effectiveness of large woody debris in stream...

Ecological Engineering 18 (2001) 211–226 www.elsevier.com/locate/ecoleng

Effectiveness of large woody debris in stream rehabilitationprojects in urban basins

Marit G. Larson a,*, Derek B. Booth b, Sarah A. Morley c

a City of New York Department of Parks and Recreation, 1234 Fifth A�enue, Room 234, New York, NY 10029, USAb Center for Urban Water Resources Management, Department of Ci�il and En�ironmental Engineering,

Uni�ersity of Washington Box 35270, Seattle, WA 98195-2700, USAc Northwest Fisheries Science Center, National Marine Fisheries Ser�ice, 2725 Montlake Boule�ard East, Seattle,

WA 98112-2097, USA

Received 18 July 2000; received in revised form 23 January 2001; accepted 16 February 2001

Abstract

Urban stream rehabilitation projects commonly include log placement to establish the types of habitat featuresassociated with large woody debris (LWD) in undisturbed streams. Six urban in-stream rehabilitation projects wereexamined in the Puget Sound Lowland of western Washington. Each project used in-stream log placement as theprimary strategy for achieving project goals; none included systematic watershed-scale rehabilitation measures. Theeffectiveness of LWD in these projects was evaluated by characterizing physical stream conditions using commonmetrics, including LWD frequency and pool spacing, and by sampling benthic macroinvertebrates. In all projectreaches where pre-project data existed, pool spacing narrowed after LWD installation. All project sites exhibitedfewer pools for a given LWD loading, however, than has been reported for forested streams. In project reaches wherethe objective was to control downstream sedimentation, only limited success was observed. At none of the sites wasthere any detectable improvement in biological conditions due to the addition of LWD. Our results indicate that,although LWD projects can modestly improve physical habitat in a stream reach over a time scale of 2–10 years, theyapparently do not achieve commensurate improvement in biological conditions. © 2001 Elsevier Science B.V. Allrights reserved.

Keywords: Stream restoration; Urban streams; Stream monitoring; Large woody debris; Watershed restoration; Biological monitor-ing; Invertebrate monitoring

1. Introduction

Urban streams in the Puget Sound Lowland(PSL) in northwest Washington state are uniquein that many still support salmon, although indeclining numbers. With the listing of PugetSound Chinook (Oncorhynchus tshawytscha) as

* Corresponding author. Tel.: +1-212-3601415; Fax: +1-212-3601426.

E-mail address: [email protected] (M.G. Larson).

0925-8574/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved.

PII: S0925-8574(01)00079-9

M.G. Larson et al. / Ecological Engineering 18 (2001) 211–226212

‘endangered’ under the federal Endangered Spe-cies Act in 1999 — the first time a species hasbeen listed within a major metropolitan area — agreater importance has been placed on reversingurban impacts on waterways in the PSL. Forthis reason, engineered projects now being de-signed and constructed in urban streams are in-creasingly required to benefit salmon habitat, evenif they are motivated primarily by more tradi-tional concerns such as flood control or propertyprotection.

In-stream rehabilitation projects are commonlybuilt in response to problems that result fromboth local sources and diffuse watershed degrada-tion. Local problems, such as an improperly sizedculvert, are relatively easily identified and cor-rected. Reversing the consequences of watersheddegradation, such as channel widening and inci-sion, is much more difficult because conditionsthat led to stream degradation usually remainunchecked. Yet, because of numerous recognizedproblems on streams in urban or urbanizingbasins, the interest of local communities in restor-ing the amenities these streams provide (MacDon-ald, 1995; Riley, 1998), and the relative ease andeconomy of site-specific in-stream work, largeamounts of money are being spent on in-streamprojects.

While the number of urban stream restorationprojects is increasing across the nation, compara-tively few examples of monitoring programs de-signed to measure the success of such efforts aredocumented (Kerschner, 1997). In western Wash-ington, salmon recovery is an objective of themajority of rehabilitation projects, yet very rarelyare salmon or any other element of stream biotadirectly monitored to assess restoration success.Of over 300 rehabilitation projects installed overthe past decade in urban areas around the PugetSound Lowland, less than 5% were evaluated withany sort of biological or habitat suitability data(Kropp, 1998).

This study investigates the effectiveness of onecommon technique, placement of in-stream largewoody debris (LWD), to reverse local effects ofwatershed degradation in the absence of any sys-tematic watershed-scale approaches to rehabilita-tion. To accomplish this, six stream rehabilitation

projects in western Washington that employLWD were examined with the objective of an-swering the following questions.� Does in-stream placement of LWD produce

physical channel characteristics typical ofstreams in less disturbed watersheds?

� Does biological condition improve immediatelydownstream of LWD addition?

� How can LWD project designs be improved?� Does watershed disturbance exert greater con-

trol over physical and biological recovery ofthe channel than the local in-channel condi-tions addressed by added LWD?

1.1. Impacts of urbanization

Because of its permanence, urbanization posesa particularly ominous obstacle for stream protec-tion and rehabilitation. The process of urbaniza-tion usually results in irreversible changes to theland surface and drainage pattern by increasingthe impervious area, reducing vegetation cover,compacting soil, reducing areas of depressionstorage, concentrating and re-routing runoff, andstraightening and piping streams. These changescause a progressively greater fraction of precipita-tion to enter the channel rapidly as Horton over-land flow (Leopold, 1968; Graf, 1975; Hollis,1975). Particularly in geologic regions such as thePSL, where overland runoff is rarely generatedand subsurface flow predominates, these changescan alter stream channels (Harr, 1976; Booth,1990, 1991). Direct alterations to channels andriparian corridors include removing bank vegeta-tion, clearing woody debris from streams, armor-ing stream beds and banks, straighteningchannels, reducing floodplain storage, and divert-ing extensive stretches of streams through cul-verts. Removing LWD, in particular, can modifychannel morphology and increase sediment dis-charge, and has led to the widespread loss ofonce-prevalent wood in urban channels (fore ex-ample, Booth et al., 1997; Horner et al., 1997).

1.2. The use of LWD in channel enhancement

LWD plays prominent roles in regulating mor-phology and habitat in Pacific Northwest chan-

M.G. Larson et al. / Ecological Engineering 18 (2001) 211–226 213

nels, from steep and narrow headwater streams towide low-gradient rivers. These functions make ita critical component of current stream and riverrestoration, and enhancement efforts. The effectsof LWD on moderate-gradient streams (0.5–4%slope) with bankfull widths of about 4–40 m havebeen studied in greatest detail. The steeper ofthese channels, dominated by riffle and glide fea-tures, are classified as ‘plane-bed’ by Montgomeryand Buffington (1998). Lower gradient streamscommonly display regularly alternating riffles andpools in a meandering planiform. The transitionfrom ‘plane-bed’ to these flatter ‘pool-riffle’ chan-nels, under a particular sediment-supply and flowregime, can be controlled by the presence of ob-structions, notably LWD, that form scour pools,bars, sediment storage sites, and steps in channelsthat would otherwise maintain a relatively flatand uniform bed. In these channels, LWD caninfluence bank stability, pool and bar formation,sediment retention, and grade (Montgomery etal., 1995; Beechie and Sibley, 1997; Nelson, 1998).

The reported effects of re-introducing LWD (orother physical habitat structures) to a channel,particularly when watershed conditions have notbeen simultaneously addressed, is ambiguous.Several studies have reported that, at least in theshort term, such habitat elements as pools, sub-strate for fish spawning or invertebrate colonizers,and cover for fish can be improved using in-stream structures alone (House and Boehne, 1986;Gortz, 1998; Hilderbrand et al., 1998; Shields etal., 1998). Some studies have shown that macroin-vertebrate community structure changes and di-versity increases when structures are added(Hilderbrand et al., 1997; Gortz, 1998), but theyalso report ambiguous responses in fish usage(Shields et al., 1998). In forested watersheds, stud-ies have reported higher fish counts after habitatstructures were installed in forested streams(House and Boehne, 1986; Crispin et al., 1993). Incontrast, a comprehensive evaluation of in-streamrehabilitation projects (Frissell and Nawa, 1992)documented a high incidence of structure failureafter floods of no more than a 10-year recurrenceinterval, and it emphasized the local and largelyshort-term benefits of LWD addition. House(1996) found similar shortcomings of in-stream

structures. No studies have yet evaluated whetherwood placed in urban streams can provide therange of functions observed in natural streams,such as increasing organic matter retention andattenuating the geomorphic effects of high flows.

2. Approach and methods

2.1. Study sites

Six in-stream rehabilitation projects were cho-sen for this study based on their location inwatersheds with urban development, their use ofin-stream log placement as a primary strategy forachieving local rehabilitation goals, and their in-clusion of fish habitat enhancement as an objec-tive (Fig. 1). The effectiveness of LWD in theseprojects was evaluated according to criteria forphysical improvements, for biological response,and for meeting stated design performancestandards.

The six evaluated projects lay in physicallysimilar watersheds but with widely different levelsof human disturbance (Table 1). But for the useof anchored LWD (i.e. buried or cabled) in themost developed basins, individual project charac-teristics were otherwise unrelated to watershed orstream characteristics. The streams in this studyall had perennial flow and were classified as allu-vial plane-bed or pool-riffle channels (Mont-gomery and Buffington, 1998).

The six rehabilitation projects included exam-ples of both anchored and unanchored LWD, aswell as different types of LWD, such as smoothlogs, root wads, and tops of trees. Five of theprojects had been built within 4 years of thisstudy; the project at Forbes Creek was 10 yearsold (Table 2). Project lengths ranged from 210 to430 m, except at Soosette Creek where wood wasplaced over a 1430 m long reach. Specific projectobjectives varied. However, all but one projectidentified ‘habitat enhancement’ as a primary orsecondary goal, and most had some flood andsediment control objective. The degree to whichthe stream channels were constrained by resi-dences, roads, or bank armor varied greatly be-tween projects.


2.2. Analysis

Development intensity in the watershed areacontributing to each project was determined fromgeographical information system (GIS) land-coverdata layers based on 1995 Landsat satellite im-agery, classified by King County Land and WaterResources Division (Jeff Burke, personal commu-nication, 1998) at 30 m resolution. For this analy-sis, land cover was considered developed ifclassified as ‘high-intensity’, ‘medium-intensity’,or ‘low-intensity’ development; or ‘bare rock/con-crete’, ‘bare ground/asphalt’, or ‘recently cleared’land. The percentage of developed land in eachbasin upstream of the downstream end of theproject was calculated using ArcView GIS.

To determine whether projects were successfulin improving physical and biological conditions,monitoring was conducted in reaches at and up-stream of where LWD was added. At the threeprojects where pre-project physical data existed(Laughing Jacobs, Soosette, and Swamp Creeks),

pre- and post-project stream conditions were com-pared. For projects without pre-project data,post-project conditions were compared with datareported in the literature for Puget Sound Low-land streams (Montgomery et al., 1995; May,1996; Karr and Chu, 2000) and/or to reachesupstream of each project. At most sites, the up-stream reaches varied somewhat from pre-projectconditions due to differences in valley width,slope, or encroachment, but all were geomorphi-cally similar.

Two sections of stream, at least 20 times thebankfull width, were surveyed in the project andjust upstream of the project on each stream. Allpieces of wood greater than 10 cm diameter andlonger than 1 m in any portion of the bankfullchannel were counted, based on criterion forLWD used by Montgomery et al. (1995), Green-berg (1995), May (1996). Root wads greater than0.02 m3 in volume were also counted. The diame-ter and length of every piece was estimated; everyfive to ten pieces the lengths and widths were

Fig. 1. Study location: western Washington state and the six rehabilitation project watersheds.


Table 1Watershed and stream characteristics

Thornton Swamp Hollywood Hills Laughing Jacob’s SoosetteForbes

Stream characteristicsAverage bankfull width (m) 3.5 5.4 10.4 4.1 6.4 8.7

0.006 0.005 0.046 0.028 0.019Bed slope 0.03718 14 1114 39Median grain size (mm) 51

3.5Upstream drainage (km2) 25.4 53.6 2.2 11.1 13

Percent upstream de�elopment93 72 6282 52Watershed (%) 58

70Riparian buffer (100 m) (%) 75 47 44 43 34 (45)a

45Basin relief (m) 45 150 50 40 400.005 0.003 0.002 0.0080.009 0.008Basin gradient (relief/basin length)

a Percent development in riparian buffer of the upstream end of the project.

measured with a tape to calibrate the visual esti-mate. ‘Key pieces’ were also identified; they arethose pieces of LWD defined as being indepen-dently stable within the bankfull channel (i.e. notheld or trapped by other material) and retainingor having the ability to retain other LWD(WFPB, 1997). The position of LWD in the pro-ject reaches with respect to the directions of flowwas sketched in the field, as was any obviousassociation with bank scour.

Where pre-project data were available, themovement of LWD in the project reaches wasestimated. Photographic documentation, sketched‘as-built’ positions of logs, notes from longitudi-nal surveys, or notes on aerial photographs, wereused to ascertain the mobility of the wood. Toevaluate the flow conditions under which anydocumented LWD movement occurred, the mag-nitude of the largest flow after the project installa-tion was determined from available gauges.

Residual pool depths (RPDs), the differencebetween depth of water in the pool and at the topof the downstream riffle (Lisle, 1987), were mea-sured in the field and calculated from longitudinalthalweg surveys. Only depressions with a RPD ofat least 25% of the bankfull depth and a minimumpool length at least 10% of the bankfull widthwere included in the final tally of pools (Mont-gomery et al., 1995). Pre-project information onpool numbers was taken from the Fish HabitatRelation surveys conducted by the King CountyWater and Land Resources Division at three of

the projects. These counts were converted to anaverage pool spacing over each measured reachand expressed as the distance between pools inunits of bankfull channel widths. This alloweddirect comparisons between streams in this studyand those in previously published studies.

The influence of LWD on the formation ofpools was determined by field observation. Twocategories were identified: pools formed by woodand pools formed by some other mechanisms(such as scour around a boulder or lateral scouragainst an armored bank). The number of piecesof wood associated with a given pool was alsocounted.

The influence of LWD on in-channel sedimentstorage was described by characterizing bars aseither ‘self-formed’ if no LWD was present, or‘associated with LWD’ if they were covered withor located immediately upstream or downstreamof LWD. Where LWD was trapping sediment, thedepth of the stored sediment was estimated bymeasuring the height of the step created by theLWD, and the total volume was estimated assum-ing the sediment was deposited as a wedge behindthe LWD. The overall volume of sediment in thereach was estimated by measuring the total vol-ume of all alluvial bars in the channel.

The influence of LWD in controlling grade in agiven reach was estimated using a longitudinalprofile of the channel thalweg. Steps in the longi-tudinal profile occurring at locations of LWDwere specified in the survey. The change in the

M.G

.L

arsonet

al./E

cologicalE

ngineering18

(2001)211

–226

216

Table 2Project characteristics

Unanchored LWDAnchored LWDProject characteristics

SoosetteForbes Laughing Jacob’sHollywood HillsSwampThornton

1994Year constructed 1988 1997 1997 1996 19951430370 430Project length (m) 280210 240

Anchored asCabled and in Unanchored, by Unanchored, byCabled and in Unanchored, byLWD placementcranedeflectors craneweirsweirs helicopter

80 28030048Number of pieces of LWD 2518added

$280c$350b $120c$160cNA $580dApproximate cost ($/m)a

Project objecti�esX XFlood control X X XX X XXSediment/erosion control X

XHabitat enhancement X XXX

a Costs based on preliminary estimates of construction costs divided by project length.b Source: Parametrix (1988).c Source: KCDNR (1995, 1996, 1997).d John Bethe, KCDNRl, personal communmication.


water surface elevation at LWD steps weresummed and divided by the total drop in eleva-tion over the reach to yield the percent of theelevation change controlled by LWD.

Biological condition at project sites was as-sessed with the benthic index of biological in-tegrity (B-IBI) (Karr and Chu, 1999). The B-IBI iscomposed of ten metrics of taxa richness andevenness, disturbance tolerance, and life historyattributes of benthic invertebrates. The B-IBIranges in value from 10 to 50 and can detect fivecategories of watershed condition, from undis-turbed to highly degraded (Doberstein et al.,2000).

Benthic invertebrates were collected at four ofthe rehabilitation projects. Monitoring sites werelocated immediately upstream and downstream ofeach restoration project. These paired sites wereselected to be as similar as possible in all regardsexcept for the presence of added LWD. All butHollywood Hills were sampled in 1998, and twoof the more recently completed projects (Swampand Laughing Jacob’s Creeks) were re-sampled in1999.

Invertebrates were collected from each site inSeptember when flows are typically stable andtaxa richness is high (Fore et al., 1996). At each

site, a Surber sampler (500-�m mesh, 0.1 m2

frame) was used to collect three samples along themid-line of a riffle (see Morley, 2000). Insectnymphs and larvae were identified to genus wherepractical; non-insect taxonomic identificationvaried from family to phylum. Following proce-dures first for fish (Karr et al., 1986), and then forinvertebrates (Kerans and Karr, 1994; Fore et al.,1996), values from each of the ten metrics weresummed to calculate a B-IBI score at each site.

3. Results

3.1. Installed LWD frequency, size, and mobility

LWD frequency was used to compare physicalconditions in-stream and upstream of the projectreach to those of geomorphically similar referencestreams (Fig. 2). The forested reference streamsallowed comparison with pre-urban conditions;urban reference streams allowed a comparisonwith a wider range of disturbed conditions thanrepresented by the six study sites.

In the study reaches, LWD loadings werehighest — in the range of least degraded urbanstreams — in unanchored projects (Fig. 3). Only

Fig. 2. Pool spacing versus LWD frequency in urban and forested streams in northwest Washington state (data from Montgomeryet al. 1995; May 1996).


Fig. 3. Pool spacing versus LWD frequency: comparison over space (project and upstream sites) and time (pre- and post-project).Forbes, Thornton, and Swamp Creeks have anchored LWD. At the other streams, LWD is not anchored.

one of the streams (Hollywood Hills), however,had a LWD frequency considered ideal for anatural stream (Bisson et al., 1987; WFPB, 1997).At the projects where the wood was anchored,LWD loadings were lower and typical of moder-ate to highly degraded urban streams and clear-cuts.

The size of added LWD varied greatly betweenprojects but showed no relationship either to thesize of the project stream or to whether the LWDwas anchored (Fig. 4). In particular, no logs ofsufficient size to be ‘key’ pieces were added to thewidest stream (Soosette) where LWD was notanchored, or to the widest streams (Swamp)where LWD was anchored. However, in thesmallest stream, Hollywood Hills, roughly fourtimes the number of ‘key’ pieces was added thantypically found in forested streams (Bisson et al.,1987; WFPB, 1997).

At all projects, typically less than one-third ofthe added LWD came in contact with the low-flow channel or obstructed at least one-third ofthe channel width. Where LWD was anchored, asat Swamp Creek, or was large, as at Hollywood

Hills, a higher fraction of the LWD was in thelow-flow channel and did obstruct flow. WhereLWD was the most mobile (at Soosette Creek),less than one-fifth of the pieces obstructed morethan 30% of the flow.

Since the projects were installed, all experienceddischarges greater than 2-year flows, and threesites experienced approximately 10-year flows. Noanchored LWD moved at any of the projectreaches and, where over 50% of the unanchoredLWD were key pieces (Hollywood Hills), therewas also no significant LWD movement. In theprojects with unanchored LWD and few or nokey pieces, however, LWD movement was docu-mented. At Swamp Creek, two unanchored logswere transported out of the project reach. AtLaughing Jacob’s Creek, numerous pieces ofLWD were transported tens of meters down-stream, moved across the channel, or abandonedon adjacent banks. At Soosette Creek, dozens ofsmooth logs (as well as logs with branches drilledinto them, ostensibly to mimic the form of ‘real’trees) moved several tens of meters and severaldebris piles re-mobilized.


3.2. Pool spacing, formation, and depth

Post-project pool spacing was not correlated toLWD loading (Fig. 3). At three sites (HollywoodHills and Laughing Jacob’s and Soosette creeks),pool spacing was wide compared with the leastdegraded urban streams with the same amount ofLWD. Forbes and Swamp creeks, where theLWD was anchored, had pool spacings most sim-ilar to those found in forested and least-degradedurban streams despite low LWD frequencies. Inthe three project reaches where pre-project dataexisted, pool spacing narrowed after LWD fre-quency was added.

Where there was no pre-project information onpools, the influence of added LWD on pools wasevidenced by the high proportion of pools formedby added LWD. In each of the project reaches,50–80% of the pools were formed by addedLWD, which was comparable with forestedstreams (Montgomery et al., 1995; Beechie andSibley, 1997; Nelson, 1998). More of the addedLWD was associated with pools where the LWDwas anchored (30–70%) than projects whereLWD was unanchored (15–18%). In forestedstreams, 20–40% of LWD was associated withpools, suggesting that unanchored LWD in urbanchannels is not having an identical geomorphiceffect as in undisturbed watersheds. Both types ofLWD addition raised pool numbers, at least

slightly, towards those of less disturbed streams.There was little difference in pool depth betweenpools formed by LWD or by other obstructions,between plunge or scour pools, or between poolsformed by natural LWD or added LWD.

3.3. Sediment storage and grade control

About one-third of the estimated in-channelsediment storage was associated with LWD atmost sites, although LWD generally did not retainsediment in the form of a discrete wedge. In allbut one stream, sediment storage associated withLWD increased by 50–100% where LWD fre-quency increased. Added LWD contributed mostto grade control (11–23%) on the highest gradientstreams where wood spanned the full width of thechannel, but it contributed little to grade controlon the low-gradient streams. Although severalprojects sought to reduce downstream sedimenta-tion by retaining sediment in the reach, onlylimited success was observed. LWD can retainsediment in some channel positions, but this stor-age was exceeded by high sediment loads.

3.4. Benthic index of biological integrity

Addition of LWD had little demonstrable effecton biological condition (see Fig. 5). Two projectshad been in place only 1 year when this study was

Fig. 4. Size distribution of added LWD and upstream natural LWD, listed from smallest to largest stream (no LWD was foundupstream at Thornton Creek).


Fig. 5. Paired B-IBI scores at projects. Scores must differ by atleast four points to conclude that sites are significantly differ-ent (Doberstein et al. 2000).

3.5. E�aluation of project design

Limited project effectiveness may result whenprojects are built with an inadequate design, or ifappropriate designs yield inadequate results. Toweigh these two possibilities, criteria for the de-sign of in-stream structures using LWD were ex-tracted from research on the function of LWD inforested streams (Table 3). For example, studieshave investigated the size ranges of stable logs ata given stream width (Bilby, 1984; Bisson et al.,1987; Hilderbrand et al., 1998), the obstructionangle associated with pool formation (Lisle, 1986;Cherry and Bilby, 1989; Gippel et al., 1996), therelationship between LWD frequency and poolspacing (Montgomery et al., 1995; Beechie andSibley, 1997), the mechanics of log jam stability(Abbe et al., 1997), and the hydraulics of logspacing (Gippel et al., 1996).

These ‘design criteria’, inferred from researchon LWD, were compared with conditions at thesix rehabilitation projects to evaluate the successof each project in meeting them (Table 4). Meet-ing design criteria, however, is worthwhile only ifthose criteria are relevant to attaining actual pro-ject objectives. For example, ‘obstruction width’was an important factor in pool formation atmost sites; when that design criterion was met(shaded box in Table 4), this objective was alsousually met (+ ). Conversely, ‘key piece fre-quency’ was also important; at most sites thiscriterion was not met (white box, Table 4) andnor were the associated objectives (− ).

Yet most criteria are not sufficient to ensurethat specific project objectives will be met. Forexample, the implied criteria for appropriatecross-bed position and burial of LWD wereachieved at several projects (shaded box, Table 4)but the project objective of sediment retentionwas not (− ). The same was true of log loading,where even adequate amounts of LWD did notensure that all project objectives were achieved.

Finally, some criteria were almost universallynot achieved at the project sites (blank box, Table4), and yet the associated objectives were usuallystill met. This implies that these factors (such asobstruction angle for pool formation) are not asimportant as other design criteria (such as, in thisexample, obstruction width).

conducted and, although invertebrates rapidly re-colonize the benthos following disturbance, thisprocess may take longer if sources of colonizersare more distant (Gore, 1885) or if the channel isstill re-equilibrating (Booth et al., 1997). Addi-tional sampling in 1999, however, still showed noimprovement in biological condition. There wasalso no detectable improvement in B-IBI scoreswhen sampling sites were located within, ratherthan downstream of, project boundaries.

Local physical channel characteristics, particu-larly LWD frequency but also pool spacing andstored sediment upstream of the benthic inverte-brate sampling sites, generally were not correlatedwith B-IBI. There was only a very weak positiverelationship between B-IBI and extent of bankerosion, median grain size, or bed stability (indi-cated by a ratio of critical shear stress toboundary shear stress: see Olsen, 1997). In con-trast, B-IBI was much better correlated with thelevel of urban development in the local riparianzone (Fig. 5), and with overall watershed urban-ization. In total, these scores indicate that, al-though projects several hundred of meters longmay improve some measures of physical habitat(i.e. pool spacing) in a stream reach over theevaluated time scales of 2–10 years, they have hadlittle influence on the benthic invertebratecommunity.


4. Discussion

The presented results have allowed us to ad-dress the original four questions of this study.

4.1. Does in-stream placement of LWD producephysical channel characteristics typical of streamsin less disturbed watersheds?

In general, LWD is expected to have the great-est influence and range of functions on moderate-slope (0.01–0.03) alluvial channels classifiedmorphologically as pool-riffle or plane-bed, whichincludes the streams studied in this study. Particu-larly in PSL streams, which tend to lack boulderor bedrock obstructions, and in urban streams,lacking deep-rooted woody bank vegetation,LWD is the primary pool-forming mechanism.Indeed, adding LWD to the urban streams in thisstudy produced more physical channel character-istics typical of undisturbed streams, such as poolsand sediment storage sites formed by LWD. Poolhabitat and channel complexity also increased inmost of the project reaches. However, any in-crease in sediment storage and grade control inthese moderate-slope alluvial channels was lessassured. The steepest project reaches studied inthe present paper did not store more sediment,although LWD did provide more grade control inthe steepest reaches. Stabilizing or retaining sedi-ment to reduce downstream sedimentation andassociated flooding was not accomplished byadding LWD to the channel.

4.2. Does biological condition impro�eimmediately downstream of LWD addition?

Since fish habitat creation is usually named asthe primary or secondary goal of an in-streamproject utilizing LWD, project monitoring gener-ally focuses on the physical expressions of thisgoal. In urban streams, however, structural habi-tat may be only one of numerous conditions thatare lacking for fish survival, as well as survival ofother aquatic species (such as benthic inverte-brates) that are critical links in the aquatic foodweb. Since the ultimate goal of habitat restorationis biological recovery, direct measures of the biotaare needed to determine whether structural mea-sures alone are likely to be sufficient in the face ofwatershed-wide disturbances in biogeophysicalprocesses. For the four projects of this study soevaluated, B-IBI analysis detected no positive ef-

Table 3Design criteria based on data reported for stable LWD innatural channels

Description Inferred LWD design criteria

Value Reference

Obstruction widthLisle (1986)Obstructing flow or �30%

Wbka

Obstructing flow �10% Cherry and Bilby(1989)

Log spacing�10 logSpacing in any Gippel et al.

direction diameter (1996)Downstream distance �3Wbk Lisle (1986)

Key piece sizes andfrequency

Bisson et al.0.4 m3Wbk=3–5 m(1987)

Wbk=6–10 m 0.8 m3 Bisson et al.(1987)Bisson et al.Minimum number of 0.13(1987)LWD/mkey pieces per

meter1 m3 WFPB (1997)Wbk=0–5 m2.4 m3Wbk=6–10 m WFPB (1997)

WFPB (1997)Minimum number of 0.3LWD/Wbkkey pieces per Wbk

Log loading0.01Volume per channel Lisle (1986)

area m3/m2

Montgomery et0.035Pieces per channelnumber/m2area al. (1995)

Cross-bed position90°Angled to flow; on Gippel et al.

bed (1996)

Burial and angleAngled to flow; Bisson et al.�30°

mostly buried (1987)

Obstruction angle90°Angled to flow; Cherry and Bilby

partially elevated (1989)

a Wbk, Bankfull width.


Table 4Objectives and LWD design criteria evaluated at project sites

fect on biological condition from the restorationactivities at the time scales sampled. The physicalcharacteristics in the reach that did change dis-played no clear relationship to biologicalcondition.

4.3. Does watershed disturbance exert greatercontrol o�er physical and biological reco�ery ofthe channel than the local in-channel conditionsaddressed by added LWD?

The influence of watershed disturbance onphysical channel response was evident and, inseveral instances, overwhelmed any potentialbenefits of LWD. High sediment loads buriedsome LWD, and high flows transported seeminglyappropriately sized LWD out of the channel orincised underneath LWD formerly within thebankfull channel. Biological integrity was not af-

fected by LWD additions, but it was stronglyinfluenced by overall watershed disturbance. B-IBI here varied inversely with the increase inpercent developed land, in full accord with otherregional studies (Karr and Chu, 2000; Morley,2000).

The degree to which watershed conditions influ-ence project effectiveness varies with specific pro-ject objectives. Where the long-term goal of thesehabitat enhancement projects is to promote bio-logical recovery, the extent of overall watersheddevelopment strongly limits the success of localLWD placement projects, independent of projectdesign or execution. If the creation of more com-plex bedforms and stable banks is the main objec-tive, watershed degradation can also limit projectsuccess. For example, several LWD structuresintended to scour pools in the most urban streamsin the study instead were filled with sediment, and


severe bank erosion continued even where vegeta-tion was planted for protection. Where projectobjectives included enhancement of riparian vege-tation, however, watershed conditions were lessimportant than local vegetation management.

At the projects in the least degraded watersheds(52–58% development), there was little evidencethat the urban-modified flow pattern were under-mining in-stream efforts to enhance the channel.The undeveloped riparian corridor in the leastdeveloped basins also aided the channel en-hancement effort. Trees were naturally recruitedto the stream from these buffers (although com-monly not meeting minimum size criteria). Buriednatural LWD and existing vegetation con-tributed significantly to pool formation and bankresistance. The deepest pools in several reacheswere formed by formerly buried natural LWD,and by trees and remnant stumps on the banks.These features also emphasize the long-lastinginfluence of LWD in the channel and on thefloodplain.

4.4. How can LWD project designs be impro�ed?

In the relatively short time that most of theprojects evaluated in this study have been inplace, none experienced complete structural fail-ure. Yet no project met all articulated objectives,most importantly because those objectives weretoo ambitious to be achieve simply by addingLWD. Where design problems hindered evenmodest improvements, the most common short-comings were ineffective log spacing, failure ofadded LWD to remain within the bankfull chan-nel, or placement of LWD too small to affect theflow. Studies documenting the size distributionand frequency of LWD found in natural streamsshould be used to inform designs in projectswhere LWD is added. Particularly where suffi-ciently large LWD is absent or sparse, anchoringshould be considered to insure that LWD influ-ences the flow in desired locations. When andwhere LWD is added should also be guided by anunderstanding of local sediment supply and trans-port conditions, tolerance for LWD movement,and articulated priorities for LWD function (e.g.grade control, high flow refuge, direct cover, etc.).

In our experience, pre-existing project monitor-ing was often inadequate for evaluating success inmeeting project objectives, as other researchershave found (for example, Kondolf and Micheli,1995). Only one of the projects studied here(Swamp Creek) specified any criteria for evaluat-ing project performance (KCDNR, 1997). Fur-thermore, the project objectives named at most ofthe sites, particularly those related to habitatquality and complexity or to aquatic species re-covery, could not be adequately evaluated usingstandard monitoring routines. Pool and LWDcounts, for example, are not sufficient to evaluatebiological or habitat conditions, as we haveshown in the present study (see also Poole et al.,1997). Biologically, placing logs devoid of bark,roots, branches, or leaves into urban streams isnot equivalent to natural recruitment, where in-stream obstructions and cover are but one benefitof a forested riparian corridor, and the organicinput comes in a variety of forms, sizes, andconfigurations (Bilby and Ward, 1989; Gregory etal., 1991).

4.5. Management implications

Adding LWD to a degraded urban stream willnot make it perform the same functions as in anundisturbed stream system. Whereas addingLWD is likely to achieve a number of specificobjectives, a proliferation of in-stream LWDprojects will not address the multitude of urbanstressors nor ensure habitat protection or biologi-cal recovery. Therefore, it is critical to identify theprimary factors causing degradation in a reach, toevaluate existing channel conditions, and to deter-mine which, if any, objectives could be met within-stream enhancement projects.

Added LWD can improve physical conditionsin a channel, but only if the LWD affects condi-tions that were significant causes of initial degra-dation. These might include channel straighteningand clearing that resulted in increased flow veloc-ities or simplified habitat, or historic removal ofLWD from the channel that resulted in loss ofbank stability or cover. When the source of aproblem originates from upstream or from water-shed-scale disturbances, however, adding LWD


does not appear either to solve the problem or tomitigate its consequences.

Not all placement techniques and types ofLWD are equally suited to meet specific designobjectives. For example, the potentially higherconstruction costs of anchored LWD may bejustified along a constrained channel where bankprotection is an objective, whereas larger unan-chored LWD might be more desirable where ac-cess is limited and channel complexity is desired.The time frame over which improvements aredesired must also be determined. UnanchoredLWD that ends up in the floodplain might beuseful additions to the riparian system in thelong-term, but it will not contribute to channelenhancement objectives in the short term.

Although we found no obvious detriments toadding LWD to urban streams, resources shouldnot be allocated to stream projects in the name ofphysical or biological recovery while overlookingsource control of problems in the watershed.

5. Conclusions

This work evaluates the effectiveness of in-stream projects using LWD in urban streamswhere no systematic effort had been made toreduce degradation at the watershed scale. Thesetypes of projects are increasingly popular, particu-larly in the Pacific Northwest, where LWD isrecognized as an important element in physicalhabitat for salmonids. Yet there is little evidencethat these in-stream projects can reverse even thelocal expressions of watershed degradation in ur-ban channels. Adding LWD to urban streams canbe successful in increasing the number of pools ina reach, but it cannot be expected to increasesediment retention or to stabilize in-channel sedi-ment over the time frames observed in this study.Biological conditions, as assessed by benthicmacroinvertebrates, did not improve as a result ofthe in-stream rehabilitation projects; instead, theydirectly relate to the level of development in theupstream watershed. Although better placement,size and frequency of added LWD can increasethe chances of meeting some project objectives,the most important consideration for these

projects is the establishment of feasible goals inthe context of existing and anticipated watersheddisturbance.

Acknowledgements

This project was funded by the Center forUrban Water Resources Management (CUWRM)and the Center for Streamside Studies at theUniversity of Washington, and by the UnitedState Environmental Protection Agency throughgrant award R-82528-01-0 from the Waters andWatersheds Program. The authors thank ChrisKonrad, Jim Karr, Stephen Burges and SallySchuman for their work on the Urban StreamRehabilitation Project. Ecologists and engineersat the King County Department of Natural Re-sources in Seattle, WA provided critical informa-tion on the projects studied, particularly JohnBethel, Kerry Bauman, Kathryn Neal, and JeffBurkey. The authors also thank Steve Kropp,Suzanne Stephens, and Ron Larson for their fieldassistance.

References

Abbe, T., Montgomery, D.R., Petroff, C., 1997. Design ofstable in-channel wood debris structures for bank protec-tion and habitat restoration: an example from the CowlitzRiver, WA. In: Wang, S.S., Langendoen, E.J., Shields,F.D. (Eds.), Proceedings of the Conference on Manage-ment of Landscapes Disturbed by Channel Incision, TheCenter for Computational Hydroscience and Engineering,University of Mississippi, Oxford, MS, pp. 809–815.

Beechie, T.J., Sibley, T.H., 1997. Relationship between chan-nel characteristics, LWD, and fish habitat in NW Washing-ton streams. Transactions of the American FisheriesSociety 126, 217–229.

Bilby, R.E., 1984. Removal of woody debris may affect streamchannel stability. Journal of Forestry 82, 609–613.

Bilby R.E., Ward, J.W., 1989. Changes in characteristics andfunction of woody debris with increasing stream size inwestern streams. Transactions of the American FisheriesSociety 118 pp. 368–378.

Bisson, P.A., Bilby, R.E., Bryant, M.D., Dolloff, C.A., Grette,G.B., House, R.A., Murphy, M.L., Koski, K.V., Sedell,J.R., 1987. Large woody debris in forested streams of thePacific Northwest past, present and future. In: Salo, E.,Cundy, T.W. (Eds.), Streamside Management: Forestryand Fishery Interactions: Contribution 57. University of


Washington Institute of Forest Resources, Seattle, WA,pp. 143–190.

Booth, D.B., 1990. Stream-channel incision followingdrainage-basin urbanization. Water Resources Bulletin 26(3), 407–417.

Booth, D.B., 1991. Urbanization and the natural drainagesystem — impacts, solutions, and prognoses. The North-west Environmental Journal 7, 93–118.

Booth, D.B., Montgomery, D.R., Bethel, J., 1997. Largewoody debris in urban streams of Pacific Northwest. InProceedings of the Conference on Effects of WatershedDevelopment and Management on Aquatic Ecosystems,American Society of Civil Engineers, pp. 179–197.

Cherry, J., Bilby, R.L., 1989. Coarse woody debris and chan-nel morphology: a flume study. Water Resources Bulletin25 (5), 1031–1036.

Crispin, V., House, R., Roberts, D., 1993. Changes in in-stream habitat, large woody debris, and salmon habitatafter the restructuring of a coastal Oregon stream. NorthAmerican Journal of Fisheries Management 13 (1), 96–102.

Doberstein, C.P., Karr, J.R., Conquest, L.L., 2000. The effectof fixed-count subsampling on macroinvertebrate biomoni-toring in small streams. Freshwater Biology 44 (2), 355–371.

Fore, L.S., Karr, J.R., Wisseman, R.W., 1996. Assessing inver-tebrate responses to human activities: evaluating alterna-tive approaches. Journal of the North AmericanBenthological Society 15 (2), 212–231.

Frissell, C.A., Nawa, R.K., 1992. Incidence and cause ofphysical failure of artificial habitat structure in streams ofWestern Oregon and Washington. North American Jour-nal of Fisheries Management 12, 182–197.

Gippel, C.J., O’Neil, I.C., Finalyson, B.J., Schnatz, I., 1996.Hydraulic guidelines for the re-introduction and manage-ment of large woody debris in lowland rivers. RegulatedRivers: Research and Management 12, 223–236.

Gore, J.A., 1885. Mechanisms of colonization and habitatenhancement for benthic macroinvertebrates in restoredriver channels. In: Gore, J.A. (Ed.), The Restoration ofRivers and Streams: Theories and Experience. Butter-worth, Boston, MA, pp. 81–101.

Gortz, P., 1998. Effects of stream restoration on the macroin-vertebrate community in the River Esrom, Denmark.Aquatic Conservation: Marine and Freshwater Ecosystems8 (1), 115–130.

Graf, W.L., 1975. The impact of suburbanization on fluvialgeomorphology. Water Resources Research 11, 690–692.

Greenberg, E.S., 1995. The influence of large woody debris onbenthic invertebrate communities in two King County,WA streams. M.Sc. Thesis. University of Washington,Seattle, WA.

Gregory, S.V., Swanson, F.J., McKee, W.A., Cummins, K.W.,1991. An ecosystem perspective of riparian zones: focus onlinks between land and water. BioScience 41, 540–551.

Harr, R.D., 1976. Hydrology of small forest streams in west-ern Oregon. US Forest Service General Technical Report

PNW-55. Pacific Northwest Forest and Range ExperimentStation, Portland, OR.

Hilderbrand, R.H., Lemly, A.D., Dolloff, C.A., Harpster,K.L., 1997. Effects of LWD placement on stream channelsand benthic macroinvertebrates. Canadian Journal of Fish-eries and Aquatic Science 54, 931–939.

Hilderbrand, R.H., Lemly, A.D., Dolloff, C.A., Harpster,K.L., 1998. Design considerations for large woody debrisplacement in stream enhancement projects. North Ameri-can Journal of Fisheries Management 18 (1), 161–167.

Hollis, G.E., 1975. The effects of urbanization on floods ofdifferent recurrence intervals. Water Resources Research11, 431–435.

Horner, R.A., Booth, D.B., Azous, A., May, C.W., 1997.Watershed determinants of ecosystem functioning. In: Pro-ceedings of the Conference of Effects of Watershed Devel-opment and Management on Aquatic Ecosystems,American Society of Civil Engineers, pp. 251–274.

House, R.A., 1996. An evaluation of stream restoration struc-tures in a coastal Oregon stream, 1981–1993. North Amer-ican Journal of Fisheries Management 16 (2), 272–281.

House, R.A., Boehne, P.L., 1986. Effects of instream struc-tures on salmonids habitat and populations in Tobe Creek,Oregon. North American Journal of Fisheries Manage-ment 6, 38–46.

Karr, J.R., Chu, E.W., 1999. Restoring Life in RunningWaters: Better Biological Monitoring. Island Press, Wash-ington, DC, p. 206.

Karr, J.R., Chu, E.W., 2000. Sustaining living rivers. Hydrobi-ologia 422, 1–14.

Karr, J.R., Fausch, K.D., Angermeier, P.L., Yant, P.R.,Schlosser, I.J., 1986. Assessment of biological integrity inrunning waters: a method and its rationale. Illinois NaturalHistory Survey Special Publication 5.

KCDNR, 1995. Capital Improvement Projects (CIP) Monitor-ing Programs Annual Report. Soosette Creek channel sta-bilization, King County Department of Natural Resources.

KCDNR, 1996. Capital Improvement Projects (CIP) Monitor-ing Programs Annual Report. Lower Laughing Jacob’sCreek sediment management and habitat improvement,year one, King County Department of Natural Resources.

KCDNR, 1997. Capital Improvement Projects (CIP) Monitor-ing Programs Annual Report. Swamp Creek debris jamremoval and channel restoration project, year one, KingCounty Department of Natural Resources.

Kerans, B.L., Karr, J.R., 1994. A benthic index of biologicalintegrity (B-IBI) for rivers of the Tennessee Valley. Ecolog-ical Applications 4, 768–785.

Kerschner, J.L., 1997. Monitoring and adaptive management.In: Williams, J.E., Wood, C.A., Dombeck, M.P. (Eds.),Watershed Restoration: Principles and Practices. AmericanFisheries Society, Bethesda, MD, pp. 116–131.

Kondolf, G.M., Micheli, E.R., 1995. Evaluating streamrestoration projects. Environmental Management 19 (1),1–15.

Kropp, S., 1998. Puget Sound stream rehabilitation projectinformation system. Center for Urban Water ResourcesManagement, University of Washington, Seattle, WA.


Leopold, L.B., 1968. Hydrology for urban land planning — aguidebook on the hydrologic effects of urban land use. USGeological Survey Circular 554, 1–18.

Lisle, T.E., 1986. Stabilization of a gravel channel by largestreamside obstructions and bedrock bends, Jacobs Creek,northwestern California. Geological Society of AmericaBulletin 97, 999–1011.

Lisle, T.E., 1987. Using ‘residual depths’ to monitor pooldepths independently of discharge. USDA Forest ServicePacific Southwest Forest and Range Experiment Station,Research Note PSW-394, December.

MacDonald, M., 1995. A combination on behalf of restora-tion: the coalition to restore urban waters. Restoration andManagement Notes 13 (1), 98–103.

May, C.W., 1996. Assessment of cumulative effects of urban-ization on small streams in the Puget Sound Lowlandecoregion: implications for salmonid resource manage-ment. Ph.D. Thesis. University of Washington, Seattle,WA.

Montgomery, D.R., Buffington, J.M., 1998. Channel pro-cesses, classification, and response potential. In: Naiman,R.J., Bilby, R.E. (Eds.), River Ecology and Management.Springer-Verlag, New York, pp. 13–42.

Montgomery, D.R., Buffington, J.M., Smith, R.D., Schmidt,K.S., Pess, G., 1995. Pool spacing in forest channels.Water Resources Research 13 (4), 1097–1105.

Morley, S.A., 2000. Effects of urbanization on the biologicalintegrity of Puget Sound lowland streams: restoration witha biological focus. M.Sc. Thesis. University of Washing-ton, Seattle, WA.

Nelson, K., 1998. The influence of sediment supply and largewoody debris on pool characteristics and habitat diversity.M.Sc. Thesis. University of Washington. Seattle, WA.

Olsen, D.S., Whitacker, A.C., Potts, O.F., 1997. Assessingstream stability thresholds using flow competence estimatesat bankfull stage. Journal of American Water ResourcesAssociation. 33(6), 1197–1207.

Parametrix, 1988. Unpublished Forbes Creek promotionalsheet. Parametrix, Inc. Kirkland, WA.

Poole, G.C., Frissell, C.A., Ralph, S.C., 1997. Instream habi-tat unit classification: inadequacies for monitoring andsome consequences for management. Journal of the Amer-ican Water Research Association 33 (4), 879–896.

Riley, A.L., 1998. Restoring Streams in Cities: A Guide forPlanners, Policy Makers, and Citizens. Island Press, Wash-ington, DC, p. 423.

Shields, F.D., Knight, S.S., Cooper, C.M., Smith, R.H., 1998.Rehabilitation of aquatic habitats in warmwater streamsdamaged by channel incision in Mississippi. Hydrobiologia382, 63–86.

WFPB, 1997. Board Manual: Standard Methodology for Con-ducting Watershed Analysis. Version 4.0, November,Washington Forest Practices Board.

effectiveness of large woody debris in stream...

Documents