle-dominant discharge may not be ap-

pool sequences, are formed and main-plicable in all flow regimes (Wolman

tained by dominant, or bankfull, dis-and Gerson 1978). Furthermore, in

charges. These discharges are flowssome flow regimes, the flows that

that can move significant quantitiesbuild the channel may differ from

of bed or bank sediment and thatthose that build the f loodplain. For

occur frequently enough (e.g., everyexample, in rivers with a wide range

several years) to continually modifyof flood flows, floodplains may ex-

the channel (Wolman and Millerhibit major bar deposits, such asberms of boulders along the channel,

772

or other features that are left byinfrequent high-magnitude floods(e.g., Miller 1990).

Over periods of years to decades,a single river can consistently pro-vide ephemeral, seasonal, and per-sistent types of habitat that rangefrom free-flowing, to standing, to nowater. This predictable diversity ofin-channel and floodplain habitattypes has promoted the evolution ofspecies that exploit the habitat mo-saic created and maintained by hy-drologic variability. For many river-ine species, comple t i on of the l i f ecycle requires an array of differenthabitat types, whose availability overtime is regulated by the flow regime(e.g., Greenberg et al. 1996, Reeveset al. 1996, Sparks 1995). Indeed,adaptation to this environmental dy-namism allows aquatic and flood-plain species to persist in the face ofseemingly harsh conditions, such asfloods and droughts, that regularlydestroy and re-create habitat elements.

From an evolutionary perspective,the pattern of spat ial and temporalhabitat dynamics influences the rela-tive success of a species in a particu-lar environmental setting. This habi-t a t t empla te (Southwood 1977),which is dictated largely by flowregime, creates both sub& and L&o-found differences in the natural his-tories of species in different segmentsof their ranges. It also influencesspecies distribution and abundance,as well as ecosystem function (Poffand Allan 1995, Schlosser 1990,Sparks 1992, Stanford et al. 1996).Human alteration of flow regimechanges the established pattern ofnatural hydrologic variation and dis-turbance, thereby altering habitatdynamics and creating new condi-tions to which the native biota maybe poorly adapted.

Human alteration offlow regimes

Human modification of natural hy-drologic processes disrupts the dy-namic equilibrium between the move-ment of water and the movement ofsediment that exists in free-flowingrivers (Dunne and Leopold 1978).This disruption alters both gross-and fine-scale geomorphic featuresthat constitute habitat for aquaticand riparian species (Table 1). After

Bioscience Vol. 47 No. 11

Table 1. Physical responses to altered flow regimes.

Source(s) of alteration Hydrologic change(s) Geomorphic response(s) Reference(s)

Dam Capture sediment movingdownstream

Dam, diversion Reduce magnitude and frequencyof high flows

Urbanization, tiling, drainage Increase magnitude and frequencyof high flows

Reduced infiltration into soil

Levees and channelization Reduce overbank flows

~ Groundwater pumping Lowered water table levels

Downstream channel erosion andtributary headcutting

Chien 1985, Petts 1984, 1985,Williams and Wolman 1984

Bed armoring (coarsening) Chien 1985

Deposition of fines in gravel Sear 1995, Stevens et al. 1995

Channel stabilization and Johnson 1994, Williams andnarrowing Wolman 1984

Reduced formation of point bars, Chien 1985 Csecondary channels, oxbows, Fenner ec al: 1s”~~ 1989’and changes in channel planform

Bank erosion and channel widening Hammer 1972

Downward incision and floodplaindisconnection

Prestegaard 1988

Reduced baseflows

Channel restriction causingdowncutting

Leopold 1968

Daniels 1960, Prestegaardet al. 1994

Floodplain deposition anderosion prevented

Sparks 1992

Reduced channel migration and Shankman and Drake 1990formation of secondary channels

Streambank erosion and channel Kondolf and Curry 1986downcutting after loss of vegetationstability

such a disruption, it may take centu-ries for a new dynamic equilibriumto be attained by channel and flood-

ri. plain adjustments to the new flow;~i regime (Petts 1985); in some cases, ai:j, new equilibrium is never attained,

and the channel remains in a state ofcontinuous recovery from the mostrecent flood event (Wolman andGerson ‘1978). These channel andfloodplain adjustments are some-times overlooked because they can

-/ be confounded with long-term re-sponses of the channel to changing

i climates (e.g., Knox 1972). Recogni-i t ion of human-caused phys ica lchanges and associated biologicalconsequences may require many

years, and physical restoration ofthe river ecosystem may call for dra-

‘, matic action (see box on the Grand! Canyon flood, page 774).i Dams, which are the most obvi-ous direct modifiers of river flow,

capture both low and high flows forI flood control, electrical power gen-, eration, irrigation and municipal

.i water needs, maintenance of recre-@I ational reservoir levels, and naviga-

December ~997

tion. More than 85% of the inlandwaterways within the continentalUnited States are now artificiallycontrolled (NRC 1992), includingnearly 1 million km of rivers that areaffected by dams (Echeverria et al.1989). Dams capture all but the fin-est sediments moving down a river,with many severe downstream con-sequences. For example, sediment-depleted water released from damscan erode finer sediments from thereceiving channel. The coarsening ofthe streambed can, in turn, reducehabitat availability for the manyaquatic species living in or usinginterstitial spaces. In addition, chan-nels may erode, or downcut, trigger-ingrejuvenation of tributaries, whichthemselves begin eroding and mi-grating headward (Chien 1985, Petts1984). Fine sediments that are con-tributed by tributaries downstreamof a dam may be deposited betweenthe cqarse particles of the streambed(e.g., Sear 1995). In the absence ofhigh flushing flows, species with lifestages that are sensitive to sedimenatation, such as the eggs and larvae of

many invertebrates and fish, can suf-fer high mortality rates.

For many rivers, it is land-useactivities, including timber harvest,livestock grazing, agriculture, andurbanization, rather than dams, thatare the primary causes of alteredflow regimes. For example, loggingand the associated building of roadshave contributed greatly to degrada-tion of salmon streams in the PacificNorthwest, mainly through effectson runoff and sediment delivery(NRC 1996). Converting forest orprairie lands to agricultural landsgenerally decreases soil infiltrationand results in increased overlandflow, channel incision, floodplain iso-lation, and headward erosion ofstream channels (Prestegaard 1988).Many agricultural areas were drainedby the construction of ditches or tile-and-drain systems, with the resultthat many channels have become en-trenched (Brookes 1988).

These land-use practices, com-bined with extensive draining ofwetlands or overgrazing, reduce re-tention of water in watersheds and,

773

A controlled flood in the Grand Canyon1.1

Since the Glen Canyon dam first began to store water in 1963, creatingLake Powell, some 430 km (270 miles) of the Colorado River, including

Grand Canyon National Park, have been virtually bereft of seasonal floods.Before 1963, melting snow in the upper basin produced an average peakdischarge exceeding 2400 mJ/s; after the dam was constructed, releaseswere generally maintained at less than 500 m3/s. The building of the damalso trapped more than 95% of the sediment moving down the ColoradoRiver in Lake Powell (Collier et al. 1996).

This dramatic change in flow regime produced drastic alterations in thedynamic nature of the historically sediment-laden Colorado River. Theannual cycle of scour and fill had maintained large sandbars along the riverbanks, prevented encroachment of vegetation onto these bars, and limitedboulderv debris deposits from constricting the river at the mouths oftributaries (Collier et al. 1997). When flows were reduced, the limitedamount of sand accumulated in the channel rather than in bars farther upthe river banks, and shallow low-velocity habitat in eddies used by juvenilefishes declined. Flow regulation allowed for increased cover of wetland andriparian vegetation, which expanded into sites that were regularly scouredby floods in the constrained fluvial canyon of the Colorado River; however,much of the woody vegetation that established after the dam’s constructionis composed of an exotic tree, salt cedar (7’umarix sp.; Stevens et al. 1995).Restoration of flood flows clearly would help to steer the aquatic andriparian ecosystem toward its former state and decrease the area of wetlandand riparian vegetation, but precisely how the system would respond to anartificial flood could not be predicted.

In an example of adaptive management (i.e., a planned experiment toguide further actions), a controlled, seven-day flood of 1274 m3/s wasreleased through the Glen Canyon dam in late March 1996. This flow,roughly 35% of the pre-dam average for a spring flood (and far less thansome large historical floods), was the maximum flow that could passthrough the power plant turbines plus four steel drainpipes, and it costapproximately $2 million in lost hydropower revenues (Collier et al. 1997).The immediate result was significant beach building: Over 53% of thebeaches increased in size, and just 10% decreased in size. Full documenta-tion of the effects will continue to be monitored by measuring channelcross-sections and studying riparian vegetation and fish populations.

instead, route it quickly downstream,increasing the size and frequency offloods and reducing baseflow levelsduring dry periods (Figure 3b; Leo-polcl 1968). Over time, these prac-tices degrade in-channel habitat foraquatic species. They may also iso-late the floodplain from overbankflows, thereby degrading habitat forriparian species. Similarly, urban-ization and suburbanization associ-ated with human population expan-sion across the landscape createimpermeable surfaces that directwater away from subsurface path-ways to overland flow (and ofteninto storm drains). Consequently,floods increase in frequency and in-tensity @even 1986), banks erode,and channels widen (Hammer 1972),

774

and baseflow declines during dry pe-riods (Figure 3~).

Whereas dams and diversions af-fect rivers of virtually all sizes, andland-use impacts are particularly evi-dent in headwaters, lowland riversare greatly influenced by efforts tosever channel-floodplain linkages.Flood control projects have short-ened, narrowed, straightened, andleveed many river systems and cutthe main channels off from their flood-plains (NRC 1992). For example,channelization of the Kissimmee Riverabove Lake Okeechobee, Florida, bythe US Army Corps of Engineerstransformed a historical 166 kmmeandering river with a 1.5 to 3 kmwide floodplain into a 90 km longcanal flowing through a series of five

impoundments, resulting in great lossof river channel habitat and adjacentfloodplain wetlands (Toth 1995).Because levees are designed to pre-vent increases in the width of flow,rivers respond by cutting deeperchannels, reaching higher velocities,or both.

C h a n n e l i z a t i o n a n d w e t l a n ddrainage can actually increase themagnitude of extreme floods, be-cause reduction in upstream storagecapacity results in accelerated waterdelivery downstream. Much of thedamage caused by the extensiveflooding along the Mississippi Riverin 1993 resulted from levee failure asthe river reestablished historic con-nections to the floodplain. Thus, al-though elaborate storage dam andlevee systems can “reclaim” thefloodplain for agriculture and hu-man settlement in most years, theoccasional but inevitable large floodswill impose increasingly high disas-ter costs to society (Faber 1996). Thesevering of floodplains from riversalso stops the processes of sedimenterosion and deposition that regulatethe topographic diversity of flood-plains. This diversity is essential formaintaining species diversity onfloodplains, where relatively smalldifferences in land elevation result inlarge differences in annual inunda-tion and soil moisture regimes, whichregulate plant distribution and abun-dance (Sparks 1992).

Ecological functions of thenatural flow regime

Naturally variable flows create andmaintain the dynamics of in-channeland floodplain conditions and habi-tats that are essential to aquatic andriparian species, as shown schemati-cally in Figure 4. For purposes ofillustration, we treat the componentsof a flow regime individually, al-though in reality they interact incomplex ways to regulate geomor-phic and ecological processes. In de-scribing the ecological functions as-sociated with the components of aflow regime, we pay particular at-tention to high- and low-flow events,because they often serve as ecologi-cal “bottlenecks” that present criti-cal stresses and opportunities for awide array of riverine species (Poffand Ward 1989).

BioScience Vol. 47 No. 11

The magnitude and frequency ofhigh and low flows regulate numer-ous ecological processes. Frequent ,moderately high flows effectivelytransport sediment through the chan-nel (Leopold et al. 1964). This sedi-ment movement, combined with theforce of moving water, exports or-ganic resources, such as detritus andattached algae, rejuvenating the bio-logical community and allowingmany species with fast life cycles andgood colonizing ability to reestab-lish (Fisher 1983). Consequently, thecomposition and relative abundanceof species that are present in a streamor river often reflect the frequencyand intensity of high flows (Meffeand Minckley 1987, Schlosser 1985).

High flows provide further eco-logical benefits by maintaining eco-system productivity and diversity.For example, high flows remove andtransport fine sediments that wouldotherwise fill the interstitial spacesin productive gravel habitats (Beschtaand Jackson 1979). Floods importwoody debris into the channel (Kellerand Swanson 1979), where it createsnew, high-quality habitat (Figure 4;Moore and Gregory 1988, Wallaceand Benke 1984). By connecting thechanne l to the f loodpla in , h ighoverbank f l o w s a l s o m a i n t a i nbroader productivity and diversity.Floodplain wetlands provide impor-tant nursery grounds for fish andexport organic matter and organ-isms back into the main channel (Junket al. 1989, Sparks 1995, Welcomme1992). The scouring of floodplainsoils rejuvenates habitat for plantspecies that germinate only on bar-ren, wetted surfaces that are free ofcompetition (Scott et al. 1996) orthat require access to shallow watertables (Stromberget al. 1997). Flood-resistant, disturbance-adapted ripar-ian communities are maintained byflooding along river corridors, evenin river sections that have steep banksand lack fl.oodplains (Hupp andOsterkamp 1985).

Flows of low magnitude also pro-vide ecological benefits. Periods ofllow flow may present recruitmentopportunities for riparian plant spe-cies in regions where floodplains arefrequently inundated (Wharton etal. 1981). Streams that dry tempo-rarily, generally in arid regions, haveaquatic (Williams and Hynes 1977)

December 1997

Freauency

Centennial

Decadal

Annual

Figure 4. Geomorphic and ecological functions provided by different levels of flow.Water tables that sustain riparian vegetation and that delineate in-channel baseflowhabitat are maintained by groundwater inflow and flood recharge (A). Floods ofvarying size and timing are needed to maintain a diversity of riparian plant speciesand aquatic habitat. Small floods occur frequently and transport fine sediments,maintaining high benthic productivity and creating spawning habitat for fishes (B).Intermediate-size floods inundate low-lying floodplains and deposit entrained sedi-ment, allowing for the establishment of pioneer species (C). These floods also importaccumulated organic material into the channel and help to maintain the characteristicform of the active stream channel. Larger floods that recur on the order of decadesinundate the aggraded floodplain terraces, where later successional species establish(D). Rare, large floods can uproot mature riparian trees and deposit them in the channel,creating high-quality habitat for many aquatic species (E).

and riparian (Nilsen et al. 1984) spe-cies with special behavioral or physi-ological adaptations that suit themto these harsh conditions.

The duration of a specific flowcondition often determines its eco-logical significance. For example, dif-ferences in tolerance to prolongedflooding in riparian plants (Chapmanet al. 1982) and to prolonged low flowin aquatic invertebrates (Williams andHynes 1977) and fishes (Closs andLake 1996) allow these species topersist in locations from which theymight otherwise be displaced bydominant, but less tolerant, species.

The timing, or predictability, offlow events is critical ecologicallybecause the life cycles of manyaquatic and riparian species are timedto either avoid or exploit flows ofvariable magnitudes. For example,the natural timing of high or lowstreamflows provides environmen-tal cues for initiating life cycle tran-sitions in fish, such as spawning(Montgomery et al. 1983, Nesler etal. 1988), egg hatching (Nzsje et al.1995), rearing (Seegrist and Gard1978), movement onto the flood-plain for feeding or reproduction(Junk et al. 1989, Sparks 199.5,Welcomme 1992), or migration up-stream or downstream (Trepanier etal. 1996). Natural seasonal varia-tion in flow conditions can prevent

the successful establishment of non-native species with flow-dependentspawning and egg incubation require-ments, such as striped bass (Moronesaxat i l i s ; Turner and Chadwick1972) and brown trout (Salmo trtrtta;Moyle and Light 1996, Strange et al.1992).

Seasonal access to floodplain wet-lands is essential for the survival ofcertain river fishes, and such accesscan directly link high wetland produc-tivity with fish production in the streamchannel (Copp 1989, Welcomme1979). Studies of the effects on streamfishes of both extensive and limitedfloodplain inundation (Finger andStewart 1987, Ross and Baker 1983)indicate that some fishes are adaptedto exploiting floodplain habitats, andthese species decline in abundancewhen floodplain use is restricted.Models indicate that catch rates andbiomass of fish are influenced byboth maximum and minimum wet-l and a rea (Power e t a l . 199 .5 ,Welcomme and Hagborg 19771, andempirical work shows that the areaof floodplain water bodies duringnonflood periods influences the spe-cies richness of those wetland habi-tats (Halyk and Balon 1983). Thetiming of floodplain inundation isimportant for some fish because mi-gratory and reproductive behaviorsmust coincide with access to and avail-

775

Table 2. Ecological responses to alterations in components of natural flow regime.”

Flow component Specific alteration Ecological response Reference(s)

Magnitude andfrequency

Increased variation Wash-out and/or strandingLoss of sensitive species

Flow stabilization

Increased algal scour and wash-out oforganic matter

Life cycle disruption

Altered energy flowInvasion or establishment of exotic species,leading to:

Local extinctionThreat to native commercial speciesAltered communities

Reduced water and nutrients to floodplainplant species, causing:

Seedling desiccationIneffective seed dispersalLoss of scoured habitat patches and second-ary channels needed for plant establishment

Encroachment of vegetation into channels

Timing

Duration

Loss of seasonal flow peaks Disrupt cues for fish:Spawning

Prolonged low flows

Egg hatchingMigration

Loss of fish access to wetlands or backwatersModification of aquatic food web structureReduction or elimination of riparian plantrecruitmentInvasion of exotic riparian species Horton 1977Reduced plant growth rates Reily and Johnson 1982

Fausch and Bestgen 1997,Montgomery et al. 1993, Nesleret al. 1988Narsje et al. 1995Williams 1996Junk et al. 1989, Sparks 1995Power 1992, Wootton et al. 1996Fenner et al. 1985

Concentration of aquatic organismsReduction or elimination of plant coverDiminished plant species diversityDesertification of riparian speciescompositionPhysiological stress leading to reduced plantgrowth rate, morphological change,or mortality

Prolonged baseflow “spikes”

Altered inundation duration

Prolonged inundation

Rate of change Rapid changes in river stage

Accelerated flood recession

Downstream loss of floating eggs

Altered plant cover types

Change in vegetation functional typeTree mortalityLoss of riffle habitat for aquatic species

Wash-out and stranding of aquatic species

Failure of seedling establishment

Cushman 1985, Petts 1984Gehrke et al. 1995, Kingsolvingand Bain 1993, Travnichek etal. 1995Petts 1984

Scheidegger and Bain 1995

Valentin et al. 1995

Kupferberg 1996, Meffe 1984Stanford et al. 1996Busch and Smith 1995, Moyle1986, Ward and Stanford 1979

Duncan 1993Nilsson 1982Fenner et al. 1985, Rood et al1995, Scott et al. 1997,Shankman and Drake 1990Johnson 1994, Nilsson 1982

Cushman 1985, Petts 1984Taylor 1982Taylor 1982Busch and Smith 1995, Stromberget al. 1996Kondolf and Curry 1986, Perkins etal. 1984, Reily and Johnson 1982,Rood et al. 1995, Stromberg et al.1992

Robertson 1997

Auble et al. 1994

Bren 1992, Connor et al. 1981Harms et al. 1980Bogan 1993

Cushman 1985, Petts 1984

Rood et al. 1995

,‘Only representative studies are listed here. Additional references are located on the Web at http://lamar.colostate.edu/-poff/natfIow.htmI.

ability of floodplain habitats (Wel-comme 1979). The match of reproduc-tive period and wetland access alsoexplains some of the yearly variationin stream fish community composition(Finger and Stewart 1987).

Many riparian plants also havelife cycles that are adapted to theseasonal timing components of natu-

ral flow regimes through their “emer-gence phenologies”-the seasonalsequence of flowering, seed dispersal,germination, and seedling growth.The interaction of emergence phe-nologies with temporally varyingenvironmental stress from floodingor drought helps to maintain highspecies diversity in, for example,

southern floodplain forests (Strenget al. 1989). Productivity of riparianforests is also influenced by flowtiming and can increase when short-duration flooding occurs in the grow-ing season (Mitsch and Rust 1984,Molles et al. 1995).

The rate of change, or flashiness,in flow conditions can influence spe-

Biokience Vol. 47 No. 11

- -

ties persistence and coexistence. Inmany streams and rivers, particu-larly in a.rid areas, flow can changedramatically over a period of hoursdue to heavy storms. Non-nativefishes generally lack the behavioraladaptations to avoid being displacedd o w n s t r e a m b y s u d d e n f l o o d s(Minckley and Deacon 1991). In adramatic example of how floods canbenefit native species, Meffe (1984)documented that a native fish, the Gilatopminnow (Poecifiopsis occidentalis),was locally extirpated by the intro-duced predatory mosquitofish (Gam-busia crffinis) in locations where natu-ral flash floods were regulated byupstream dams, but the native speciespersisted in naturally flashy streams.

Rapid flow increases in streams ofthe central and southwestern UnitedStates often serve as spawning cuesfor native minnow species, whoserapidly developing eggs are eitherbroadcast into the water column orattached to submerged structures asfloodwaters recede (Fausch and Best-gen 1997, Robertson in press). Moregradual, seasonal rates of change inflow conditions also regulate the per-sistence of many aquatic and riparianspecies. Cottonwoods (Po@.ts spp.),for example, are disturbance speciesthat establish after winter-springflood flows, during a narrow “win-dow of opportunity” when competi-tion-free alluvial substrates and wetsoils are available for germination.A certain rate of floodwater reces-sion is critical to seedling germina-tion because seedling roots must re-main connected to a receding watertable as they grow downward (Roodand Mahoney 1990).

Ecological responses to alteredflow regimes

Modification of the natural flow re-g i m e dram.atically a f f e c t s b o t haqua t i c and r ipa r i an spec ies instreams and rivers worldwide. Eco-logical responses to altered flow re-gimes in a specific stream or riverdepend on how the components offlow have changed relative to thenatural flow regime for that particu-lar stream or river (Poff and Ward1990) and how specific geomorphicand ecological processes will respondto this relative change. As a result of

December 1997

variation in flow regime within andamong rivers (Figure 2), the samehuman activity in different locationsmay cause different degrees of changerelative to unaltered conditions and,therefore, have different ecologicalconsequences.

Flow alteration commonly changesthe magnitude and frequency of highand low flows, often reducing vari-ability but sometimes enhancing therange. For example, the extreme dailyvariations below peaking power hy-droelectric dams have no naturalanalogue in freshwater systems andrepresent, in an evolutionary sense,an extremely harsh environment offrequent, unpredictable flow distur-bance. Many aquatic populations liv-ing in these environments suffer highmortality from physiological stress,from wash-out during high flows,and from stranding during rapid de-watering (Cushman 1985, Petts1984). Especially in shallow shore-line habitats, frequent atmosphericexposure for even brief periods canresult in massive mortality of bot-tom-dwelling organisms and subse-quent severe reductions in biologicalproductivity (Weisberg et al. 1990).Moreover, the rearing and refugefunctions of shallow shoreline orbackwater areas, where many smallfish species and the young of largespecies are found (Greenberg et al.1996, Moore and Gregory 1988),are severely impaired by frequentflow fluctuations (Bain et al. 1988,Stanford 1994). In these artificiallyfluctuating environments, specializedstream or river species are typicallyreplaced by generalist species thattolerate frequent and large varia-tions in flow. Furthermore, life cyclesof many species are often disruptedand energy flow through the ecosys-tem is greatly modified (Table 2).Short-term flow modifications clearlylead to a reduction in both the natu-ral diversity and abundance of manynative fish and invertebrates.

At the opposite hydrologic ex-treme, flow stabilization below cer-tain types of dams, such as watersupply reservoirs, results in artifi-cially constant environments thatlack natural extremes. Although pro-duction of a few species may in-crease greatly, it is usually at theexpense of other native species andof systemwide species diversity

(Ward and Stanford 1979). Manylake fish species have successfullyinvaded (or been intentionally estab-lished in) flow-stabilized river envi-ronments (Moyle 1986, Moyle andLight 1996). Often top predators,these introduced fish can devastatenative river fish and threaten com-mercially valuable stocks (Stanfordet al. 1996). In the southwesternUnited States, virtually the entirenative river fish fauna is listed asthreatened under the EndangeredSpecies Act, largely as a consequenceof water withdrawal, flow stabiliza-tion, and exotic species prolifera-tion. The last remaining strongholdsof native river fishes are all in dy-namic, free-flowing rivers, whereexotic fishes are periodically reducedby natural flash floods (Minckleyand Deacon 1991, Minckley andMeffe 1987).

Flow stabilization also reduces themagnitude and frequency of overbankflows, affecting riparian plant speciesand communities. In rivers with con-strained canyon reaches or multipleshallow channels, loss of high flowsresults in increased cover of plantspecies that would otherwise be re-moved by flood scour (Ligon et al.1995, Williams and Wolman 1984).Moreover, due to other related ef-fects of flow regulation, includingincreased water salinity, non-nativevegetation often dominates, such asthe salt cedar (Tarn&x sp.) in thesemia r id wes te rn Uni ted S ta tes(Busch and Smith 1995). In alluvialvalleys, the loss of overbank flowscan greatly modify riparian commu-nities by causing plant desiccation,reduced growth, competitive exclu-sion, ineffective seed dispersal, orfailure of seedling establishment(Table 2).

The elimination of flooding mayalso affect animal species that de-pend on terrestrial habitats. For ex-ample, in the flow-stabilized PlatteRiver of the United States GreatPlains, the channel has narroweddramatically (up to 85%) over aperiod of decades (Johnson 1994).This narrowing has been facilitatedby vegetative colonization of sand-bars that formerly provided nest-ing habitat for the threatened pip-ing p lover (Charndk melodies)and endangered least tern (Sternaantillarum; Sidle et al. 1992). Sand-

777

hill cranes (Grus canadensis), whichmade the Platte River famous, haveabandoned river segments that havenarrowed the most (Krapuet al. 1984).

Changes in the duration of flowconditions also have significant bio-logical consequences. Riparian plantspecies respond dramatically to chan-nel dewatering, which occurs fre-quently in arid regions due to surfacewater diversion and groundwaterpumping. These biological and eco-logical responses range from alteredleaf morphology to total loss of ri-parian vegetation cover (Table 2).Changes in duration of inundation,independent of changes in annualvolume of flow, can alter the abun-dance of plant cover types (Auble etal. 1994). For example, increasedduration of inundation has contrib-uted to the conversion of grasslandto forest along a regulated Austra-lian river (Bren 1992). For aquaticspecies, prolonged flows of particu-lar levels can also be damaging. Inthe regulated Pecos River of NewMexico, artificially prolonged highsummer flows for irrigation displacethe floating eggs of the threatenedPecos bluntnose shiner (Notropis siniuspecosensis) into unfavorable habitat,where none survive (Robertson inpress).

Modification of natural flow tim-ing, or predictabili ty, can affectaquatic organisms both directly andindirectly. For example, some nativefishes in Norway use seasonal flowpeaks as a cue for egg hatching, andriver regulation that eliminates thesepeaks can directly reduce local popu-lation sizes of these species (Naesje etal. 1995). Furthermore, entire foodwebs, not just single species, may bemodified by altered flow timing. Inregulated rivers of northern Califor-nia, the seasonal shifting of scouringflows from winter to summer indi-rectly reduces the growth rate of juve-nile steelhead trout (Oncorhyncusmykiss) by increasing the relativeabundance of predator-resistant in-vertebrates that divert energy awayfrom the food chain leading to trout(7JC’ootton et al. 1996). In unregu-lated rivers, high winter flows re-duce these predator-resistant insectsand favor species that are more pal-atable to fish.

Riparian plant species are alsostrongly affected by altered flow tim-

Pnor to ,776. wdespread beaver dams naturally control Streemflow: dams gradually disappear as beavers are huntedto near ext,nct,on: ml, dams replace beaver clams as iernlory 1s settled.

,824 - Cremon of Army Corps of Engmeers. wth task af keepmg rwers nawgable: federal government beglns suoportof commerctaf nzwgat~on on the MISSISSIPPI.

1825 _ Complemn of Ens Canal. creabng tansporr route from the Hudson Riverto the Great Lakes.

1849. 1~350. 1860 - Swamp Land Acts. transfernng 65 mlflion acres of wetlands I” 15 states from federal to stateadmtnlstratlon for purpose of dramage;. 1650 Act gwes Everglades to Florida.

1660’s - ditchmg and drafnmg of wetlands I” Inbutanes to the Misslsslppt Rwer begms.

1901 canel bulk from Colorado River to Salton Sink and the impenal Valley IS born. floods of 1904-1905 createSalton Sea. and the ““er IS put back I” Its onglnal channel.

1902 Reclamabon Prolect Act. establishing Reclamabon Sewre to “nat!onalize the works ot irngatton’*.

1920 - Federal Power Act authorizes l,ce”sl”g of non-federal hydropower dams.

1927 MiwssIppI River floods, prowng exlsllng levees Inadequate and leadng to 1926 Flood Control Act.1926 - Colorado River Compact ratified. pamtlonmg the w&s water1933 -Tennessee Valley Authonty Act passed. and nanon embarks on first multipurpose prefect for contmlfing and

us,ng a river..1935 - Hoover Oam dedrcaled by FOR.1930.1940 - U.S. Army Corps constructs P-Foot Channel Protect. tur”!ng upper M~ssisstpp~ into an ~ntra-contmentalc channel.1940 -channel stralghtemng of tnbutanes to the MiswsIppI Awer begms.1944 . Flood Control Act authorizes federal part~wbon in flood control prowcts. and esiaokshes recreation as a fulli purpose for flood control pro,ects.

f 950

+

1953 - bulldmg of flood contml dams begms on the MiesissIppI River. 750 miles channelzed upstream from mouth.1954. Watershed Protection and Flood Prevention Act. beglns actwe Soti Conservatton Serwce wwolvement I” helpmg

farmers to channellze streams.

1963 - Glen Canyon Dam completed: 1964 - U.S. and Canada ratify Columbia River Treaty: 1965 - Callfamla StateWater Prafect approved.

1968 . Wild and Scemc Rwers Act passed to preserve ce”am ““ers I” “free-flowng Condltlon”.

1976 - PURPA passed. prowdng market for small-scale hydropower genemoon.

1966 Electnc Consumers Protecnon Act - amends Federal Power Act. requxes FERC to give equal conslderatlon topower generatm potential and fish. wlldllfe. recreanon. and other aspects of enwronmental qualny dung damlIcenslng/rellcenslng.

1992 legislation approved for federal purchase and removal of 2 pnvate dams on the Elwha River, to restore fishpZlSS=Q~.

-1993 mafor flood on MISSISSIPPI River causes exfenswe damage

i-1996 Conrrollea IlOOd of Colatado Rwer at Grano Canyon: resloral~o” of Everglaoes begns.

2oooLFigure 5. A brief history of flow alteration in the United States.

ing (Table 2). A shift in timing ofpeak flows from spring to summer,as often occurs when reservoirs aremanaged to supply irrigation water,has prevented reestablishment of theF r e m o n t c o t t o n w o o d (Pop~llusfremontii), the dominant plant spe-cies in Arizona, because flow peaksnow occur after, rather than before,its germination period (Fenner et al.1985). Non-native plant species withless specific germination require-ments may benefit from changes inf lood t iming . For example , sa l tcedar’s (Tamarix sp.) long seed dis-persal period allows it to establishafter floods occurring any time duringthe growing season, contributing to itsabundance on floodplains of the west-ern United States (Horton 1977).

Altering the rate of change in flowcan negatively affect both aquaticand riparian species. As mentionedabove, loss of natural f lashiness

threatens most of the native fish faunaI

of the American Southwest (Minckleyand Deacon 1991), and artificially \increased rates of change caused by 1peaking power hydroelectric damson historically less flashy rivers cre-

1

ates numerous ecological problems i(Table 2; Petts 1984). A modifiedrate of change can devastate riparian Ispecies, such as cottonwoods, whosesuccessful seedling growth dependson the rate of groundwater recessionfollowing floodplain inundation. Inthe S t . Mary River in Alber ta ,Canada, for example, rapid draw-

[1

downs of river stage during springhave prevented the recruitment ofyoung trees (Rood and Mahoney1990). Such effects can be reversed,however. Restoration of the springflood and its natural, slow recessionin the Truckee River in Californiahas allowed the successful establish-ment of a new generation of cotton- I

778 BioScience Vol. 47 No. 11I

T;lbic 3. Rcxnt projects in which restoration of some component(s) of natural flow regimes has occurred or been proposedfor specific ecological benefits.

Location

Trinity River, CA

Flow component(s)

Mimic timing and magnitude of peakflow

Ecological purpose(s) Reference

Rejuvenate in-channel gravel habitats; restore Barinaga 1996”early riparian succession; provide migrationflows for juvenile salmon

Truckee River, CA Mimic timing, magnitude, and duration Restore riparian trees, especially cottonwoods Klotz and Swansonof peak flow, and its rate of changeduring recession

1997

Owens River, CA Increase base flows; partially restoreoverbank flows

Restore riparian vegetation and habitat fornative fishes and non-native brown trout

Rush Creek, CA (and other Increase minimum flowstributaries to Mono Lake)

Restore riparian vegetation and habitat forwaterfowl and non-native fishes

Oldman River and tributaries, Increase summer flows; reduce rates of Restore rioarian vegetation (cottonwoods)southern Alberta, Canada postflood stage decline; mimic natural

flows in wet years

I

and cold-water (trout) fisheries

Green River, UT

San Juan River, UT/NM

Gunnison River. CO

Rio Grande River, NM

Pecos River, NM

Colorado River, AZ

Bill Williams River, AZ Mimic natural flood peak timing(proposed) and duration

Pemigewasset River, NH Reduce frequency (i.e., to no morethan natural frequency) of high flowsduring summer low-flow season; reducerate of change between low and highflows during hydropower cycles

Roanoke River. VA

Kissimmee River, FL

Mimic timing and duration of peak flowand duration and timing of nonpeakflows; reduce rapid baseflow fluctu-ations from hydropower generation

Mimic magnitude, timing, and durationof peak flow; restore low winterbaseflows

Mimic magnitude, timing, and durationof peak flow; mimic duration and timingof nonpeak flows

Mimic timing and duration of flood-plain inundation

Regulate duration and magnitude ofsummer irrigation releases to mimicspawning flow “spikes”; maintainminimum flows

Mimic magnitude and timing

Restore more natural patterning ofmonthly flows in spring; reduce rate ofchange between low and high flowsduring hydropower cycles

Mimic magnitude, duration, rate ofchange, and timing of high- and low-flow periods

Recovery of endangered fish species; enhanceother native fishes

Recovery of endangered fish species

Recovery of endangered fish species

Ecosystem processes (e.g., nitrogen flux,microbial activity, litter decomposition)

Determine spawning and habitat needsfor threatened fish species

Restore habitat for endangered fish speciesand scour riparian zone

Promote establishment of native trees

Enhance native Atlantic salmon recovery

Increased reproduction of striped bass

Restore floodplain inundation to recoverwetland functions; reestablish in-channelhabitats for fish and other aquatic species

Hill and Platts inpress

LADWP 1995

Rood et al. 1995

Stanford I994

-b

-b

Molles et al. 1995

Robertson 1997

Collier et al. 1997

USCOE 1996

FERC 1995

Rulifson and ManoochI993

Toth 1995

“J. Poles, 1997, personal communication. US Fish & Wildlife Service, Arcata, CA.bF. Pfeifer, 1997, personal communication. US Fish & Wildlife Service, Grand Junction, CO.

/

wood trees (Klotz and Swanson primarily on one or a few species opinions of the minimal flow needs1997). that live in the wetted river channel. for certain fish species (e.g., Larson I

Most of these methods have the nar- 1981).Recent approaches to row intent of establishing minimum A more sophisticated assessment

streamflow management allowable flows. The simplest make of how changes in river flow affectuse of easily analyzed flow data, of aquatic habitat is provided by the jl

Methods to estimate environmental assumptions about the regional simi- Instream Flow Incremental Method- 1’flow requirements for rivers focus larity of rivers, and of professional ology (IFIM; Bovee and Milhous

December .I 997 779

1978). IFIM combines two models, abiological one that describes the physi-cal habitat preferences of fishes (andoccasionally macroinvertebrates) interms of depth, velocity, and substrate,and a hydraulic one that estimateshow the availability of habitat forfish varies with discharge. IFIM hasbeen widely used as an organiza-tional framework for formulatingand evaluating alternative watermanagement options related to pro-duction of one or a few fish species(Stalnaker et al. 199.5).

As a predictive tool for ecologicalmanagement, the IFIM modelingapproach has been criticized both interms of the statistical validity of itsphysical habitat characterizations(Williams 1996) and the limited re-alism of its biological assumptions(Castleberry et al. 1996). Field testsof its predictions have yielded mixedresults (Morehardt 1986). Althoughthis approach continues to evolve,both by adding biological realism(Van Winkle et al. 1993) and byexpanding the range of habitatsmodeled (Stalnaker et al. 1995), inpractice it is often used only to estab-lish minimum flows for “important”(i.e., game or imperiled) fish species.But current understanding of riverecology clearly indicates that fishand other aquatic organisms requirehabitat features that cannot be main-tained by minimum flows alone (seeStalnaker 1990). A range of flows isnecessary to scour and revitalizegravel beds, to import wood andorganic matter from the floodplain,and to provide access to productiveriparian wetlands (Figure 4). Inter-annual variation in these flow peaksis also critical for maintaining chan-nel and riparian dynamics. For ex-ample, imposition of only a fixedhigh-flow level each year would sim-ply result in the equilibration of in-channel and floodplain habitats tothese constant peak flows.

Moreover, a focus on one or a fewspecies and on minimum flows failsto recognize that what is “good” forthe ecosystem may not consistentlybenefit individual species, and thatwhat is good for individual speciesmay not be of benefit to the ecosys-tem. Long-term studies of naturallyvariable systems show that some spe-cies do best in wet years, that otherspecies do best in dry years, and that

780

overall biological diversity and eco-system function benefit from thesevariations in species success (Tilmanet al. 1994). Indeed, experience inriver restoration clearly shows theimpossibility of simultaneously en-gineering optimal conditions for allspecies (Sparks 1992, 1995, Toth1995). A holistic view that attemptsto restore natural variability in eco-logical processes and species success(and that acknowledges the tremen-dous uncertainty that is inherent inattempting to mechanistically modelall species in the ecosystem) is neces-sary for ecosystem management andrestoration (Franklin 1993).

Managing toward a naturalflow regime

The first step toward better incorpo-rating flow regime into the manage-ment of river ecosystems is to recog-nize that extensive human alterationof river flow has resulted in wide-spread geomorphic and ecologicalchanges in these ecosystems. The his-tory of river use is also a history offlow alteration (Figure 5). The earlyestablishment of the US Army Corpsof Engineers is testimony to the im-portance that the nation gave to de-veloping navigable water routes andto controlling recurrent large floods.However, growing understanding ofthe ecological impacts of flow alter-ation has led to a shift toward anappreciation of the merits of free-flowing rivers. For example, the Wildand Scenic Rivers Act of 1968 recog-nized that the flow of certain riversshould be protected as a nationalresource, and the recent blossomingof natural flow restoration projects(Table 3) may herald the beginningof efforts to undo some of the dam-age of past flow alterations. The nextcentury holds promise as an era forrenegotiating human relationshipswith rivers, in which lessons from pastexperience are used to direct wise andinformed action in the future.

A large body of evidence hasshown that the natural flow regimeof virtually all rivers is inherentlyvariable, and that this variability iscritical to ecosystem function andnative biodiversity. As we have al-ready discussed, rivers with highlyaltered and regulated flows lose theirability to support natural processes

and native species. Thus, to protectpristine or nearly pristine systems, itis necessary to preserve the naturalhydrologic cycle by safeguardingagainst upstream river developmentand damaging land uses that modifyrunoff and sediment supply in thewatershed.

Most rivers are highly modified,of course, and so the greatest chal-lenges lie in managing and restoringrivers that are also used to satisfyhuman needs. Can reestablishing thenatural flow regime serve as a usefulmanagement and restoration goal?We believe that it can, although tovarying degrees, depending on thepresent extent of human interven-tion and flow alteration affecting aparticular river. Recognizing thenatural variability of river flow andexplicitly incorporating the five com-ponents of the natural flow regime(i.e., magnitude, frequency, duration,timing, and rate of change) into abroader framework for ecosystemmanagement would cons t i tu te amajor advance over most presentmanagement, which focuses on mini-mum flows and on just a few species.Such recognition would also con-tribute to the developing science ofstream restoration in heavily alteredwatersheds, where, all too often,physical channel features (e.g., barsand woody debris) are re-createdwithout regard to restoring the flowregime that will help to maintainthese re-created features.

Just as rivers have been incremen-tally modified, they can be incre-mentally restored, with resultingimprovements to many physical andbiological processes. A list of recentefforts to restore various componentsof a natural flow regime (that is, to“naturalize” river flow) demon-strates the scope for success (Table3). Many of the projects summarizedin Table 3 represent only partial stepstoward full flow restoration, but theyhave had demonstrable ecologicalbenefits. For example, high floodflows followed by mimicked naturalrates of flow decline in the OldmanRiver of Alberta, Canada, resulted ina massive cottonwood recruitmentthat extended for more than 500 kmdownstream from the Oldman Dam.Dampening of the unnatural flowfluctuations caused by hydroelectricgeneration on the Roanoke River in

BioScience Vol. 47 No. 11

Virginia has increased juvenile abun-dances of native striped bass. Mim-icking short-duration flow spikes thatare historically caused by summerthunderstorms in the regulated PecosRiver of New Mexico has benefitedthe reproductive success of the Pecosbluntnose shiner.

We also recognize that there arescientific limits to how precisely thenatural flow regime for a particularriver can be defined. It is possible tohave only an approximate knowl-edge of the historic condition of ariver, both because some human ac-tivities may have preceded the instal-lation of flow gauges, and becauseclimate conditions may have changedover the past century or more. Fur-thermore, in many rivers, year-to-year differences in the timing andquantity of flow result in substantialvariability around any average flowcondition. Accordingly, managingfor the “average” condition can bemisguided. For example, in human-altered rivers that are managed forincremental improvements, restoringa flow pattern that is simply propor-tional to the natural hydrograph inyears with little runoff may providefew if any ecological benefits, be-cause many geomorphic and eco-logical processes show nonlinear re-sponses to flow. Clearly, half of thepeak discharge will not move half ofthe sediment, half of a migration-motivational flow will not motivatehal f of the f i sh , and hal f of anoverbank flow will not inundate halfof the floodplain. In such rivers, moreecological benefits would accruefrom capitalizing on the natural be-tween-year variability in flow. Forexample, in years with above-aver-age flow, “surplus” water could beused to exceed flow thresholds thatdrive critical geomorphic and eco-logical processes.

If full flow restoration is impos-sible, mimicking certain geomorphicprocesses may provide some ecologi-cal benefits. Well-timed irrigationcould stimulate recruitment of val-ued riparian trees such as cotton-woods (Friedman et al. 1995). Stra-tegically clearing vegetation fromr ive r banks cou ld p rov ide newsources o f g rave l fo r sediment-starved regulated rivers with reducedpeak flows (e.g., Ligon et al. 1995).In all situations, managers will be

December 1997

required to make judgments aboutspecific restoration goals and to workwith appropriate components of thenatural flow regime to achieve thosegoals. Recognition of the natural flowvariability and careful identificationof key processes that are linked tovarious components of the flow re-gime are critical to making thesejudgments.

Setting specific goals to restore amore natural regime in rivers withaltered flows (or, equally important,to preserve unaltered flows in pristinerivers) should ideally be a cooperativeprocess involving river scientists, re-source managers, and appropriatestakeholders. The details of this pro-cess will vary depending on the spe-cific objectives for the river in ques-tion, the degree to which its flowregime and other environmental vari-ables (e.g., thermal regime, sedimentsupply) have been altered, and thesocial and economic constraints thatare in play. Establishing specific cri-teria for flow restoration will be chal-lenging because our understandingof the interactions of individual flowcomponents with geomorphic andecological processes is incomplete.However, quantitative, river-specificstandards can, in principle, be devel-oped based on the reconstruction ofthe natural flow regime (e.g., Rich-ter et al. 1997). Restoration actionsbased on such guidelines should beviewed as experiments to be moni-tored and evaluated-that is, adap-tive management-to provide criti-cal new knowledge for creativemanagement of natural ecosystemvariability (Table 3).

To manage rivers from this newperspective, some policy changes areneeded. The narrow regulatory fo-cus on minimum flows and singlespecies impedes enlightened rivermanagement and restoration, as dothe often conflicting mandates of themany agencies and organizations thatare involved in the process. Revi-sions of laws and regulations, andredefinition of societal goals and poli-cies, are essential to enable managersto use the best science to develop ap-propriate management programs.

Using science to guide ecosystemmanagement requires that basic andapplied research address difficultquestions in complex, real-world set-tings, in which experimental con-

trols and statistical replication areoften impossible. Too little attentionand too few resources have been de-voted to clarifying how restoringspecific components of the flow-re-gime will benefit the entire ecosys-tem. Nevertheless, it is clear that,whenever possible, the natural riversystem should be allowed to repairand maintain itself. This approach islikely to be the most successful andthe least expensive way to restoreand maintain the ecological integrityof flow-altered rivers (Stanford et al.1996). Although the most effectivemix of human-aided and natural re-covery methods will vary with theriver, we believe that existing knowl-edge makes a strong case that restor-ing natural flows should be a corner-stone of our management approachto river ecosystems.

Acknowledgments

We thank the following people forreading and commenting on earlierversions of this paper: Jack Schmidt,LouToth, Mike Scott, David Wegner,Gary Meffe , Mary Power , Kur tFausch, Jack Stanford, Bob Naiman,Don Duff , John Epi fan io , Lor iRobertson, Jeff Baumgartner, TimRandle, D a v i d Harpman, M i k eArmbruster, and Thomas Payne.Members of the Hydropower Re-form Coalition also offered construc-tive comments. Excellent final re-views were provided by Greg Auble,Carter Johnson, an anonymous re-viewer, and the editor of Bioscience.Robin Abel1 contributed to the de-velopment of the timeline in Figure5, and graphics assistance was pro-vided by Teresa Peterson (Figure 3),Matthew Chew (Figure 4) and RobinAbel1 and Jackie Howard (Figure 5).We also thank the national offices ofTrout Unlimited and American Riv-ers for encouraging the expression ofthe ideas presented here. We espe-cially thank the George Gund Foun-dation for providing a grant to holda one-day workshop, and The Na-ture Conservancy for providing lo-gistical support for several of theauthors prior to the workshop.

References cited

Abramovitz JN. 1996. Imperiled waters, im-poverished future: the decline of freshwa-

781

ter ecosystems. Washington (DC): World-watch Institute. Worldwatch paper nr 128.

Allan JD, Flecker AS. 1993. Biodiversitycon-servation in running waters. Bioscience43: 32-43.

Arthington AH, King JM, O’Keefe JH, BunnSE, Day JA, Pusey BJ, Bluhdorn DR, ThameR. 1991. Development of an holistic ap-proach for assessing environmental flowrequirements of riverine ecosystems. Pages69-76 in Pigram JJ, Hooper BA, eds. Waterallocation for the environment: proceed-ings of an international seminar and work-shop. University of New England Armidale(Australia): The Centre for Water PolicyResearch.

Auble GT, Friedman JM, Scott ML. 1994.Relating riparian vegetation to present andfuture streamflows. Ecological Applications4: 544-554.

BainMB,Finn JT, BookeHE. 1988. Streamflowregulation and fish community structure.Ecology 69: 382-392.

Barinaga M. 1996. A recipe for river recovery?Science 273: 1648-1650.

Beschta RL, Jackson WL. 1979. The intrusionof fine sediments into a stable gravel bed.Journal of the Fisheries Research Board ofCanada 36: 207-210.

Beven KJ. 1986. Hillslope runoff processes andflood frequency characteristics. Pages 187-202 in Abrahams AD, ed. Hillslope pro-cesses. Boston: Allen and Unwin.

Bogan AE. 1993. Freshwater bivalve extinc-tions (Mollusca: Unionida): a search forcauses. American Zoologist 33: 599-609.

Bovee KD, Milhous R. 1978. Hydraulic simu-lation in instream flow studies: theory andtechniques. Ft. Collins (CO): Office of Bio-logical Services, US Fish & Wildlife Service.Instream Flow Information Paper nr 5, FWS/OBS-78/33.

Bren L.J. 1992. Tree invasion of an intermittentwetland in relation to changes in the flood-ing frequency of the River Murray, Austra-lia. Australian Journal of Ecology 17: 395-408.

Brookes A. 1988. Channelized rivers, perspec-tives for environmental management. NewYork: John Wiley & Sons.

Busch DE, Smith SD. 1995. Mechanisms asso-ciated with decline of woody species inriparian ecosystems of the SouthwesternUS. Ecological Monographs 65: 347-370.

Castleberry DT, et al. 1996. Uncertainty andinstream flow standards. Fisheries 21: 20-21.

Chapman RJ, Hinckley TM, Lee LC, TeskeyRO. 1982. Imoact of water level changes onwoody riparian and wetland commumties.Vol. 10. Kearneysville (WV): US Fish &Wildlife Service. Publication nr OBS-82/83.

Chien N. 1985. Changes in river regime afterthe construction of upstream reservoirs.Earth Surface Processes and Landforms 10:143-159.

Gloss GP, Lake PS. 1996. Drought, differentialmortality and the coexistence of a nativeand an introduced fish species in a southeast Australian intermittent stream. Envi-ronmental Biology of Fishes 47: 17-26.

Collier M, Webb RH, Schmidt JC. 1996. Damsand rivers: primer on the downstream ef-fects of dams. Reston (VA): US GeologicalSurvey. Circular nr 1126.

Collier MP, Webb RH, Andrews ED. 1997.

782 BioScience Vol. 47 No. 11

Experimental flooding in the Grand Can-yon. Scientific American 276: 82-89.

Connor WH, Gosselink JC, Parrondo RT. 198 1.Comparison of the vegetation of three Loui-siana swamp sites with different floodingregimes. American Journal of Botany 68:320-331.

Copp GH. 1989. The habitat diversity and fishreproductive function of floodplain ecosys-tems. Environmental Biology of Fishes 26:l-27.

Cushman RM. 1985. Review of ecological ef-fects of rapidly varying flows downstreamfrom hydroelectric facilities. North Ameri-can Journal of Fisheries Management 5:330-339.

Danieis RB. 1960. Entrenchment of the willowdrainage ditch, Harrison County, Iowa.American Journal of Science 258: 161-176.

Duncan RP. 1993. Flood disturbance and thecoexistence of species in a lowland podocarpforest, south Westland, New Zealand. Jour-nal of Ecology 81: 403-416.

DunneT, Leopold LB. 1978. Water in Environ-mental Planning. San Francisco: W. H. Free-man and Co.

Echeverria JD, Barrow P, Roos-Collins R. 1989.Rivers at risk: the concerned citizen’s guideto hydropower. Washington (DC): IslandPress.

Faber S. 1996. On borrowed land: public poli-cies for floodplains. Cambridge (MA): Lin-coln Institute of Land Policy.

Fausch KD, Bestgen KR. 1997. Ecology offishes indigenous to the central and south-western Great Plains. Pages 131-166 inKnopf FL, Samson FB, eds. Ecology andconservation of Great Plains vertebrates.New York: Springer-Verlag.

[FERC] Federal Energy Regulatory Commis-sion. 1995. Relicensing the Ayers Islandhydroelectric project in the PemigewassedMerrimack River Basin. Washington (DC):Federal Energy Regulatory Commission.Final environmental impact statement,FERC Project nr 2456-009.

Fenner P, Brady WW, Patten DR. 1985. Effectsof regulated water flows on regeneration ofFremont cottonwood. Journal of RangeManagement 38: 135-138.

Finger TR, Stewart EM. 1987. Response offishes to flooding in lowland hardwoodwetlands. Pages 86-92 in Matthews WJ,Heins DC, eds. Community and evolution-ary ecology of North American stream fishes.Norman (OK): University of OklahomaPress.

Fisher SG. 1983. Succession in streams. Pages7-27 in Barnes JR, Minshall GW, eds. Streamecology: application and testing of generalecological theory. New York: Plenum Press.

Franklin JF. 1993. Preserving biodiversity: spe-cies, ecosystems, or landscapes? EcologicalApplications 3: 202-205.

Friedman JM, Scott ML, Lewis WM. 1995.Restoration of riparian forest using irriga-tion, artificial disturbance, and naturalseedfall. Environmental Management 19:547-557.

Gehrke PC, Brown P, Schiller CB, Moffatt DB,Bruce AM. 1995. River regulation and fishcommunities in the Murray-Darling riversystem, Australia. Regulated Rivers: Re-search & Management 11: 363-375.

Greenberg L, Svendsen P, Harby A. 1996. Avail-ability of microhabitats and their use by

brown trout (Salmo trutta) and grayling(Thymal[~s rhyma[ltrs) in rhe River Vojman,Sweden. Regulated Rivers: Research &Management 12: 287-303.

Halyk LC, Balon EK. 1983. Structure and eco-logical production of the fish taxocene of asmall floodplain system. Canadian Journalof Zoology 61: 2446-2464.

Hammer TR. 1972. Stream channel enlarge-ment due to urbanization. Water ResourcesResearch S: 1530-1540.

Harms WR, Schreuder HT, Hook DD, BrownCL, Shropshire FW. 1980. The effects offlooding on the swamp forest in LakeOklawaha,Florida,Ecology61: 1412-1421.

Hill MT, Platts WS. In press. Restoration ofriparian habirat with a multiple flow regimem the Owens River Gorge, California. Jour-nal of Restoration Ecology.

Hill MT, Platts WS, Beschta RL. 1991. Ecologi-cal and geomorphological conceprs forinstream and out-of-channel flow require-ments. Rivers 2: 198-210.

Helling CS, Meffe GK. 1996. Command andcontrol and the pathology of natural re-source management. Conservation Biology10: 328-337.

Horton JS. 1977. The development and per-petuation of the permanent tamarisk type inthe phreatophyte zone of the Southwest.USDA Forest Service. General TechnicalReport nr RM-43: 124-127.

Hughes FMR. 1994. Environmental change,disturbance, and regeneration in semi-aridf loodpla in fores ts . Pages 321-345 inMillington AC, Pye K, eds. Environmentalchange in drylands: biogeographical andgeomorphologica l perspect ives . NewYork: John Wiley & Sons.

Hughes RM, Noss RF. 1992. Biological diver-sity and biological integrity: current con-cerns for lakes and streams. Fisheries 17:11-19.

Hupp CR, Osterkamp WR. 1985. Bottomlandvegetation distribution alone. Passage Creek,Virginia, in relation to flu~ial landforms:Ecology 66: 670-681.

Johnson WC. 1994. Woodland expansion inthe Platte River, Nebraska: patterns andcauses. Ecological Monographs 64: 45-84.

Johnson WC, Burgess RL, Keammerer WR.1976. Forest overstory vegetation and envi-ronment on the Missouri River floodplainin North Dakota. Ecological Monographs46: 59-84.

Junk WJ, Bayley PB, Sparks RE. 1989. Theflood pulse concept in river-floodplain sys-tems. Canadian Special Publication of Fish-eries and Aquatic Sciences 106: 110-127.

Karr JR. 1991. Biological integrity: a long-neglected aspect of water resource manage-ment. Ecological Applications 1: 66-84.

Karr JR, Toth LA, Dudley DR. 1985. Fishcommunities of midwestern rivers: a historyof degradation. Bioscience 35: 90-95.

Keller EA, Swanson FJ. 1979. Effects of largeorganic material on channel form and flu-vial processes. Earth Surface Processes andLandforms 4: 351-380.

Kingsolving AD, Bain MB. 1993. Fish assem-blage recovery along a riverine disturbancegradient. Ecological Applications 3: 531-544.

Klotz JR, Swanson S. 1997. Managed instreamflows for woody vegetation recruitment, acase study. Pages 483-489 in Warwick J,

ed. Symposium proceedings: water resourceseducation, training, and practice: opportu-nities for the next century. American WaterResources Association, Universities Coun-cil on Water Resources, American WaterWorks Association; 29 Jun-3 Jul; Keystone,c o .

Knox JC. 1972. Valley alluviation in south-western Wisconsin. Annals of the Associa-tion of ,4merican Geographers 62: 401-410.

Kondolf GM, Curry RR. 1986. Channel ero-sion along the ‘Carmel River, MontereyCounty, California. Earth SurfaceProcessesand Landforms 11: 307-319.

Krapu GL, Facey DE, Fritzell EK, Johnson DH.1984. Habitat use by migrantsandhill cranesin Nebraska. Journal of Wildlife Manage-ment 48: 407-417.

Kupferberg SK. 1996. Hydrologic and geomor-phic factors affecting conservation of a river-breeding frog (Ram boylii). EcologicalApplications 6: 1332-1344.

Larson HN. 1981. New England flow policy.Memorandum, interim regional policy forNew England stream flow recommenda-tions. Boston: US Fish Br Wildlife Service,Region 5.

Leopold LB. 1968. Hydrology for urban landplanning: a guidebook on the hydrologiceffects of land use. Reston (VA): US Geo-logical Survey. Circular nr 554.

Leopold LB, Wolman MG, Miller JP. 1964.Fluvial processes in geomorphology. SanFrancisco: W. H. Freeman & Sons.

Ligon FK, Diet-rich WE, Trush WJ. 1995. Down-stream ecological effects of dams, a geo-morphic perspective. Bioscience 45: 183-192.

[LADWPl Los Angeles Department of WaterandPower. 1995. DraftMono Basinstreamand channel restoration plan. Los Angeles:Department of Water and Power.

Meffe GK. 1984. Effects of abiotic disturbanceon coexistence of predator and prey fishspecies. Ecology 65: 1525-1534.

Meffe GK, Minckley WL. 1987. Persistenceand stability of fish and invertebrate assem-blages in a repeatedly disturbed SonoranDesert stream. American Midland Natural-ist 117: 177-191.

LMiller AJ. 1990. Flood hydrology and geomor-phic effectiveness in the central Appala-chians. Earth Surface Processes and Land-forms 15: 119-134.

Minckley ML, Deacon JE, ed. 1991. Battleagainst extinction: native fish managementin the American West. Tucson (AZ): Uni-versity of Arizona Press.

Mincklev WL. Meffe GK. 1987. Differentialselecrion by flooding in stream-fish commu-nities of the arid American Southwest. Pages93-104 in Matthews WJ, Heins DC, eds.Community and evolutionary ecology ofNorth American stream fishes. Norman(OK): Universitv of Oklahoma Press.

Mitsch WJ? Rust WC. 1984. Tree growth re-sponses to flooding in a bottomland forestin northern Illinois. Forest Science 30: 499-510.

Molles MC, Crawford CS, Ellis LM. 1995.Effects of an experimental flood on litterdynamics in the Middle Rio Grande ripar-ian ecosystem. Regulated Rivers: Research& Management 11: 275-281.

Montgomer; WL, McCormick SD, Naiman

December 1997

RJ, Whoriskey FG, Black GA. 1983. Springmigratory synchrony of s&~onid, cato-stomid, and cyprinid fishes in Riviere j laTruite, Quebec. Canadian Journal of Zool-ogy 61: 2495-2502.

America: patterns and ecological impact.Pages 27-43 in Moonev HA. Drake IA. eds.Ecology of biological ‘invasions 0; North

Moore KMS, Gregory SV. 1988. Response ofyoung-of-the-year cutthroat trout to ma-nipulations of habitat structure in a small

America and Hawaii. New York: Springer-

stream. Transactions of the American Fish-eries Society 117: 162-170.

Verlag.

Morehardt JE. 1986. Instream flow method-ologies. Palo Alto (CA): Electric Power Re-search Institute. Report nr EPRIEA-4819.

Moyle PB. 1986. Fish introductions into North

Moyle PB, Light T. 1996. Fish invasions inCalifornia: do abiotic factors determinesuccess? Ecology 77: 1666-1669.

Naesie T, Jonsson B, Skurdal J. 1995. Springflood: a primary cue for hatching of riverspawning Coregoninae. Canadian Journalof Fisheries and Aquatic Sciences 52: 2190-2196.

Naiman RJ, Magnuson JJ, McKnight DM,Stanford JA. 1995. The freshwater impera-tive: a research agenda. Washington (DC):Island Press.

[NRC] National Research Council. 1992. Res-toration of aquatic systems: science, tech-nology,andpublicpohcy. Washington (DC):National Academy Press.

1996. Upstream: salmon and society inthe’pacific Northwest. Washington (DC):

National Academy Press.Nesler TP, Muth RT, Wasowicz AF. 1988.

Evidence for baseline flow spikes as spawn-ing cues for Colorado Squawfish in theYamaa River. Colorado. American Fisher-ies Society Symposium 5: 68-79.

Nilsen ET, Sharifi MR, Rundel PW. 1984.Comparative water relations of phreato-phytes in the Sonoran Desert of California.k&logy 65: 767-778.

Nilsson C. 1982. Effects of stream reeulationon riparian vegetation. Pages 931106 inLillehammer A;Saltveit SJ,eds. Regulatedrivers. New York: Columbia UniversityPress.

Perkins DJ, Carlsen BN, Fredstrom M, MillerRH. Rofer CM. Rueeerone GT. Zim-merman CS. 1984. Thi”effects of iround-water pumping on natural spring communi-ties in Owens Vallev. Pages 515-527 inWarner RE, Hendrix KMyeds. Californiaripariansystems: ecology, conservation, andproductive management. Berkeley (CA):University of California Press.

Petts GE. 1984. Impounded rivers: perspectivesfor ecological management. New York: JohnWiley & Sons.

- . i98S. Time scales for ecological con-tern in regulated rivers. Pages 2.57-266 inCraig JF, fernper JB, eds. Regulated streams:advances in ecology. New York: PlenumPress.

Pickett STA, Parker VT, Fiedler PL. 1992. Thenew paradigm in ecology: implications forconservation biology above the species level.Pages 66-88 in Fiedler PL, Jain SK, eds.Conservation biology. New York: Chapman& Hall.

Poff NL. 1996. A hydrogeography of unregu-

lated streams in the United States and anexamination of scale-dependence in somehydrological descriptors. Freshwater Biol-ogy 36: 101-121.

Poff NL, Allan JD. 1995. Functional organiza-tion of stream fish assemblages in relationto hydrological variability. Ecology 76: 606-627.

Poff NL, Ward JV. 1989. Implications ofstreamflow varia bility and predictability forlotic community structure: a regional analy-sis of streamflow patterns. Canadian Jour-nal of Fisheries and Aquatic Sciences 46:1805-1818.

Power ME. 1992. Hydrologic and trophic con-

-’

trols of seasonal algal blooms in northernCalifornia rivers. Archiv fur Hydrobiologie

1990. The physical habitat template of

125: 385-410.

lotic systems: recovery in rhe context ofhistorical pattern of spatio-temporal het-erogeneity. Environmental Management 14:629-646.

Power ME, Sun A, Parker M, Dietrich WE,Wootron IT. 1995. Hvdraulic food-chainmodels: an approach to the study of food-web dynamics in large rivers. Bioscience 45:159-167.

Prestegaard KL. 1988. Morphologicaicontrolsonsediment delivery paihwayi. Pages 533-540 in Walling DE. ed. Sediment budgets.Wallingford (UK): IAHS Press. Internati&alAssociation of Hydrological Sciences Publi-cation nr 174.

Prestegaard KL, Matherne AM, Shane B,Houghton K, O’Connell M, Katyl N. 1994.Spatial variations in the magnitude of the1993 floods, Raccoon River Basin, Iowa.Geomorphology 10: 169-182.

Reeves GH, Benda LE, Burnett KM, Bisson PA,Sedell JR. 1996. A disturbance-based eco-systemapproach to maintaining and restor-ing freshwater habitats of evolutionarilysignificant units of anadromous salmonidsin the Pacific Northwest. American Fisher-ies Society Symposium 17: 334-349.

Reily PW, Johnson WC. 1982. The effects ofaltered hydrologic regime on tree growthalong the Missouri River in North Dakota.Canadian Journal of Botany 60: 2410-2423.

Resh VH, Brown AV, Covich AP, Gurtz ME, LiHW, Minshall GW, Reice SR, Sheldon AL,Wallace JB, Wissmar R. 1988. The role ofdisturbance in stream ecology. Journal ofthe North American Benthological Society7: 433-455.

Richter BD, Baumgartner JV, Powell J, BraunDP. 1996. A method for assessing hydro-logic alteration within ecosystems.Conser-vation Biology 10: 1163-1174.

Richter BD, Baumgartner JV, Wigington R,Braun DP. 1997. How much water does ariver need? Freshwater Biology 37: 231-249.

Robertson L. In press. Water operations onthe Pecos River, New Mexico and thePecos bluntnose shiner, a federally-listedminnow, US Conference on Irrigation andDrainage Symposium.

Rood SB, Mahoney JM. 1990. Collapse ofriparian poplar forests downstream fromdams in western prairies: probable causesand prospects for mitigation. Environ-mental Management 14: 451-464.

Rood SB, Mahoney JM, Reid DE, Zilm L.1995. Instream flows and the decline of

783

riparian cortonwoods along the St. .LlaryRiver, Alberta. Canadian Journal of Botany73: 1250-1260.

Ross ST, Baker JA. 1983. The response of fishesto periodic spring floods in a southeasternstream. American Midland Naturalist 109:l -14 .

Rulifson RA, Manooch CS III, eds. 1993. Roa-noke River water flow committee report for1991-1993. Albemarle-Pamlico estuarinestudy. Raleigh (NC): US Environmental Pro-tection Agency. Project nr APES 93-18.

Scheidegger KJ, Bain MB. 1995. Larval fish innatural and regulated rivers: assemblagecomposition and microhabitat use. Copeia1995: 125-135.

Schlosser IJ. 1985. Flow regime, juvenile a%un-dance, and the assemblage structure ofstream fishes. Ecology 66: 1484-1490.

1 9 9 0 . E n v i r o n m e n t a l v a r i a t i o n , l i f ehistory attributes, and community struc-ture in stream fishes: implications for envi-ronmental management assessment. Envi-ronmental Management 14: 621-628.

Scott ML, Friedman JM, Auble GT. 1996.Fluvial processes and the establishment ofbottomland trees. Geomorphology 14: 327-339.

Scott, ML, huble GT, Friedman JM. 1997.Flood dependencv of cottonwood establish-ment along the IMissouri River, Montana,USA. Ecolouical Aoolications 7: 677-690.

Sear DA. 1995.YMorph’dlogical and sedimento-logical changes in a gravel-bed river follow-ing 12 years of flow regulation for hydro-power. Regulated Rivers: Research &Management 10: 247-264.

Seegrist DW, Gard R. 1972. Effects of floods ontrout in Sagehen Creek. California. Trans-actions of the American Fisheries Society101: 478482.

Shankman D. Drake DL. 1990. Channel mirrra-tion and regeneration of bald cypress inwestern Tennessee. Physical Geography 11:343-352.

Sidle JC, Carlson DE, Kirsch EM, Dinan JJ.1992. Flooding mortalin, and habitat re-newal for least terns and piping plovers.Colonial Waterbirds 15: 132-136.

Southwood TRE. 1977. Habitat, the templetfor ecological strategies? Journal of AnimalEcology 46: 337-365.

Sparks RE. 1992. Risks of altering the hydro-logic regime of large rivers. Pages 119-152in Cairns J, Niederlehner BR, Orvos DR,teds. Predicting ecosystem risk. Vol XX.Advances in modern environmental toxi-cology. Princeton (NJ): Princeton Scientificpublishing Co.

- - . 1995. Need for ecosystem manage-ment of large rivers and their floodplains.Bioscience 45: 168-182.

Stalnaker CB. 1990. Minimum flow is a myth.pages 31-33 in Bain MB, ed. Ecology andassessment of warmwater streams: work-shop synopsis. Washington (DC): US Fish &Wildlife Service. Biological Report nr 90(5).

Stalnaker C, Lamb BL, Henriksen J, Bovee K,Bartholow J. 1995. The instream flow in-cremental methodology: a primer for IFIM.Ft. Collins (CO): National Biological Ser-vice, US Department of the Interior. Bio-logical Report nr 29.

Stanford JA. 1994. Instream flows to assist therecovery of endangered fishes of the UpperColorado River Basin. Washington (DC):US Department of the Interior, NationalBiological Survey. Biological Report nr 24.

Stanford JA, Ward JV, Liss WJ, Frissell CA,Williams RN, Lichatowich JA, Coutant CC.1996. A general protocol for restoration ofregulated rivers. Regulated Rivers: Research& Management 12: 391-414.

Stevens LE, Schmidt JC, Brown BT. 1995. Flowregulation, geomorphology, and ColoradoRiver marsh development in the GrandCanyon, Arizona. Ecological Applications5: 1025-1039.

Strange EM, Moyle PB, Foin TC. 1992. Inter-actions between stochastic and determinis-tic processes in scream fish community as-sembly. Environmental Biology of Fishes36: l-15.

SrrengDR,Glitzenstein JS,HarcombePA. 1989.Woody seedling dynamics in an East Texasfloodplain forest. Ecological Monographs59: 177-204.

Stromberg JC. Tress JA, Wilkins SD, Clark S.1992. Response of velvet mesquite togroundwater decline. Journal of Arid Envi-ronments 23: 45-58.

Stromberg JC, Tiller R, Richter B. 1996. Effectsof groundwater decline on riparian vegeta-tion of semiarid regions: the San PedroRiver, Arizona, USA. Ecological Applica-tions 6: 113-131.

Stromberg JC, Fry J, Patten DT. 1997. Marshdevelopment after large floods in an alluvial,arid-land river. Wetlands 17: 292-300.

Taylor DW. 1982. Eastern Sierra riparian veg-etation: ecological effects of stream diver-sion. Mono Basin Research Group Contri-bution nr 6, Report to Inyo National Forest.

[TNC] The Nature Conservancy. 1996. Troubledwaters: protecting our aquatic heritage. Ar-lington (VA): The Nature Conservancy.

Tilman D, Downing JA, Wedin DA. 1994.Does diversity beget stability? Nature 371:257-264.

Toth LA. 1995. Principles and guidelines forrestoration of river/floodplain ecosystems-Kissimmee River, Florida. Pages 49-73 inCairns J, ed. Rehabilitating damaged eco-sysrems. 2nd ed. Boca Raton (FL): LewisPublishers/CRC Press.

Travnichek VH, Bain MB, Maceina MJ. 1995.Recovery.of a warmwater fish assemblageafter the initiation of a minimum-flow re-lease downstream from a hydroelectric dam.Transactions of the American Fisheries So-ciety 124: 836-844.

Trepanier S, Rodriguez MA, Magnan P. 1996.Spawning migrations in landlocked Atlan-tic salmon: time series modelling of riverdischarge and water temperature effects.Iournal of Fish Bioloev 48: 925-936.

Turner JL, Chadwick HI?. 1972. Distributionand abundance of young-of-the-year stripedbass, Morone saxatilis, in relation to riverflow in the Sacramento-San Joaquin estu-ary. Transactions of the American FisheriesSociety 101: 442-452.

Tvus HM. 1990. Effects of altered stream flowson fishery resources. Fisheries 15: 18-20.

[USCOE] US Army Corps of Engineers, LosAngeles District. 1996. Reconnaissance re-

port, rcvlew of existing project: Alnmc Like,Arizona.

Valentin S, Wasson JG, Philippe M. 1995.Effects of hydropower peaking on epilithonand invertebrate community trophic struc-ture. Regulated Rivers: Research & .Man-agement 10: 105-l 19.

Van Winkle W, Rose KA, Chambers RC. 1993.Individual-based approach to fish popula-tion dynamics: an -overview. Transactionsof the American Fisheries Society 122: 397-403.

Walker KF, Sheldon F, Puckridge JT. 199.5. Aperspective on dryland river ecosystems.Regulated Rivers: Research & Management11: 85-104.

Wallace JB, Benke AC. 1984. Quantification ofwood habitat in subtropical coastal plainsstreams. Canadian Journal of Fisheries andAquatic Sciences 41: 1643-1652.

Ward IV. Stanford IA. 1979. The ecoloqv ofI I _.

regulated streams. New York:Plenum Press.Weisberg SB, Janicki AJ, Gerrirsen J, Wilson

H T . 1 9 9 0 . Enhancemenr o f b e n t h i cmacroinvertebrates by minimum flow froma hydroelecrric dam. Regulated Rivers: Re-search s( Management 5: 265-277.

WelcommeRL. 1979,Fisheriesecologyofflood-plain rivers. New York: Longman.

-.1992. Riverconservation-futurepros-pects. Pages 454-462 in Boon PJ, Calow P,Petts GE, eds. River conservation and man-agement. New York: John Wiley s( Sons.

Welcomme RL, Hngborg D. 1977. Towards amodel of a floodplain fish population andits fishery. Environmental Biology of Fishes2: 7-24.

Wharton CH, Lambou VW,Newsome J, WingerPV, Gaddy LL, Mancke R. 1981. The faunaof bottomland hardwoods in the sourheast-ern United States. Pages 87-160 in ClarkJR, Benforado J, eds. -Wetlands of bottom-land hardwood forests. New York: ElsevierScientific Publishing Co.

Williams JG. 1996. Lost in space: minimumconfidence intervals for idealized PHABSIMstudies. Transactions of the American Fish-eries Society 125: 458-465.

Williams DD, Hynes HBN. 1977. The ecologyof temporary streams. II. General remarkson temporary streams. Internationale Revuedes gesampten Hydrobiologie 62: 53-61.

Wilhams GP, Wolman MG. 1984.Downstreameffects of dams on alluvial rivers. Reston(VA): US Geological Survey. ProfessionalPaper nr 1286.

Williams RN, Calvin LD, Coutant CC, ErhoMW, Lichatowich JA, Liss WJ, McConnahaWE, Mundy PR, Stanford JA, Whitney RR.1996. Return to the river: restoration ofsalmonid fishes in the Columbia River eco-system. Portland (OR): Northwest PowerPlanning Council.

Wolman MG, Gerson R. 1978. Relative scalesof time and effectiveness of climate in wa-tershed geomorpholdgy. Earth Surface Pro-

cesses and Landforms 3: 189-208.Wolman MG, Miller JP. 1960. Magnitude and

frequency of forces in geomorphic processes.Journal of Hydrology 69: 54-74:

Woorton JT, Parker MS, Power ME. 1996.Effects of disturbance on river food webs.Science 273: 1555-1561.

784 BioScience Vol. 47 NO. 11