wood and sediment storage and dynamics in river corridors · wood and sediment storage and dynamics...

EARTH SURFACE PROCESSES AND LANDFORMSEarth Surf. Process. Landforms (2016)Copyright © 2016 John Wiley & Sons, Ltd.Published online in Wiley Online Library(wileyonlinelibrary.com) DOI: 10.1002/esp.3909

State of Science

Wood and sediment storage and dynamics in rivercorridorsEllen Wohl* and Daniel N. ScottDepartment of Geosciences, Colorado State University, Fort Collins, CO, USA

Received 3 March 2015; Revised 19 January 2016; Accepted 20 January 2016

*Correspondence to:Ellen Wohl, Department of Geosciences, Colorado State University, Fort Collins, CO 80523-1482, USA. E-mail: [email protected]

ABSTRACT: Large wood along rivers influences entrainment, transport, and storage of mineral sediment and particulate organicmatter. We review how wood alters sediment dynamics and explore patterns among volumes of in-stream wood, sediment storage,and residual pools for dispersed pieces of wood, logjams, and beaver dams. We hypothesized that: volume of sediment per unit areaof channel stored in association with wood is inversely proportional to drainage area; the form of sediment storage changesdownstream; sediment storage correlates with wood load; the residual volume of pools created in association with wood correlatesinversely with drainage area; and volume of sediment stored behind beaver dams correlates with pond area. Lack of data from largerdrainage areas limits tests of these hypotheses, but the analyses suggest that sediment volume correlates positively with drainage areaand wood volume. The form of sediment storage in relation to wood appears to change downstream, with wedges of sedimentupstream from jammed steps most prevalent in small, steep channels andmore dispersed sediment storage in lower gradient channels.Pool volume correlates positively with wood volume and negatively with channel gradient. Sediment volume correlates well withbeaver pond area.More abundant in-stream wood and beaver populations present historically equated to greater sediment storage within river

corridors and greater residual pool volume. One implication of these changes is that protecting and re-introducing wood and beaverscan be used to restore rivers. This review of the existing literature on wood and sediment dynamics highlights the lack of studies onlarger rivers. Copyright © 2016 John Wiley & Sons, Ltd.

KEYWORDS: in-stream wood; large wood (LW); rivers; sediment; pools

Introduction

Large wood (LW) along river corridors creates numerousphysical and ecological effects that influence sediment dyna-mics (e.g. Keller and Tally, 1979; Harmon et al., 1986; Kelleret al., 1995; Wohl, 2013). LW refers to downed, dead woodpieces ≥ 1 m long and 10 cm in diameter. The river corridor iscomposed of the active channel and adjacent floodplains.Sediment dynamics includes the entrainment (initiation ofmotion), transport, deposition, and storage of particulate matter,which here includes mineral sediment and particulate organicmatter (POM).Wood along river corridors creates hydraulic resistance that

can decrease flow velocity and transport capacity in the vicinityof the wood (Shields and Smith, 1991; Davidson and Eaton,2013) or, if sufficient wood is dispersed throughout the corridor,along an entire river reach (Brooks et al., 2003, 2006;Webb andErskine, 2003; Wallerstein and Thorne, 2004; Umazano et al.,2014). By deflecting flow, LW can also locally enhance entrain-ment of bed material and scour of the channel bed and banks(Buffington et al., 2002). The magnitude of the effects of LW

on sediment dynamics depends on factors such as wood load(volume of wood per unit area of channel), orientation andstability of LW, and sediment supply relative to transportcapacity (Wallerstein and Thorne, 2004; Yarnell et al., 2006;Magilligan et al., 2008; Fisher et al., 2010; Cadol and Wohl,2011). The cumulative effects on river corridors of persistentLW have been described as a mediated equilibrium conditionin which water and sediment fluxes and river corridor geometryare substantially influenced by riparian vegetation and LW(Brooks and Brierley, 2002).

Numerous case studies now document the effects of LW onsediment dynamics, but the information presented in these casestudies has not been effectively synthesized. The primaryobjectives of this paper are to review what is known of woodand sediment dynamics and to collect previously publishedinformation from diverse rivers to evaluate whether patternsof wood–sediment interactions can be discerned in relation tocharacteristics such as channel geometry, channel gradient,drainage area, or wood load. The presence of consistentpatterns across diverse channels and spatial scales could beused to more quantitatively predict the effects of LW on

E. WOHL AND D. N. SCOTT

sediment storage and to develop management goals forrestoring LW in river corridors and facilitating sediment storageand channel stability. This review focuses on bedload mineralsediment because the majority of published work examinesthe effects of wood on bedload processes. However, becausesuspended mineral sediment and POM exert such importantinfluences on river ecosystems (e.g. Bilby and Likens, 1980;Wallace et al., 1995), the review also includes these forms ofsediment.We hypothesize that the volume of sediment per unit area of

channel stored in association with in-stream LW declines in adownstream direction as drainage area increases (Figure 1A).Drainage area here is a proxy for several other variables thattypically exhibit progressive downstream changes; specifically,increasing discharge and channel width and decreasingchannel gradient and average bed grain size. As channels growprogressively larger and discharge increases downstream,transport capacity for in-stream LW should increase (Gurnell,2003). Increasing transport capacity reflects smaller values ofthe ratios of LW piece length to channel width and piecediameter to flow depth, each of which positively correlates withmobility of in-stream LW (Braudrick et al., 1997; Welber et al.,2013). Larger channels with smaller bed material are also lesslikely to have protruding obstacles, such as large boulders or

Figure 1. Hypothesized relations between sediment storage andpotential control variables. (A) Sediment stored per unit area of channelcould correlate inversely with drainage area. In this figure, we assumethat increasing drainage area correlates positively with floodplain extentand inversely with gradient and average size of channel substrate. Insetovals represent different forms of in-stream wood and associatedsediment storage. (B) Sediment stored per unit area of channel couldcorrelate primarily with in-stream wood load, with either a linear orexponential relationship.

Copyright © 2016 John Wiley & Sons, Ltd.

ramped wood pieces extending across a substantial fractionof total channel width, which can trap wood in transport(Braudrick and Grant, 2001; Bocchiola et al., 2006; Beckmanand Wohl, 2014b). If wood mobility increases downstream,then we expect to see decreasing sediment storage downstreambecause of less stable wood that can retain sediment (Wohl et al.,2009; Cadol and Wohl, 2011). Rivers can be categorized aslarge based on drainage area, channel dimensions, discharge,or other parameters. In the context of LW and sediment dyna-mics, we use the definition of Piégay (2003): a large river has awidth several times greater than the height of the trees in itsriparian area.

We also hypothesize that the manner in which sediment isstored in association with LW changes downstream (Figure 1A).Steep, laterally confined, smaller channels are likely to haverelatively closely spaced, channel-spanning steps (Chin andWohl, 2005; Church and Zimmermann, 2007), at least someof which are created by in situ logjams (Abbe and Montgomery,2003). Each of these steps creates a backwater that stores awedge of sediment upstream. In steep channels, the down-stream thickness of this wedge can be comparable to averagechannel width (Andreoli et al., 2007; Ryan et al., 2014), creat-ing substantial sediment storage per unit area of channel. Thesize of log steps and the resultant thickness of the sedimentwedge behind them also depend on the presence of coarsebed material and the diameter of the wood forming the step(Wohl et al., 1997; Scott et al., 2014). As the channel becomesless steep and laterally confined, as well as wider, fluvial trans-port capacity for in-stream wood increases. LW is likely to bedispersed along a reach or to form transport jams (Marcuset al., 2002; Abbe and Montgomery, 2003; Wohl and Jaeger,2009). Dispersed wood can create sufficient hydraulic resis-tance to facilitate deposition and storage of dispersed sedimentand transport jams can create local upstream and downstreamdeposition, but the volume of sediment stored per unit area ofchannel is likely to be lower than in the steep segments of theriver network because the sediment is finer grained and hencemore likely to continue downstream or to move onto the flood-plain. Large, low-gradient, floodplain rivers in forested regionshistorically had extensive wood rafts (Triska, 1984; Wohl,2014) that could block even large channels sufficiently tocreate enhanced overbank deposition and sediment storageon floodplains. These rafts, although still present in a fewrivers (Webster et al., 2002; Martín-Vide et al., 2014; Boivinet al., 2015), are now very rare. Although there is stratigraphicevidence that wood rafts could cause substantial overbankdeposition (Barrett, 1996; Patterson et al., 2003), this depositionwas spread across broad floodplains. The deposition thus likelyresulted in lower sediment storage per unit area than jammedsteps in small, steep channels, but greater volume of depositionper unit length of channel.

An alternate hypothesis is that the presence of LW is the mostimportant influence on the volume of sediment stored andtherefore sediment storage correlates most strongly with woodload (Figure 1B). This relationship could take the form of alinear increase if more wood per unit area of channel simplycauses greater sediment storage. The relationship could alsoresemble an exponential increase if non-linear interactions thatoccur with increasing wood load, such as the formation oflogjams, also create non-linear responses in sediment storage.The most likely non-linear interaction would be the formationof a channel-spanning logjam or, in large rivers, a wood raftthat facilitates overbank flow and deposition, as well asformation of a multithread channel planform (Triska, 1984;Brummer et al., 2006; Wohl, 2011; Collins et al., 2012).

We hypothesize that the residual volume of pools created inassociation with LW declines in a downstream direction

Earth Surf. Process. Landforms, (2016)

Figure 2. Hypothesized relation between residual pool volume anddrainage area. Inset ovals represent different forms of in-stream woodand associated bed scour to form pools. igure 3. Downstream view of a beaver pond perched along the

alley side wall ~3 m above the active channel of the Duke Riverut of sight here in the right foreground), Yukon Territory, Canada.

WOOD AND SEDIMENT DYNAMICS IN RIVER CORRIDORS

(Figure 2). Pools are a readily measured manifestation of theeffects of LW on sediment entrainment and removal. LW stepscan create backwater pools upstream from the step crest andscoured plunge pools downstream from the step crest (Bilbyand Ward, 1989). These pools can be substantial (Richmond,1994; Zelt, 2002; Buffington et al., 2002), despite the higherosional resistance of coarse-grained streambeds. Dispersedwood can create local scour of the bed and transport jamscan create backwater and local scour, but at lower channelgradients, LW is less likely to form tall jams with extensivebackwaters or plunging flow. Our limited experience withrelatively small wood rafts on channels in central Alaskasuggests that deeper bed scour can be present beneath thesefeatures, but the resulting pool volume is likely to be smallerper unit area of channel than in steep, narrow channels.We test these hypotheses and the underlying assumptions

about interactions between LW, stream flow, and channelboundaries by evaluating trends in LW, sediment storage, andpool volume in relation to drainage area, channel gradient,and bankfull width from published datasets. We focus ondatasets from channels with minimal human alteration ofwood recruitment, in-stream and floodplain LW, and flowregime. Diverse human activities have substantially altered landcover and reduced recruitment of LW to river corridors, aswell as directly removing LW from channels and floodplains(Montgomery et al., 2003a, 2003b; Wohl, 2014). The netresult is that most river corridors in forested environments arewood-poor relative to conditions prior to intensive humanresource use. Comparisons of LW in (i) old-growth, (ii) youngerbut natural, and (iii) managed forests indicate that old-growthforests have the largest wood loads within river corridors, butnaturally disturbed forests are closer to old-growth conditionsthan are managed forests and river corridors that have ex-perienced timber harvest and LW removal from rivers (BeckmanandWohl, 2014b). The datasets used in our analyses come fromrivers within old-growth forests or forests with only naturaldisturbances.We also consider the effects of beaver dams on sediment

storage within river corridors. Beaver dams form a distinctivesubset of in-stream wood because beavers actively create andmaintain dams. Beavers recruit wood to the channel and flood-plain by cutting down living trees and beaver dams commonlyincorporate wood pieces of a size that would otherwise likelybe mobile within the channel. Beavers can also create pondsthat are perched along the valley sides more than a meterabove the active channel and primary floodplain (Figure 3).Because beavers actively shape dams that obstruct flow


Fv(o

and create upstream backwaters, we hypothesize that thesediment volume behind beaver dams will correlate positivelywith pond area.

Existing Knowledge of Wood and SedimentDynamics

Before testing the hypotheses presented earlier, it is useful toreview what is known of how LW in channels and floodplainsinfluences sediment dynamics. Very few studies quantify factorssuch as residence times of sediment stored in association withLW, but at least qualitative understanding of the effects of LWon sediment dynamics can be gleaned from existing literature.The following sections review what is known of how LW influ-ences sediment entrainment and transport within channels, sed-iment storage within channels, floodplain sedimentation, POMdynamics, and effects of wood removal on sediment dynamics.

Influences of wood on in-channel entrainmentthresholds and transport rates

By increasing hydraulic resistance within channels, LW typicallydecreases flow velocity (MacFarlane and Wohl, 2003; Andreoliet al., 2007; Yochum et al., 2012), sediment entrainment, andsuspended and bedload transport at spatial scales > 101 m(Assani and Petit, 1995). However, by creating flow separationand locally accelerated and plunging flow, LW can enhanceentrainment of bed and bank sediment at scales of 100 to 101

m and create persistent scour features such as pools (Lisle,1995; Abbe and Montgomery, 1996; Buffington et al., 2002;Klaar et al., 2011).

The effects of LWon sediment entrainment and transport maybe stage dependent. Working in gravel-bed Mack Creek ofwestern Oregon, USA, Faustini and Jones (2003) found thatstream reaches with abundant LWexhibited less cross-sectionalchange than reaches lacking LW during moderate flows (returnperiod < approximately five years), but responded similarly tolarger peak flows with return periods of approximately 10 to25 years.

When LW that is storing sediment breaks or becomes mobile,a substantial pulse of sediment transport can result. Bugosh andCuster (1989) documented how breakup of a logjam on gravel-bed Squaw Creek in Montana, USA (drainage area 106 km2,



channel width 8–20 m, gradient 0.02–0.03) resulted in bedloadtransport two times greater than that measured at similardischarges prior to the breakup. Similar observations of acce-lerated sediment transport and channel change following LWmovement or jam breakup come from a stream in the ColoradoRocky Mountains (Adenlof and Wohl, 1994).The effects of LW on sediment entrainment and transport

also depend on channel substrate and the distribution of thewood. Flow deflection and acceleration associated with LW,especially LW accumulations that form channel-spanning stepsin steep channels, can be particularly important in locallyenhancing entrainment in gravel- to boulder-bed channelswith relatively high values of critical shear stress. Dispersedwood in lower-gradient, sand-bed channels can reduce near-bed and average velocity and associated sediment transportcapacity and result in increased storage (e.g. Brooks et al.,2003), which presumably reflects substantially reducedentrainment and transport. Most studies that directly documentthe effects of LW on sediment transport, however, focus oncoarser-grained streambeds, as reflected in the literature citedearlier.The effects of beaver dams on in-channel entrainment thresh-

olds and transport rates are similar to those described for otherforms of LW. Beaver dams decrease sediment entrainment andtransport within the backwater zone upstream from the dam(Butler and Malanson, 1995; John and Klein, 2004; De Visscheret al., 2014), enhance bed scour where flow plunges over a dam(Pollock et al., 2014), enhance spatial variation in average bedgrain size (Curran and Cannatelli, 2014), and can createsubstantial pulses of sediment transport when a dam fails (Butlerand Malanson, 2005; Russell et al., 2009).

igure 4. Channel-spanning logjams that create local sedimenttorage. (A) Upstream view of the base of a logjam on an unnamedibutary to the Upper Rio Chagres in Panama. The jam creates a wedgef sediment, approximately 6 m thick at the downstream end, whichpers upstream. Person holding 2-m-tall survey rod at upper left forcale, highlighted with white oval. (B) Downstream view of wedge ofand-sized and finer sediment and particulate organic matter storedpstream from a jam that is partially breached along North St Vrainreek, Colorado, USA. Channel is approximately 15 m wide.

Influences of wood on in-channel sediment storage

The majority of studies that examine how wood affectssediment storage focus on material transported as bedload. Afew studies, however, have explicitly examined sedimenttransported in suspension. Skalak and Pizzuto (2010) docu-ment fine-grained deposits that accumulate downstream fromLWalong the margins of the gravel-bed South River in Virginia,USA. These deposits include a total mass equivalent to 17% to43% of the annual suspended sediment load, with an averageturnover time of 1.75 years for sediment. Along this river, mar-ginal LW creates sufficient flow separation to facilitate deposi-tion and storage of sand, silt, clay, and POM that wouldotherwise remain in transport.Individual LW pieces or jams can create flow obstructions,

resulting in backwater areas in which particulate matter isstored (Keller and Swanson, 1979; Megahan, 1982; Webband Erskine, 2003; Andreoli et al., 2007; Scott et al., 2014)(Figure 4). In addition to promoting sediment storage, LW canfacilitate deposition of finer sediment on the streambed thanis present in segments of channel without wood (Faustini andJones, 2003; Dumke et al., 2010). The orientation and stabilityof pieces and jams is particularly important: pieces orientedparallel to flow are less likely to facilitate storage of particulatematter (Magilligan et al., 2008). LW that is frequently mobilizedmay create only temporary sediment storage (Wohl et al., 2009;Cadol and Wohl, 2011), whereas deposition around stablewood can completely bury the wood and create a streambedwith alternating strata of LWand sediment (Gippel et al., 1992).LW commonly creates high spatial variability in average bed

grain size in relatively steep, coarse-grained channels. Thisspatial variability takes the form of relatively homogeneousbut spatially limited textural patches (Hogan, 1989; Buffingtonand Montgomery, 1999). Working in a relatively small stream


FstrotassuC

(bankfull width 15 m) in British Columbia, Canada, however,Haschenburger and Rice (2004) found that individual largejams could influence bed material texture over lengths of upto 100 times the bankfull width.

Studies of the effects of LWon storage of sediment moving asbedload include several different foci. Many studies of smallerstreams focus on correlations between volumes of LW andstored sediment per unit area of channel (Table I). Studies inthe Pacific Northwest region of the United States have alsoexamined how LW can alter regional patterns of channelsubstrate in relation to drainage area and reach-scale channelgradient. In the steep channel networks of the Oregon CoastRange, USA, the distribution of bedrock and alluvial streamreaches reflects the influences of drainage area, channel gradi-ent, sediment trapping by LW jams, and boulders deposited bydebris flows (Montgomery et al., 2003a, 2003b). Bedrock andfree-formed alluvial reaches can be distinguished based onslope–area relations (Montgomery et al., 1996), but large,persistent LW jams can convert bedrock reaches to alluvialreaches by storing sediment (Massong and Montgomery, 2000).

Studies of LW storage in small, steep channels also empha-size the influence of disturbances such as debris flows. Debris


Table I. Published values for volumes of in-channel wood and stored sediment.

SiteDrainagearea (km2)

Gradient(m/m)

Bankfullwidth (m)

Reachlength (m)

Riparianstand agea

Woodvol(m3/ha)

Sedimentvol(m3/ha) Reference

Tres Arroyos,Chile

9.1 0.7 7.8 1500 old-growth 656 to 710 1600 Andreoli et al.,2007

Tonghi Creek,Australia

187 0.002 14.5 715 No human disturbance;wildfires during1960s-70s

576 5464 Webb and Erskine,2003

Bruces Creek,Australia

68 NA ~ 10 310 not reported 751 < 10 645b Webb andDragovich, 2004

Western 5.5 0.11 18.1 500 old-growth 570 408 Nakamura andSwanson,1993

Oregon 10.7 0.07 15.6 500 340 3593

Malaysia 0.33 0.027 2.9 500 no logging since1960s

50 to 60 20 to 40(varies over time)

Gomi et al.,2006

Colorado,USA

2.70 0.09 2 100 old-growth 225 117 Ryan et al.,20140.4 0.09 1 100 289 43

0.81 0.16 1.9 80 353 3351.24 0.06 1.8 140 319 110

Colorado,USA

15.7 0.035 3.6 37 old-growth 335 47 Jackson andWohl,201537.3 0.077 6.1 57 566 443

37.3 0.039 6.5 73 217 5829.0 0.020 6.6 35 226 2725.9 0.031 5.0 65 435 025.8 0.018 5.0 50 144 16218.2 0.025 5.2 45 51 1926.7 0.046 6.0 46 210 027.2 0.036 6.9 70 388 2927.5 0.049 4.4 50 470 7

Wyoming,USA

49.5 0.022 8.9 2000 not reportedfire in 1988

252 58 Zelt and Wohl,200466.8 0.016 11.2 166 53.6

DucktrapRiver, Maine

94 0.0045 10.2 5700 not reportedold-growth

22.6 1.86 Fisher et al.,2010

Queen CharlotteIslands,BC, Canada

6.9 0.0125 20.2 244 570 287 Hogan, 19866.8 0.0199 21.1 250 390 6643.9 0.0207 18.8 230 430 345

10.5 0.0092 37.8 270 280 2665.4 0.0191 27.7 223 380 4163.9 0.0490 21 461 530 1632

20.2 0.0125 29.3 261 520 46645.2 0.0065 45.4 300 390 732

Buena Esperanza,Argentina

12.9 0.065 6.3 1850 old-growth 121 1522 Comiti et al.,2008

Mao et al.,2008

Note: NA indicates not available; sediment stored primarily as jammed steps and sediment wedges indicated with italicized typeface.aThis refers to age of riparian forest along the study reach, where reported. Sites for which riparian stand age was not reported were in some type offorest reserve with minimal human alteration during recent decades.bThis is an upper bound; it is not clear what percentage of this sediment storage is caused by wood.


flows can episodically remove stored LW and sediment fromsteep channels in high-relief drainages (Keller and Swanson,1979;Marston et al., 1997; Hart, 2002;May, 2007), but substan-tial volumes of LWand sediment can accumulate again within afew years in old-growth forests and regions of high sedimentyield (e.g. Bunn and Montgomery, 2003; May and Gresswell,2003b). In steep, coarse-grained, narrow channels, debris flowsmay be the primary transport mechanism for wood and sedi-ment (May and Gresswell, 2003a), creating very large depositswhere the debris flow loses momentum (Hogan et al., 1998).Localized scour of the stream bed and banks can also result

from accelerated flow around a wood-formed obstruction orplunging flow over the obstruction (Gurnell and Sweet, 1998;Buffington et al., 2002; Wallerstein, 2003; Webb and Erskine,2003; Hassan and Woodsmith, 2004). Bilby and Ward (1989)assessed trends in LW and pool volume across 22 segments ofstreams (drainage areas 0.8 to 68 km2) draining old-growthforest in western Washington, USA. They distinguished scour


(erosion of bed or bank caused by deflection and concentrationof flow around LW), plunge, dammed (impounded waterupstream from LW) and backwater (lower velocity downstreamfrom LW) pools associated with LW. Plunge pools were mostcommon in small streams, whereas scour pools were mostabundant in the moderate to large streams in their survey. Bilbyand Ward (1989) developed regressions between pool surfacearea, wood volume, and channel width. Although these rela-tionships were statistically significant, they explained relativelylittle variation in pool surface area (R2 values of 0.39 to 0.59).Wallerstein (2003) developed a one-dimensional (1D) modelfor predicting constriction scour associated with LW. The orien-tation of LW pieces influences scour, with pieces perpendicularto flow creating greatest scour (Cherry and Beschta, 1989).Published values of LW volumes and residual pool volumeare summarized in Table II.

Studies of larger rivers emphasize processes, rather thanvolumes, of sediment storage associated with the presence of


Table II. Published values for in-channel wood load and residual pool volume

SiteDrainagearea (km2)

Gradient(m/m)

Bankfullwidth (m)

Reachlength (m)

Riparianstand age

Woodvol(m3/ha)

Residual poolvolume (m3/ha) Reference

Wyoming, USA 56.9 0.029 12.3 209 not reported 68 24 Zelt, 200253.8 0.016 12.9 228 52 230 Zelt and Wohl,

200448.5 0.019 13.0 324 152 7639.8 0.018 11.4 126 120 51130.6 0.01 10.1 125 76 223.2 0.024 11.1 137 85 13064.5 0.029 15.4 324 302 4432.7 0.027 9.9 109 229 32212.9 0.026 8.4 177 39 6913.8 0.028 7.3 131 47 4737.6 0.018 12.0 139 184 35449.8 0.044 7.6 143 142 11331.3 0.016 7.8 238 155 9320.9 0.008 9.1 140 139 13619.2 0.028 7.8 106 252 17617.0 0.033 7.7 221 239 9428.0 0.010 10.0 187 206 20023.1 0.002 8.6 126 132 19432.5 0.001 7.7 168 133 3739.8 0.026 9.5 182 176 31

NortheasternOregon, USA

25 0.02 4.1 228 undisturbed (specificage not reported)

227 538 Carlson et al., 1990a

19 0.044 4.5 324 304 3699 0.065 3.8 126 258 1948 0.071 3.8 300 232 2297 0.074 3 300 550 233

12 0.045 4.1 300 337 23116 0.021 4 300 478 75525 0.038 6.1 300 474 5417 0.056 4 300 638 475

19 0.035 4.5 300 258 38512 0.054 2.3 300 413 284

Colorado, USA 3.4 0.027 3.7 268 old-growth 251 132 Richmond, 19942.4 0.064 3.8 285 545 2103.9 0.034 3.9 577 169 1995.1 0.006 4.1 221 312 1589.9 0.030 4.1 263 349 2238.9 0.004 4.2 260 217 529

11 0.020 4.5 277 280 19821.7 0.027 5.6 344 211 20317.4 0.033 7 314 378 26824.5 0.034 8 260 131 8629.1 0.022 10.2 300 266 214

Colorado, USA 15.7 0.035 3.6 37 old-growth 335 86 Jackson and Wohl,201537.3 0.077 6.1 57 566 1387

37.3 0.039 6.5 73 217 16929.0 0.020 6.6 35 226 75825.9 0.031 5.0 65 435 025.8 0.018 5.0 50 144 50018.2 0.025 5.2 45 51 14426.7 0.046 6.0 46 210 027.2 0.036 6.9 70 388 10027.5 0.049 4.4 50 470 64

Western Oregon,USA

6.5 0.022 7.8 1550 logged 50 yearsearlier

232 228 Andrus et al., 1988<6.5 0.049 6.1 1950 251 126<6.5 0.057 5.9 750 393 54<6.5 0.060 5.5 750 596 53

Note: Italicized typeface indicates predominantly plunge pools.aType of pool not indicated; assumed any gradient > 0.06 was predominantly plunge pools.


LW. Wood deposition can initiate the formation of bars orislands that then become stabilized by riparian vegetation(O’Connor et al., 2003; Collins et al., 2012) (Figure 5). Workby Mikus et al. (2013) on a gravel-bed river in the PolishCarpathians provides an example of how LW can enhancebed-material storage in larger rivers. Deposition of LW ongravel bars creates a sheltered spot in which woody riparianplants preferentially germinate and mature. As the plants grow


larger, they create greater hydraulic resistance and increasethe erosional resistance of the sediment in which they aregrowing, and this in turn enhances sediment deposition andstorage (Gurnell et al., 2012; Bertoldi et al., 2013). The volumesof sediment historically stored in association with LW werelikely to have been substantial on at least some larger, lowergradient rivers, although records are mostly unavailable.Gippel et al. (1992) cite historical descriptions of removing


Figure 5. Google Earth imagery of a portion of the Snake River,Wyoming, USA, showing a concentration of fluvially deposited woodat the upstream end of a bar. Flow direction is right to left and the baris approximately 300 m long.


LW from the Latrobe River in Australia. Streambed scour afterthe first round of LW removal exposed underlying layers ofwood. Up to three layers of logs had to be removed beforethe channel was cleared, resulting in as much as 2 m of bedincision.A few studies have estimated residence time of bedload

associated with LW. Fisher et al. (2010) used cosmogenicberyllium-7 (7Be) to estimate storage time of bedload associatedwith LWalong the mixed sand-gravel-bed of the Ducktrap Riverin Maine, USA. Transport-limited river segments had longerbedload storage times (~100–200 days) than supply-limited seg-ments (< 100 days). Relative to sediment stored around largeboulders, LW created more storage sites with larger sedimentvolumes and longer storage times (Fisher et al., 2010). Thepresence of abundant LW has the potential to create transport-limited conditions by substantially increasing hydraulic resis-tance, as shown on the Cann River in Australia, where removalof LW altered the channel from a sediment sink to a sedimentsource (Brooks et al., 2003).Another approach to LW-sediment storage has been to

develop empirical or stochastic models of the influence of LWon sediment retention. Working on five river segments inwestern Oregon, Nakamura and Swanson (1993) developedthe following relationship for sediment trapped by LW:

Vs ¼ a Wð Þb Sð Þc (1)

in which Vs is sediment volume per unit channel length (in m3/100 m), W is average channel width (in meters), S is averagechannel gradient (per cent), and a, b, and c are constants thatcan vary over several orders of magnitude. Lancaster et al.(2003) modeled debris flow runout lengths for scenarios inwhich debris-flow velocity was reduced by wood entrainmentor by wood-induced changes in flow direction. Modelingdebris flows at the network scale indicates that, in scenarioswith wood, debris-flow runout lengths are shorter anddebris-flow deposits are widely distributed throughout the net-work, whereas the absence of wood increases runout lengthsand basin sediment yield. Eaton et al. (2012) developed a sto-chastic model of wood and sediment dynamics in gravel-bedstreams. They used an equation for sediment trapping basedon laboratory experiments (Davidson, 2011; Davidson andEaton, 2013):

ΔVsed ¼ Qbmζ bmBi

∑B

� �e-tx (2)

in which ΔVsed is the change in volume of sediment storedby an individual LW piece over a year, Qbm is the annual


bed-material transport rate entering a stream reach, ζ bm is areach-average bed material trapping efficiency, Bi is the areaof the ith LW piece that is projected across the channel,∑B is the sum of the projected areas for all pieces, and txis the length of time that the LW piece has been in itscurrent location.

As noted for sediment entrainment and transport and illus-trated in Figure 1A, the effects of LW on sediment storage varywith substrate and LW characteristics. Studies quantifying thevolume of sediment storage come primarily from relativelysteep, coarse-grained rivers, although a few studies of this typehave focused on larger, sand-bed channels in Australia. Manyof the studies of larger gravel-bed rivers focus on the nature ofLW–sediment interactions, however, rather than quantifyingwood-associated sediment storage.

The effects of LW on sediment storage can also exhibit sub-stantial variation through time as a result of three basic mecha-nisms: (1) Fluctuations in wood recruitment to a channelthrough time. As summarized by Benda and Sias (2003), woodrecruitment to a river segment results from downstream LWfluxes and from lateral inputs in the form of individual tree mor-tality, mass mortality (e.g. wildfire, windstorm, insect infesta-tion), hillslope instability (e.g. debris flows), bank erosion, andexhumation of buried logs from the floodplain. Each of theserecruitment sources can vary over timescales of 10�1 to 102

years. (2) Fluctuations in wood transport capacity within a riversegment through time as a result of processes such as floods ordebris flows. (3) Fluctuations in sediment supply to the channelas a result of downstream sediment fluxes and lateral inputsfrom the banks, floodplain, and adjacent uplands. Each of thestudies cited in relation to LW and sediment storage essentiallyrepresents a ‘snapshot’ of LW–sediment interactions over arelatively short-time interval. Although several models predictfluctuations in wood load through time in relation to factorssuch as lateral LW inputs (e.g. Bragg, 2000) or stochastic inter-actions among LW recruitment and discharge (e.g. Eaton et al.,2012), few field studies quantify changes in sediment storagethrough time (exceptions include Brooks et al. [2003] andMay and Gresswell [2003b]).

Numerous studies document increased sedimentationassociated with beaver dams via both ponding of water up-stream from the dam and sediment deposition within the pond(Naiman et al., 1986; Butler and Malanson, 1995; John andKlein, 2004; Pollock et al., 2007; Green and Westbrook,2009; Butler, 2012; Levine and Meyer, 2014) and enhancedoverbank flows and floodplain sedimentation (Westbrooket al., 2011). Scheffer (1938), for example, estimated 7.8 m3

of sediment stored per linear meter of stream along MissionCreek in Washington, USA. Individual beaver dams can retainup to 6500 m3 of sediment along streams in boreal forest nearQuebec, Canada (Naiman et al., 1986). Pollock et al. (2007)estimated 7200 m3 of sediment in 13 beaver dams along an un-specified length of Bridge Creek in Oregon, USA. Very fewstudies, however, actually quantify the resulting sedimentstorage per unit area of the pond or valley bottom (Table III).Only three of the five sets of published data for pond areaand sediment volume use direct field measurements ofsediment volume (Butler and Malanson, 1995; Meentemeyerand Butler, 1999; John and Klein, 2004). The other two studiesuse a regression equation between pond area and sedimentvolume developed by Butler and Malanson (1995). The lackof direct measurements of sediment volume in relation topond area or unit area of valley bottom is a striking gap inthe literature.

Pollock et al. (2003) proposed a relation between maximumpotential sediment storage behind beaver dams, channelgradient, and dam geometry:


Table III. Published volumes of sediment storage in beaver ponds.

SiteDrainage area

(km2)Gradient

(m/m)Bankfullwidth (m)

Pond area(m2)

Sedimentvolume (m3) Reference

Lake deltas in GlacierNational Park,Montana, USA

NA NA NA 8512 5236 Butler, 2012a

9184 56563036 181310302 63551652 9485207 32543784 23814676 28391536 8761775 1025

Sandown Creek, BritishColumbia, Canada

72 < 0.02 4.9 99 478 Green and Westbrook,2009a405 169

388 158502 230952 5111482 842291 981109 609

Jossa River, Germany NA 0.006 4-8 2274 516 John and Klein, 2004744 242592 1491495 173270 33

Glacier National Park,Montana, USA

NA NA NA 1500 1290 Butler and Malanson,19951000 345

140 31120 2650 118200 5084450 2921710 479

Glacier National Park,Montana, USA

NA NA NA 94 26 Meentemeyer and Butler,1999588 265

181 57690 267180 7780 2238 942 165

Note: NA indicates information not available.aEstimated sediment volume using a regression for pond area and sediment volume developed by Butler and Malanson (1995).


Vm ¼ 0:5H2W=S (3)

in which Vm is maximum sediment storage volume, H is damheight, W is dam width, and S is channel gradient. Because ofthe lack of appropriate parameters for several of the datasetsin Table III, we did not test this relation for these data.As with other forms of LW, the volume of sediment stored

behind a beaver dam varies with time as the valley segmentupstream from the dam gradually aggrades. The rate of fillingdepends on factors such as sediment load in the river and thegeometry and location of the beaver pond (some ponds areperched along a valley wall well above the main channel, forexample). Once a dam is no longer maintained by beaver,breaching of the dam can result in at least partial erosionof the pond sediment, although establishment of rootedaquatic macrophytes and, eventually, other wetland vegetationcommonly limits evacuation of pond sediment. Where abeaver meadow (Morgan, 1868; Polvi and Wohl, 2013) – amore extensive floodplain wetland composed of numerousactive and inactive beaver dams and ponds in various stagesof infilling – is present, the cumulative sediment storage cangreatly exceed average annual downstream sediment flux and


temporal variations in sediment storage can be dampened overtime periods of 101 to 103 years (Pollock et al., 2003; Krameret al., 2012).

Influences of wood on floodplain sedimentation

Concentrations of wood in jams or rafts can substantially re-duce channel conveyance and increase overbank flow duringhigher discharges. This can result in either deposition of partic-ulate matter on the floodplain (Jeffries et al., 2003; Sear et al.,2010) and local aggradation of the channel above the flood-plain (Kochel et al., 1987; Montgomery and Abbe, 2006), oraccelerated bank erosion and channel avulsion (Hickin,1984; Brummer et al., 2006; Phillips, 2012; Boivin et al.,2015), formation of secondary channels (O’Connor et al.,2003; Wohl, 2011; Collins et al., 2012) (Figure 6), and local-ized scour or stripping of the floodplain during higher dis-charges (Piégay, 2003). The manner in which overbank flowaffects floodplains depends on factors such as the ratio of shearstress generated by the flow versus floodplain erosional resis-tance and the concentration of suspended material in theoverbank flow (Wohl, 2013).


Figure 6. Aerial view of a naturally formed wood raft on a side chan-nel of the Slave River, Northwest Territories, Canada. White oval con-tains a person; channel width in the foreground is approximately 70m. Photograph courtesy of Natalie Kramer.


Studies of the effects of in-channel LWon floodplain sedimentdynamics emphasize how the obstructions created by LWenhance either overbank flows and vertical accretion or bankerosion, channel avulsion, and formation of secondary chan-nels. Working on the HighlandWater (gradient 0.0085, bankfullwidth 5m) in the UK, Jeffries et al. (2003) documented increasedfrequency and extent of overbank flows because of in-channellogjams. They monitored nine floods during the winter of 2000to 2001 with mean overbank deposition rates of 1.6 to 8kg/m2, although rates were as high as 28 kg/m2 in the vicinityof in-channel logjams. Overbank deposition reflected floodmagnitude and suspended sediment load, but the high deposi-tion rates around logjams were significantly above values re-corded for other lowland rivers in Europe with less LW (Jeffrieset al., 2003). Sear et al. (2010) demonstrated that enhancedfloodplain deposition along this channel can occur at sites dis-tant from the main channel when logjams divert sediment-ladenwater into floodplain channels. LW-induced overbank sedimen-tation can be a primary mechanism for vertical accretion onsome floodplains (Nanson et al., 1995; Oswald and Wohl,2008). LW can also influence lateral accretion by altering therates and mechanisms of channel avulsion and migration.Accumulations of wood can accelerate migration or avulsion,but wood can also dampen channel migration, as in thescenario when a logjam blocks a chute cutoff along a sinuouschannel (Hickin, 1984).As with other aspects of LW–sediment interactions, the

mechanisms and relative importance of wood-induced changesin floodplain sedimentation or erosion can scale with river size.Channel-spanning in situ jams (Abbe and Montgomery, 2003)are more likely to enhance channel-floodplain connectivityand associated changes in floodplain sedimentation in smallerrivers (Wohl, 2011), but wood rafts can have similar effects invery large rivers (Triska, 1984). Marginally deposited LW thatlimits or directs overbank flow (Piégay, 2003) can stronglyinfluence floodplain sedimentation in large rivers. The primaryscale effect may be that progressively larger rivers typicallyhave progressively wider and more longitudinally continuousfloodplains, although this is not always the case.Few studies of sedimentation associated with beaver dams

consider overbank deposition outside of a defined pond. Excep-tions include Scheffer (1938, who described beaver-induceddeposition of 4470 m3 of sediment behind 22 dams along a620-m-length of Mission Creek, Washington, USA, Krameret al. (2012), who used near-surface geophysical methods toquantify beaver-induced sediment storage over a 2 km2 area


in Rocky Mountain National Park, Colorado, USA, andWestbrook et al. (2011, who documented 750 m3 of sedimentdeposition over a 1.1 ha area along the upper Colorado Riverfloodplain in Rocky Mountain National Park, Colorado, USA.Each of these studies examined the cumulative sediment storageassociated with a beaver meadow.

Influences of wood on particulate organic matter(POM) dynamics

Most studies of POM dynamics have been conducted by streamecologists (e.g. Webster et al., 1999). They divide POM into finePOM (0.45 μm to 1 mm) and coarse POM (> 1 mm). Theamount of POM in streams reflects input rates, abiotic and bioticprocessing rates, stream transport capacity, and retentionstructures such as LW (Naiman and Sedell, 1979; Raikowet al., 1995). LW is thus only one of several controls on POMstorage, but several studies demonstrate that greater wood loadscorrespond to greater POM storage (Bilby, 1981; Bilby andWard, 1989; Raikow et al., 1995; Thompson, 1995; Beckmanand Wohl, 2014a). This appears to be the case even in low-gradient channels in which LW is highly mobile (Daniels,2006). In low-gradient, sand-bed, headwater streams, LW jamscan facilitate trapping and burial of POM during higherdischarges when much of the stream bed is mobile (Jones andSmock, 1991).

Experimental releases of POM demonstrate that much ofthe material retained within the channel is associated with LW(e.g. Millington and Sear, 2007). When all LW jams were exper-imentally removed from a 175-m-long segment of HubbardBrook Watershed 5 (channel width 2.8 m, gradient 0.213m/m) in New Hampshire, USA, for example, POM exportincreased from 1450 kg/year to 9150 kg/year (Bilby, 1981). Inaddition to trapping and storing POM produced from othersources, LW can be a source of finer POM. Ward and Aumen(1986) found that POM resulting from wood processing via mi-crobial activity and physical abrasion created several times thePOM contributed by leaf and needle litter in western Oregon’sgravel-bed Mack Creek. By enhancing hyporheic exchange(Sawyer et al., 2012), in-channel LW may also influencemovement of POM between the surface and subsurface.

Working in the Cascade Mountains of Oregon, USA, Naimanand Sedell (1979) showed that benthic POM was inverselyproportional to stream order. Annual values of stored POM,including LW, ranged from 26 kg/m2 in the smallest channelto 0.8 kg/m2 in the largest channel (the seventh order McKenzieRiver). In all channels, LWaccounted for> 90% of this material.Studies elsewhere also document decreasing volumes of POMstorage as channel size increases (e.g. Bilby and Ward, 1989)and a smaller proportion of POM storage being associated withLW as channel size increases (e.g. Bilby and Likens, 1980).Existing studies of in-stream POM have focused almost exclu-sively on smaller rivers draining areas smaller than about 100km2 (Tank et al., 2010), although a separate line of research inriparian ecology quantifies organic matter and carbon contentof riparian and floodplain soils on large rivers (e.g. Hoffmannet al., 2009; Cierjacks et al., 2010 ).

As with mineral sediment, temporal fluctuations inLW-induced POM storage occur as a result of fluctuations indischarge and associated trapping capacity within a river orfloodplain, POM inputs, and wood load. POM is continuallyproduced within the riparian zone and adjacent uplands,decays relatively quickly, and is typically readily mobilized.Consequently, fluctuations in POM supply and discharge canproduce substantial variations in POM storage over timescales



of less than a year (e.g. Heartsill Scalley et al., 2012), whereaschanges in wood load are likely to create variations in storageover longer timescales. Analogous to wood-induced variationsin storage of mineral sediment, few studies have quantified var-iations in wood-induced POM storage over time periodsgreater than a few years.Beaver dams are also very effective at trapping POM in the

pond associated with each dam. Examining boreal forest drain-age networks in Quebec, Canada, for example, Naiman et al.(1986) documented a benthic standing stock of coarse POMof 125 g/m2 in riffles and 4075 g/m2 in beaver ponds. Althoughthe ecologists who have primarily investigated this topic tend tofocus on fluxes rather than storage (e.g. Naiman and Melillo,1984; Correll et al., 2000), their work clearly indicates thatPOM fluxes are lower through beaver ponds than through adja-cent riffles or through the same stream segment without a bea-ver dam. As Naiman et al. (1994) demonstrated for boreal forestdrainage networks in Minnesota, USA, beaver activities shiftnutrient and ion storage from forest vegetation to sediments.

Effects of wood removal on sediment dynamics

Wood removal can occur directly when wood is physicallytaken out of a channel or floodplain for purposes of flood con-trol (Gendaszek et al., 2012), navigation (Harmon et al., 1986),or limiting potential damage by mobile wood during floods(Mazzorana et al., 2011; Mao et al., 2013; Ruiz-Villanuevaet al., 2014). Wood removal can also occur indirectly when re-cruitment of new wood to a channel or floodplain is reducedor eliminated by timber harvest or other changes in land cover(Montgomery et al., 2003a, 2003b; Wohl, 2014). Once woodis removed, the ability of a channel to retain subsequentlyrecruited wood can decrease. Wood within a channel or flood-plain, particularly less mobile pieces, creates obstructions towood in transport and can therefore help to trap otherwisemobile wood (Braudrick et al., 1997; Braudrick and Grant,2001; Bocchiola et al., 2008; Wohl and Beckman, 2014).Direct or indirect wood removal can therefore continue toinfluence wood loads along river corridors even after the woodremoval ceases.Direct and indirect wood removal has been occurring for

many centuries in Eurasia (Schama, 1995; Montgomery et al.,2003a, 2003b; Comiti, 2012) and for more than a century inthe Americas (Wohl, 2014), Australia (Brierley et al., 1999),and New Zealand (Mackay, 1991). Wood removal has affectedentire river networks, from the smallest headwater channels tothe major rivers on these continents, with the result that thegreat majority of rivers in high-income countries now havewood loads that are orders of magnitude lower than woodloads prior to the Industrial Revolution. The net effect on sedi-ment dynamics of removing LW is to reverse all of the influ-ences discussed in the earlier parts of this paper. Themagnitude of these effects has not been quantitatively esti-mated at the largest spatial and temporal scales, but smallerscale studies provide some insight into the effects of woodremoval.When LW is removed from a channel, bedload transport

rates can increase significantly. Experimental removal of LWfrom gravel-bed Bambi Creek (drainage area 155 ha, gradient0.010, channel width 3.9 m) in southeastern Alaska, USAcaused a four-fold increase in bedload transport during bankfulldischarge for a year after LW removal (Smith et al., 1993).Wood removal can result in decreased volumes of stored sedi-ment, as well as centralizing the locations of sediment storage(Klein et al., 1987). Pool volume can also decrease substantially(Bisson and Sedell, 1984), as observed during the first high


flow following LW removal along gravel-bed Salmon Creek(drainage area 900 ha, gradient 1.5%, bankfull width 11.5 m)in western Washington, USA (Bilby, 1984). Wood removalalong the sand-bed Little Sioux River (drainage area 69 km2,channel gradient 0.5%, channel width 4.7 m before LWremoval) in Wisconsin, USA resulted in 25% narrowing ofmean channel width, 32% increase in mean velocity (but notthalweg velocity), incision of the bed, 58% reduction in sandbedload, and 400% increase in coarse channel substrate atthe 300-m reach evaluated in the study (Dumke et al., 2010).

Removal of LW throughout a river corridor also stronglyinfluences floodplain dynamics. As demonstrated most thor-oughly for rivers in the Pacific Northwest region of the UnitedStates, LW removal has caused anastomosing rivers to becomesingle-thread, in association with a loss of channel-floodplainconnectivity and floodplain physical and biotic diversity(Collins et al., 2002; Gendaszek et al., 2012). Removal of his-torically occurring wood rafts on large lowland rivers in thesouth-eastern United States has resulted in decreased flood-plain sedimentation rates (Barrett, 1996; Patterson et al., 2003).

Removal of beavers and their dams has also been practicedfor centuries in association with fur trapping (Naiman et al.,1988) and, more recently, as a means of controlling floodingand the mobility of small in-stream wood along river corridors.As with in-stream LW removal, the net effects of removing bea-vers and beaver dams are to increase velocity, reduce in-streamsediment storage, reduce hyporheic exchange, reduceoverbank flows and vertical floodplain accretion, and changemulti-thread channels to single-thread planforms (Naimanet al., 1988; Marston, 1994; Green and Westbrook, 2009; Polviand Wohl, 2013).

The spatial scale and magnitude of the effects on sedimentdynamics of LWand beaver removal likely varied among differ-ent types of channels, just as the effects caused by the presenceof LW and beaver vary among channels. Removal of beaverlikely had the greatest influence on the smaller rivers (approxi-mately ≤ 30 m bankfull channel width) that beaver are morelikely to dam, although removing many beaver dams fromheadwater areas within a river network could have a substan-tial cumulative effect on sediment yield to downstream portionsof the network. Removal of LW jams may have substantially re-duced storage of bedload, and particularly the finer portion ofbedload, in steep channels. Removal of dispersed LW is likelyto have reduced in-stream sediment storage in moderatelysized, pebble to sand-bed channels, as illustrated by work inAustralia (Gippel et al., 1992; Brooks et al., 2003), and removalof LW rafts likely reduced floodplain sedimentation in largerivers (Barrett, 1996; Patterson et al., 2003).

Patterns of Wood and Sediment in Rivers

Numerous factors can potentially influence wood loads andassociated sediment dynamics in river corridors. Proximalcontrols include the recruitment of LWand delivery of sedimentto rivers and the flow regime that influences distribution of LWand sediment. Distal controls include geologic history, lithol-ogy, climate, and land use as these influence the downslopemovement of water, sediment, and LW. Despite the importanceof these variables in controlling observed quantities of LW andsediment storage along river corridors, the lack of consistentlyreported and relevant quantitative measures of these variablesin many LW studies limits our ability to statistically analyzetheir effect. Consequently, we focus on potential patternsbetween LW and sediment in relation to commonly reportedriver variables.



Volume of stored sediment

We used the data in Table I to evaluate trends in stored sedi-ment in relation to drainage area, channel gradient, bankfullchannel width, and wood load (Figure 7). Sediment storageassociated with wood was estimated over different spatialscales (channel width versus reach-scale) and using slightlydifferent field methods in all of the studies reviewed for this

Figure 7. Sediment volume per unit area of channel in relation to (A)drainage area, (B) bankfull channel width, and (C) volume of wood perunit area of channel for the data in Table I. Data point surrounded by adashed oval represents an outlier from an Australian site. Each datapoint represents a reach-scale average, so points designated as jammedsteps reflect a reach in which the majority of sediment storage was as-sociated with jammed steps.


analysis. We chose only those studies from which we couldderive volume of sediment per unit area of channel surface,either by using the data reported in the paper or, in some cases,by contacting the authors of a paper and requesting supple-mental data. Despite our efforts to standardize data presenta-tion, it is important to note that these data reflect differentfield methods and should be regarded as a reasonable approx-imation of patterns of wood and sediment storage across rivers,rather than precise and fully comparable quantities.

We hypothesized that sediment volume stored in associationwith LW would exponentially decrease downstream in relationto drainage area (Figure 1A). Figure 7A weakly supports thishypothesized trend, with the exception of an outlier fromAustralia. A bivariate plot of channel gradient versus sedimentvolume indicates no relationship. The bivariate plot of channelwidth versus sediment volume (Figure 7B) suggests a peak instored sediment at intermediate values of channel width,although this peak is based on only five data points. We alsohypothesized that sediment volume would increase with woodvolume (Figure 1B). Figure 7C weakly suggests such trends,although it remains unclear whether this relationship is betterdescribed as a linear or an exponential increase. The data inFigure 7 do not extend to sufficiently large drainage areas to in-clude the wood rafts included in Figure 1, but the hypothesizeddownstream change from jammed steps to dispersed woodappears to be present in Figures 7A and 7B. We also evaluatedwood volume per unit area against bankfull channel width(sensu Fox and Bolton, 2007). Although a slight trend was pres-ent, with wood load increasing as channel width increased,analogous to the pattern in Fox and Bolton (2007), the relation-ship was not significant at α = 0.10.

The outlier (Figure 7A) from the Australian site with largerdrainage area is important. Historical records from NorthAmerica and elsewhere suggest that substantial volumes ofLWwere present in large channels (Wohl, 2014). This LW likelycreated significant sediment storage within channels and acrossfloodplains, but sediment storage was not measured prior tointensive and sustained LW removal during the second half ofthe nineteenth century and first half of the twentieth century.If the Australian outlier in the upper right of Figure 7A is repre-sentative of other rivers with larger drainage areas, our hypoth-esized relationship in Figure 1A could be backwards, with thelargest volumes of sediment storage per unit area actuallyoccurring in larger, or at least mid-sized, rivers. At present,we do not have the data to evaluate this hypothesis.

We utilized multivariate analysis to model sediment volumeusing drainage area, channel gradient, bankfull channel width,and wood load. We performed all subsets multiple linearregression in the R statistical package (R Core Team, 2014)and used the corrected Akaike Information Criterion (AICc)weights to evaluate 16 potential models (Wagenmakers andFarrell, 2004). To accommodate the non-normality of thedataset, we used a square root transform on all variables beforeperforming multiple linear regression. We used Akaike weights(Burnham and Anderson, 2002) to rank the relative importanceof each variable in the model to better understand which pre-dictors exerted a dominant influence on residual pool volume.We found a significant (p < 0.001) correlation (Ra

2 = 0.39)(Figure 8A) between drainage area, wood volume, and storedsediment volume, such that stored sediment increases linearlywith drainage area and wood volume. Drainage area was themost important predictor (weight = 1.0), followed by woodvolume (weight = 0.79). We also tested the sediment volumeprediction relation proposed by Nakamura and Swanson(1993) (note that our model uses sediment volume per unitarea, not per unit river length) using the dataset compiled here.We were able to moderately predict sediment volume using


Figure 8. Comparison of predicted versus measured sediment volumeby applying: our multiple linear regression to the entire dataset (A), theNakamura and Swanson (1993) relationship to the entire dataset (B),and the Nakamura and Swanson (1993) relationship applied to onlyjammed steps (C). A 1:1 line is shown in each plot to represent the be-havior of an ideal model. The performance of the model can be evalu-ated by comparing the distance of each point from the 1:1 line. Noticethat the model tends to perform better for data with low measuredsediment volume.


width and gradient (Ra2 = 0.26) (Figure 8B). Applying the

Nakamura and Swanson (1993) relation to only jammed stepsresulted in a similar predictive ability (Ra

2 = 0.25) (Figure 8C).Our dataset weakly supports the idea of a power function relat-ing sediment volume to width and slope: as gradient and widthincrease, the volume of sediment behind a jam increases. It isimportant to note, however, that both the Nakamura and


Swanson (1993) relationship and our multiple linear regressionperform poorly for larger sediment volumes (Figure 8). Both ofthese models show an increase in sediment volume down-stream (positive correlation with either width or drainagearea), contradicting our hypothesis that sediment volume de-creases downstream in a river network. This could indicatethat (i) wood mobility is not tied to sediment storage, (ii) dis-persed LW is capable of creating greater sediment storageper unit area than the logjams that are common in the smallerstreams in our dataset, or (iii) sediment storage has a peakedfunction, with increased sediment storage in mid-sized riversand lower sediment storage in small and large rivers. As notedearlier, we cannot test the third possibility because of an ab-sence of data from large rivers.

In testing our hypothesis regarding the effect of wood volumeon sediment volume, we added a wood volume term to theNakamura and Swanson (1993) relation, but we found that thewood volume term was most likely not significant in the model.Thus, our hypothesis that increasing wood volume shouldincrease sediment volume is supported by our multiple linearregression, but not by the relationship of Nakamura andSwanson (1993). Wood volume likely plays some role incontrolling sediment volume, although there may be a thresholdsediment volume above which wood fails to control sedimentvolume, as indicated by the poor performance of the testedmodels at high sediment volumes. Our analysis fails to test forthe possibility of a threshold sediment storage response thatmay occur due to the presence of extremely LW jams.

One of the challenges in assembling the data for theseanalyses was the lack of consistent methods for quantifyingsediment storage. Standardized measurements that allowsediment storage associated with LW to be quantified on thebasis of per unit area of channel would be very useful.

Residual pool volume

We used the data in Table II to evaluate trends in residual poolvolume in relation to drainage area, channel gradient, bankfullchannel width, and wood load (Figure 9). We hypothesizedthat residual pool volume would decrease as drainage area in-creased (Figure 2). A bivariate plot of drainage area versus poolvolume (Figure 9A) contains so much scatter that there is littlesupport for this hypothesis. Similarly, Figure 9B does not sug-gest any correlation between channel width and pool volume.Figures 9A and 9B also do not indicate any trend with respectto the occurrence of plunge pools in relation to drainage areaor channel width. Pool volume is slightly more likely to in-crease with LW volume (Figure 9C), but there is substantialscatter in the data. Analogous to Figure 7, the data in Figure 9do not extend to sufficiently large rivers to include wood rafts(Figure 2), so we cannot evaluate the potential for decliningpool volume per unit area in much larger rivers.

We modeled residual pool volume using multiple linearregression in a similar method to our linear modeling of sedi-ment volume. The best model explained the majority of thevariance in the data (Ra

2 = 0.63, p < 0.0001) by utilizing chan-nel gradient and wood volume as predictors (Figure 10). Woodload (weight = 1.00) is the most important predictor, followedby channel gradient (weight = 0.98). Pool volume correlatesnegatively with slope and positively with wood volume, suchthat lower slopes and higher wood volumes produce greaterpool volumes. This contradicts our hypothesis that pool volumedecreases downstream, assuming that slope also decreasesdownstream. This could reflect (i) the greater ability of flowdeflected by LW to scour a bed that is likely to be composedof finer grained substrate at lower gradients or (ii) the


igure 10. Comparison of measured pool volume data to volume pre-icted by multiple linear regression using channel gradient and woodolume as predictors. A 1:1 line is shown to represent the behavior ofn ideal model. The performance of the model can be evaluated byomparing the distance of each point from the 1:1 line.

Figure 9. Residual pool volume per unit area of channel in relation to(A) drainage area, (B) bankfull channel width, and (C) volume of woodper unit area of channel for the data in Table II. Solid black data pointsrepresent plunge pools. Each data point represents a reach-scale aver-age, so points designated as plunge pools reflect a reach in which themajority of pools formed by this mechanism.


more extensive backwaters associated with logjams at lowerstream gradients.The results from the multiple linear regression also seemingly

contradict the interpretations from the bivariate plots (Figure 9).Multiple linear regression accounts for changes in the modeledvariable with respect to the association of many predictorvariables, as opposed to the individual effects of each variable.The sum of the effects of all variables might be able to predict


Fdvac

the modeled variable when each individual variable by itselfcannot. Consequently, we consider the results from themultiple linear regression analyses to be more indicative ofactual relationships between pool volume and potential controlvariables than the results from bivariate analyses.

We did not perform similar analyses for relations betweenLW volumes and POM storage because of a lack of publishedstudies that report volumes or weights of in-stream LW andPOM storage. We found only two studies (Naiman and Sedell,1979; Beckman and Wohl, 2014a) that report these types ofdata. LW clearly exerts an important influence on POM stor-age, as reviewed earlier, and the lack of studies quantifyingthis effect represents a significant gap in our understandingof how forest, channel, and LW characteristics interact toinfluence storage of particulate material within rivers.

Volume of sediment stored in beaver ponds

Data from the three studies with direct measurements ofbeaver pond area and sediment volume indicate that sedimentvolume increases with pond area (Figure 11A). We identifiedand removed two outlying data points that represent extremesediment volumes in order to obtain a better model for themajority of the dataset (low sediment volumes and smallponds). The model should not include an intercept term, con-sidering that a pond of zero area should hold no sediment. Alinear regression through the data had highly non-normal errorvariance, leading to our choice of a logarithmic transfor-mation of the data, which satisfied linear regression assump-tions (Figure 11B). The strength of this relation is uncertain,however, because of the small dataset. Insufficient data existto understand the relationship between beaver ponds andstored sediment for extremely large beaver ponds.

The stratigraphy and soil characteristics of diverse forestedregions in the northern hemisphere suggest that much largerbeaver dams and ponds existed in the past (Morgan, 1868;Mills, 1913). At least one very large beaver pond has beenidentified from remote sensing imagery in northern Canada(Discovery News, 2010), but this site has not yet beenmeasured. Systematic field measurements of the largest beaverponds still in existence would help to clarify the relationbetween beaver pond area and sediment volume.


Figure 11. Beaver pond area versus volume of stored sediment withinthe pond, using data in Table III (A) and with outliers removed andmodel fit shown (B). The Spearman correlation coefficient for (A) is0.89.


Observed patterns relative to initial hypotheses

As explained in the introductory sections, we hypothesized ei-ther that the volume of sediment per unit area of channel storedin association with in-stream LW declines in a downstream di-rection as drainage area increases or the presence of LW is themost important influence on the volume of sediment stored andtherefore sediment storage correlates most strongly with woodload. Multiple linear regression analyses indicate a significantrelation in which sediment volume increases linearly withdrainage area and wood volume. The lack of support for the hy-pothesized trend between sediment storage and drainage areacould reflect either a poor understanding of how LW influencessediment storage in larger rivers or the limitations of the dataanalyzed here, which do not adequately represent medium tolarge rivers. We also hypothesized that the residual volume ofpools created in association with LW declines in a downstreamdirection. Multiple linear regression analyses indicate thatresidual pool volume increases as LW volume increases, butis greater downstream. Finally, we hypothesized sedimentvolume behind beaver dams will correlate positively with pondarea, a relationship that the results support.An important consideration with respect to our ability to test

hypotheses regarding downstream trends in sediment storageand pool volume is the limitations of this dataset. All of thechannels for which LW and sediment data were obtained havea gradient ≥ 0.002 and most of the channels for which LW andpool volume data were obtained have a gradient ≥ 0.01. In


other words, we are analyzing data primarily from small, steep,mountain streams, presumably because such streams areamong the few remaining river segments with old-growth ornatural forest and minimal flow regulation and channel engi-neering. The trends that we initially hypothesized might exist,but quantitative measurements of LW, sediment storage, andpool volume focused on medium to large rivers are needed tofully test these hypotheses.

Conclusions

Diverse investigations of the relatively small amounts of LWand beaver dams still present in river corridors clearly indicatethe importance of these LW and dams for sediment dynamics.Obstructions from individual LW pieces to logjams, beaverdams, and wood rafts can significantly alter entrainment,transport, and deposition of POM, suspended sediment, andbedload at spatial scales from the immediate vicinity of theobstruction to river reaches 101–102 m long. Over relativelysmall drainage areas (0–200 km2), the volume of sedimentstored within channels in association with LW tends to increaseas drainage area and LW increase, and the residual pool vol-ume tends to increase with LW load and with declining gradi-ent, but data that would allow us to examine patterns at largerdrainage areas do not exist. Similarly, the volume of sedimentstored in beaver ponds increases with pond area, althoughthe details of this relationship are unclear because of a lack ofdata from larger beaver ponds. One of the more important in-sights arising from our review of the existing literature on woodand sediment dynamics in rivers is the lack of studies on largerriver systems. Most, or perhaps all, large rivers within the tem-perate zone have been so altered by systematic removal ofLW and beaver that investigations of these systems will needto focus on characteristics of floodplain stratigraphy or, wherepresent, abandoned and preserved paleochannels (Howardet al., 1999; Davies et al., 2014). Large rivers at higher or lowerlatitudes may provide useful insights into wood and sedimentdynamics, although tropical rivers, in particular, are likely tohave such rapid rates of wood decay (Cadol et al., 2009) thatwood may not influence sediment dynamics in the mannerdiscussed in this paper.

In conclusion, it is worth emphasizing that wood was oncemuch more abundant, from channel-spanning LW steps insteep, headwater channels to wood rafts and dispersed pieceswithin major lowland rivers. Similarly, millions of beavers onceinhabited river corridors throughout the northern hemisphere.Although the trends presented here between LW volume andeither stored sediment volume or residual pool volume only ap-ply to the limited data and range of channel sizes used for thissynthesis paper, all the available evidence suggests that moreabundant wood equated to greater volumes of sediment storedand greater residual pool volume in diverse channels. The neteffect of direct and indirect LW and beaver removal from rivercorridors has been to increase channel conveyance for waterand sediment and to decrease the physical complexity of chan-nels and floodplains. One of the management implications ofthese changes is that actively re-introducing LW and beaversand/or limiting active removal of naturally recruited wood fromrivers can be used to restore river process and form.

Active design and placement of individual pieces of LWor ofengineered log jams is increasingly being used as part of riverrestoration (Abbe et al., 2003; Abbe and Brooks, 2011;Gallisdorfer et al., 2014), as is re-introduction of beavers orbuilding dams that mimic those built by beavers (Pollocket al., 2014). LW and beaver dams can be used to enhance fishhabitat, trap sediment, stabilize channels, and enhance



hyporheic exchange (Shields et al., 2004; Brooks et al., 2004,2006). Understanding of how LW interacts with flow forces,sediment in flux, and channel boundaries can also be used inpassive restoration that facilitates LW recruitment and allowsLW within the river corridor to influence channel and flood-plain process and form (e.g. Piégay et al., 2000).The very limited published quantifications of in-stream LW

and associated volumes of stored sediment and residual pools,and the even more limited published quantification of beaverdams and sediment storage, highlight a data gap in the li-terature. Our analyses point to a correlation between storedsediment volume, drainage area, and wood volume, as wellas a correlation between residual pool volume, slope, andwood volume. These analyses indicate that wood volume, inconjunction with drainage characteristics, plays a major rolein controlling the geomorphically and ecologically importantfactors of volume of stored sediment and pool volume withina channel. Specific focus on quantifying relations amongwood, sediment, pools, and beaver dams, including detailedfield measurements of these variables, could help to expandthe analyses presented here and provide clearer guidelines forrestoring or protecting in-stream and floodplain LW and beaverpopulations in order to facilitate storage of sediment andcreation of pools. The negation of our hypotheses regardingsediment and pool volume highlights our current lack of under-standing of these relationships. An explicit focus on howsediment storage varies through time in relation to fluctuationsin LW recruitment and storage, as well as in relation to fluctuat-ing water and sediment discharges, is also missing from theliterature.In future studies, it would be particularly valuable to specify

the area of channel which a set of wood and sedimentmeasurements represent (e.g. channel-width versus reach-scale). Information that is integral to understanding wood andsediment relations in river corridors includes the most detaileddescription possible of land-use and natural disturbance his-tory, stand age of the riparian forest, forms of LW recruitment(e.g. bank erosion versus hillslope failure), and contemporaryconditions of the river corridor with respect to LW recruitment.The last point can help to evaluate whether LW present in theriver corridor represents a legacy of past human- (e.g. timberharvest) or naturally induced recruitment or whether theLW represents contemporary processes. Understanding theconditions under which LW is recruited can help to under-stand differences in wood load, piece size, spatial distribution(e.g. sunken logs versus unattached pieces), and decay rates, allof which can vary between old-growth forest and fast-growingpioneer tree species (Bunn and Montgomery, 2003; Chenet al., 2005; Keeton et al., 2007; Beckman and Wohl, 2014b).These differences in LW characteristics and distribution cansignificantly affect geomorphic function of the LW, and mightexplain some of the observed variability between sites, but wewere unable to gather sufficient supporting data to evaluatethese effects in the analyses reported here.

Acknowledgements—The authors thank Stuart Lane and two anony-mous reviewers for useful comments on an earlier draft of this paper.The authors also thank Sandra Ryan for sending additional data foranalyses and Ann Hess for advice on statistical analysis.

ReferencesAbbe T, Brooks A. 2011. Geomorphic, engineering, and ecologicalconsiderations when using wood in river restoration. In StreamRestoration in Dynamic Fluvial Systems: Scientific Approaches,Analyses, and Tools, Simon A, Bennett SJ, Castro JM (eds). AmericanGeophysical Union Press: Washington, DC; 419–451.


Abbe TB, Brooks AP, Montgomery DR. 2003. Wood in river rehabilita-tion and management. In The Ecology and Management of Wood inWorld Rivers, Gregory SV, Boyer KL, Gurnell AM (eds). AmericanFisheries Society: Bethesda, MD; 367–389.

Abbe TB, Montgomery DR. 1996. Large woody debris jams, channelhydraulics and habitat formation in large rivers. Regulated Rivers:Research and Management 12: 201–221.

Abbe TB, Montgomery DR. 2003. Patterns and processes of wooddebris accumulation in the Queets River basin, Washington. Geo-morphology 51: 81–107.

Adenlof KA, Wohl EE. 1994. Controls on bedload movement in asubalpine stream of the Colorado Rocky Mountains, USA. Arcticand Alpine Research 26: 77–85.

Andreoli A, Comiti F, Lenzi MA. 2007. Characteristics, distribution andgeomorphic role of large woody debris in a mountain stream of theChileanAndes. Earth Surface Processes and Landforms 32: 1675–1692.

Andrus CW, Long BA, Froehlich HA. 1988. Woody debris and its con-tribution to pool formation in a coastal stream 50 years after logging.Canadian Journal of Fisheries and Aquatic Sciences 45: 2080–2086.

Assani AA, Petit F. 1995. Log-jam effects on bed-load mobility fromexperiments conducted in a small gravel-bed forest ditch. Catena25: 117–126.

Barrett ML. 1996. Environmental reconstruction of a 19th century RedRiver raft lake: Caddo Lake, Louisiana and Texas. American Associa-tion of Petroleum Geologists Bulletin 80: 1495(abstract).

Beckman ND, Wohl E. 2014a. Carbon storage in mountainous head-water streams: the role of old-growth forest and logjams. WaterResources Research 50: 2376–2393.

Beckman ND, Wohl E. 2014b. Effects of forest stand age on the charac-teristics of logjams in mountainous forest streams. Earth SurfaceProcesses and Landforms 39: 1421–1431.

Benda LE, Sias JC. 2003. A quantitative framework for evaluating themass balance of in-stream organic debris. Forest Ecology andManagement 172: 1–16.

Bertoldi W, Gurnell AM, Welber M. 2013. Wood recruitment andretention: the fate of eroded trees on a braided river explored usinga combination of field and remotely-sensed data sources. Geomor-phology 180–181: 146–155.

Bilby RE. 1981. Role of organic debris dams in regulating the export ofdissolved and particulate matter from a forested watershed. Ecology62: 1234–1243.

Bilby RE. 1984. Removal of woody debris may affect stream channelstability. Journal of Forestry 82(10): 609–613.

Bilby RE, Likens GE. 1980. Importance of organic debris dams in thestructure and function of stream ecosystems. Ecology 61: 1107–1113.

Bilby RE, Ward JW. 1989. Changes in characteristics and function ofwoody debris with increasing size of streams in western Washington.Transactions of the American Fisheries Society 118: 368–378.

Bisson PA, Sedell JR. 1984. Salmonid populations in streams in clearcutvs old-growth forests of western Washington. In Fish and Wildlife Re-lationships in Old-growth Forests, Meehan WR, Merrell TR, HanleyTA (eds). American Institute of Fisheries Research Biologists: Juneau,AK; 121–129.

Bocchiola D, Rulli MC, Rosso R. 2006. Transport of large woody debrisin the presence of obstacles. Geomorphology 76: 166–178.

Bocchiola D, Rulli MC, Rosso R. 2008. A flume experiment on theformation of wood jams in rivers. Water Resources Research 44:W02408. DOI:10.1029/2006WR005846.

Boivin M, Buffin-Bélanger T, Piégay H. 2015. The raft of the Sain-JeanRiver, Gaspé (Québec, Canada): a dynamic feature trapping most ofthe wood transported from the catchment. Geomorphology 231:270–280.

Bragg DC. 2000. Simulating catastrophic and individualistic large woodydebris recruitment for a small riparian system. Ecology 81: 1383–1394.

Braudrick CA, Grant GE. 2001. Transport and deposition of largewoody debris in streams: a flume experiment. Geomorphology 41:263–283.

Braudrick CA, Grant GE, Ishikawa Y, Ikeda H. 1997. Dynamics of woodtransport in streams: a flume experiment. Earth Surface Processes andLandforms 22: 669–683.

Brierley GJ, Cohen TC, Fryirs KA, Brooks AP. 1999. Post-Europeanchanges to the fluvial geomorphology of Bega Catchment, Australia:implications for river ecology. Freshwater Biology 41: 839–848.



Brooks AP, Brierley GJ. 2002. Mediated equilibrium: the influenceof riparian vegetation and wood on the long-term evolution andbehavior of a near-pristine river. Earth Surface Processes andLandforms 27: 343–367.

Brooks AP, Brierley GJ, Millar RG. 2003. The long-term control ofvegetation and woody debris on channel and flood-plain evolution:insights from a paired catchment study in southeastern Australia.Geomorphology 51: 7–29.

Brooks AP, Gehrke PC, Jansen JD, Abbe TB. 2004. Experimental reintro-duction of woody debris on the Williams River, NSW: geomorphicand ecological responses. River Research and Applications 20:513–536.

Brooks AP, Howell T, Abbe TB, Arthington AH. 2006. Confronting hys-teresis: wood based river rehabilitation in highly altered riverinelandscapes of south-eastern Australia. Geomorphology 79: 395–422.

Brummer CJ, Abbe TB, Sampson JR, Montgomery DR. 2006. Influenceof vertical channel change associated with wood accumulations ondelineating channel migration zones, Washington, USA. Geomor-phology 80: 295–309.

Buffington JM, Lisle TE, Woodsmith RD, Hilton S. 2002. Controls on thesize and occurrence of pools in coarse-grained forest rivers. River Re-search and Applications 18: 507–531.

Buffington JM, Montgomery DR. 1999. Effects of hydraulic roughnesson surface textures of gravel-bed rivers. Water Resources Research35: 3507–3521.

Bugosh N, Custer SG. 1989. The effect of a log-jam burst on bedloadtransport and channel characteristics in a headwater stream. InHeadwaters Hydrology, Potts DF, Woessner WW (eds). AmericanWater Resources Association: Bethesda, MD; 203–211.

Bunn JT, Montgomery DR. 2003. Patterns of wood and sediment stor-age along debris-flow impacted headwater channels in old-growthand industrial forests of the western Olympic Mountains, Washing-ton. In Riparian Vegetation and Fluvial Geomorphology, Bennett SJ,Simon A (eds). American Geophysical Union Press: Washington,DC; 99–112.

Burnham KP, Anderson DR. 2002. Model Selection and Multimodel In-ference: A Practical Information-theoretic Approach. Springer: Berlin.

Butler DR. 2012. Characteristics of beaver ponds on deltas in amountainenvironment. Earth Surface Processes and Landforms 37: 876–882.

Butler DR, Malanson GP. 1995. Sedimentation rates and patterns inbeaver ponds in amountain environment.Geomorphology 13: 255–269.

Butler DR, Malanson GP. 2005. The geomorphic influences of beaverdams and failures of beaver dams. Geomorphology 71: 48–60.

Cadol D, Wohl E. 2011. Coarse sediment movement in the vicinity of alogjam in a neotropical gravel-bed stream. Geomorphology 128:191–198.

Cadol D, Wohl E, Goode JR, Jaeger KL. 2009. Wood distribution in neo-tropical forested headwater streams of La Selva, Costa Rica. EarthSurface Processes and Landforms 34: 1198–1215.

Carlson JY, Andrus CW, Froehlich HA. 1990. Woody debris, channelfeatures, and macroinvertebrates of streams with logged and undis-turbed riparian timber in northeastern Oregon, USA. Canadian Jour-nal of Fisheries and Aquatic Sciences 47: 1103–1111.

Chen X, Wei X, Scherer R. 2005. Influence of wildfire and harvest onbiomass, carbon pool, and decomposition of large woody debris inforested streams of southern interior British Columbia. Forest Ecologyand Management 208: 101–114.

Cherry J, Beschta RL. 1989. Coarse woody debris and channel mor-phology: a flume study. Water Resources Bulletin 25: 1031–1036.

Chin A, Wohl E. 2005. Toward a theory for step pools in streamchannels. Progress in Physical Geography 29: 275–296.

Church M, Zimmermann A. 2007. Form and stability of step-poolchannels: research progress. Water Resources Research 43W03415. DOI:10.1029/2006WR005037.

Cierjacks A, Kleinschmit B, Babinsky M, Kleinschroth F, Markert A,Menzel M, Ziechmann U, Schiller T, Graf M, Lang F. 2010. Carbonstocks of soil and vegetation on Danubian floodplains. Journal ofPlant Nutrition and Soil Science 173: 644–653.

Collins BD, Montgomery DR, Haas AD. 2002. Historical changes in thedistribution and functions of large wood in Puget Lowland rivers.Canadian Journal of Fisheries and Aquatic Sciences 59: 66–76.

Collins BD, Montgomery DR, Fetherston KL, Abbe TB. 2012. The flood-plain large-wood cycle hypothesis: a mechanism for the physical and


biotic structuring of temperate forested alluvial valleys in the NorthPacific coastal ecoregion. Geomorphology 139–140: 460–470.

Comiti F 2012. How natural are Alpine mountain rivers? Evidence fromthe Italian Alps. Earth Surface Processes and Landforms 37: 693–707.

Comiti F, Andreoli A, Mao L, Lenzi MA. 2008. Wood storage in threemountain streams of the Southern Andes and its hydro-morphological effects. Earth Surface Processes and Landforms 33:244–262.

Correll DL, Jordan TE, Weller DE. 2000. Beaver pond biogeochemicaleffects in the Maryland Coastal Plain. Biogeochemistry 49: 217–239.

Curran JC, Cannatelli KM. 2014. The impact of beaver dams on themorphology of a river in the eastern United States with implicationsfor river restoration. Earth Surface Processes and Landforms 39:1236–1244.

Daniels MD. 2006. Distribution and dynamics of large woody debrisand organic matter in a low-energy meandering stream. Geomor-phology 77: 286–298.

Davidson SL. 2011. Modeling Channel Morphodynamics Associatedwith Large Wood in an Intermediate-sized Stream, MS Thesis. Uni-versity of British Columbia, Vancouver, Canada.

Davidson SL, Eaton BC. 2013. Modeling channel morphodynamicresponse to variations in large wood: implications for stream rehabil-itation in degraded watersheds. Geomorphology 202: 59–73.

Davies NS, Gosse JC, Rybczynski N. 2014. Cross-bedded woody debrisfrom a Pliocene forested river system in the High Arctic Beaufort For-mation, Meighen Island, Canada. Journal of Sedimentary Research84: 19–25.

De Visscher M, Nyssen J, Pontzeele J, Billi P, Frankl A. 2014. Spatio-temporal sedimentation patterns in beaver ponds along the ChevralRiver, Ardennes, Belgium. Hydrological Processes 28: 1602–1615.

Discovery News. 2010. Largest Beaver Dam Seen from Space. http://news.discovery.com/animals/zoo-animals/beaver-dam-canada-space.htm [13 February 2015].

Dumke JD, Hrabik TR, Brady VJ, Gran KB, Regal RR, Seider MJ. 2010.Channel morphology response to selective wood removals in a sand-laden Wisconsin trout stream. North American Journal of FisheriesManagement 30: 776–790.

Eaton BC, Hassan MA, Davidson SL. 2012. Modeling wood dynamics,jam formation, and sediment storage in a gravel-bed stream. Journalof Geophysical Research 117 F00A05. DOI:10.1029/2012JF002385.

Faustini JM, Jones JA. 2003. Influence of large woody debris on channelmorphology and dynamics in steep, boulder-rich mountain streams,western Cascades, Oregon. Geomorphology 51: 187–205.

Fisher GB, Magilligan FJ, Kaste JM, Nislow KH. 2010. Constraining thetimescales of sediment sequestration associated with large woodydebris using cosmogenic 7Be. Journal of Geophysical Research 115F01013.

Fox M, Bolton S. 2007. A regional and geomorphic reference for quan-tities and volumes of instream wood in unmanaged forested basins ofWashington State. North American Journal of Fisheries Management27: 342–359.

Gallisdorfer MS, Bennett SJ, Atkinson JF, Ghaneeizad SM, Brooks AP,Simon A, Langendoen EJ. 2014. Physical-scale model designs forengineered log jams in rivers. Journal of Hydro-environmentResearch 8: 115–128.

Gendaszek AS,Magirl CS, CzubaCR. 2012.Geomorphic response to flowregulation and channel and floodplain alteration in the gravel-beddedCedar River, Washington, USA. Geomorphology 179: 258–268.

Gippel CJ, O’Neill IC, Finlayson BL. 1992. The Hydraulic Basis of SnagManagement. Centre for Environmental Applied Hydrology, Univer-sity of Melbourne: Melbourne.

Gomi T, Sidle RC, Noguchi S, Negishi JN, Nik AR, Sasaki S. 2006. Sed-iment and wood accumulations in humid tropical headwaterstreams: effects of logging and riparian buffers. Forest Ecology andManagement 224: 166–175.

Green KC, Westbrook CJ. 2009. Changes in riparian area structure,channel hydraulics, and sediment yield following loss of beaverdams. BC Journal of Ecosystems and Management 10: 68–79.

Gurnell AM. 2003. Wood storage and mobility. In The Ecology andManagement of Wood in World Rivers, Gregory SV, Boyer KL,Gurnell AM (eds). American Fisheries Society: Bethesda, MD; 75–91.

Gurnell AM, Bertoldi W, Corenblit D. 2012. Changing river channels:the roles of hydrological processes, plants and pioneer fluvial



landforms in humid temperate, mixed load, gravel bed rivers.Earth-Science Reviews 111: 129–141.

Gurnell AM, Sweet R. 1998. The distribution of large woody debris ac-cumulations and pools in relation to woodland stream managementin a small, low-gradient stream. Earth Surface Processes andLandforms 23: 1101–1121.

Harmon ME, Franklin JF, Swanson FJ, Sollins P, Gregory SV, Lattin JD,Anderson NH, Cline SP, Aumen NG, Sedell JR, Lienkaemper GW,Cromack K, Cummins KW. 1986. Ecology of coarse woody debris intemperate ecosystems.Advances in Ecological Research 15: 133–302.

Hart EA. 2002. Effects of woody debris on channel morphology andsediment storage in headwater streams in the Great Smoky Moun-tains, Tennessee-North Carolina. Physical Geography 23: 492–510.

Haschenburger JK, Rice SP. 2004. Changes in woody debris andbed material texture in a gravel-bed channel. Geomorphology 60:241–267.

Hassan MA, Woodsmith RD. 2004. Bed load transport in anobstruction-formed pool in a forest, gravel bed stream. Geomorphol-ogy 58: 203–221.

Heartsill Scalley T, Scatena FN, Moya S, Lugo AE. 2012. Long-termdynamics of organic matter and elements exported as coarse partic-ulates from two Caribbean montane watersheds. Journal of TropicalEcology 28: 127–139.

Hickin EJ. 1984. Vegetation and river channel dynamics. CanadianGeographer 28: 111–126.

Hoffmann T, Glatzel S, Dikau R. 2009. A carbon storage perspective onalluvial sediment storage in the Rhine catchment. Geomorphology108: 127–137.

Hogan DL. 1986. Channel Morphology of Unlogged, Logged, and De-bris Torrented Streams in the Queen Charlotte Islands, Land Manage-ment Report Number 49. British Columbia Ministry of Forests andLands: Vancouver; 94 pp.

Hogan DL. 1989. Channel response to mass wasting in the QueenCharlotte Islands, British Columbia: temporal and spatial changesin stream morphology. Watersheds 89: 125–144.

Hogan DL, Bird SA, Hassan MA. 1998. Spatial and temporal evolutionof small coastal gravel-bed streams: influence of forest managementon channel morphology and fish habitats. In Gravel-bed Rivers inthe Environment, Klingeman PC, Beschta RL, Komar PD, Bradley JB(eds). Water Resources Publications: Highlands Ranch, CO;365–392.

Howard AJ, Smith DN, Garton D, Hillam J, Pearce M. 1999. Middle tolate Holocene environments in the middle to lower Trent Valley. InFluvial Processes and Environmental Change, Brown AG, Quine TA(eds). John Wiley & Sons: Chichester; 165–178.

Jackson KJ, Wohl E. 2015. Instream wood loads in montane foreststreams of the Colorado Front Range, USA. Geomorphology 234:161–170.

Jeffries R, Darby SE, Sear DA. 2003. The influence of vegetation and or-ganic debris on flood-plain sediment dynamics: case study of a low-order stream in the New Forest, England.Geomorphology 51: 61–80.

John S, Klein A. 2004. Hydrogeomorphic effects of beaver dams onfloodplain morphology: avulsion processes and sediment fluxes inupland valley floors (Spessart, Germany). Quaternaire 15: 219–231.

Jones JB, Smock LA. 1991. Transport and retention of particulate or-ganic matter in two low-gradient headwater streams. Journal of theNorth American Benthological Society 10: 115–126.

Keeton WS, Kraft CE, Warren DR. 2007. Mature and old-growth ripar-ian forests: structure, dynamics, and effects on Adirondack streamhabitats. Ecological Applications 17: 852–868.

Keller EA, MacDonald A, Tally T, Merrit NJ. 1995. Effects of LargeOrganic Debris on Channel Morphology and Sediment Storage inSelected Tributaries of Redwood Creek, Northwestern California,US Geological Survey Professional Paper 1454-P. US GeologicalSurvey: Reston, VA; 29 pp.

Keller EA, Swanson FJ. 1979. Effects of large organic material on chan-nel form and fluvial processes. Earth Surface Processes 4: 361–380.

Keller EA, Tally T. 1979. Effects of large organic debris on channel formand fluvial processes in the coastal redwood environment. In Adjust-ments of the Fluvial System, Rhodes DD, Williams GP (eds). Kendall/Hunt Publishing Co.: Dubuque, IA; 169–197.

Klaar MJ, Hill DF, Maddock I, Milner AM. 2011. Interactions betweeninstream wood and hydrogeomorphic development within recently


deglaciated streams in Glacier Bay National Park, Alaska. Geomor-phology 130: 208–220.

Klein R, Sonnevil R, Short D. 1987. Effects of woody debris removal onsediment storage in a northwest California stream. In Erosion andSedimentation in the Pacific Rim, Beschta RL, Blinn T, Grant GE,Swanson FJ, Ice GG (eds), IAHS Publication no. 165. IAHS Press:Wallingford; 403–405.

Kochel RC, Ritter DF, Miller J. 1987. Role of tree dams in the construc-tion of pseudo-terraces and variable geomorphic response to floodsin Little River valley, Virginia. Geology 15: 718–721.

Kramer N, Wohl EE, Harry DL. 2012. Using ground penetrating radar to‘unearth’ buried beaver dams. Geology 40: 43–46.

Lancaster ST, Hayes SK, Grant GE. 2003. Effects of wood on debris flowrunout in small mountain watersheds. Water Resources Research 39:1168. DOI:10.1029/2001WR001227.

Levine R, Meyer GA. 2014. Beaver dams and channel sediment dy-namics on Odell Creek, Centennial Valley, Montana, USA. Geomor-phology 205: 51–64.

Lisle TE. 1995. Effects of coarse woody debris and its removal on a chan-nel affected by the 1980 eruption of Mount St. Helens, Washington.Water Resources Research 31: 1797–1808.

MacFarlane WA, Wohl E. 2003. Influence of step composition on stepgeometry and flow resistance in step-pool streams of the WashingtonCascades. Water Resources Research 39: 1037. DOI:10.1029/2001WR001238.

Mackay D. 1991. Working the Kauri. Random Century: Auckland.Magilligan FJ, Nislow KH, Fisher GB, Wright J, Mackey G, Laser M.2008. The geomorphic function and characteristics of large woodydebris in low gradient rivers, coastal Maine, USA. Geomorphology97: 467–482.

Mao L, Andreoli A, Comiti F, Lenzi MA. 2008. Geomorphic effects oflarge wood jams on a sub-Antarctic mountain stream. River Researchand Applications 24: 249–266.

Mao L, Andreoli A, Iroumé A, Comiti F, Lenzi MA. 2013. Dynamics andmanagement alternatives of in-channel large wood in mountain ba-sins of the southern Andes. Bosque 34: 319–330.

Marcus WA, Marston RA, Colyard CR, Gray RD. 2002. Mapping thespatial and temporal distributions of woody debris in streams ofthe Greater Yellowstone Ecosystem, USA. Geomorphology 44:323–335.

Marston RA. 1994. River entrenchment in small mountain valleys of thewestern USA: influence of beaver, grazing and clearcut logging.Revue de Geographie de Lyon 69: 11–15.

Marston RA, Fritz DE, Nordberg V. 1997. The impact of debris torrentson substrates of mountain streams. Geomorphologie: Relief,Processus, Environnement 1: 21–32.

Martín-Vide JP, Amarilla M, Zárate FJ. 2014. Collapse of the PilcomayoRiver. Geomorphology 205: 155–163.

Massong TM, Montgomery DR. 2000. Influence of sediment supply,lithology, and wood debris on the distribution of bedrock andalluvial channels. Geological Society of America Bulletin 112:591–599.

May C 2007. Sediment and wood routing in steep headwater streams:an overview of geomorphic processes and their topographic signa-tures. Forest Science 53: 119–130.

May CL, Gresswell RE. 2003a. Large wood recruitment and redistribu-tion in headwater streams in the southern Oregon Coast Range, USA.Canadian Journal of Forest Research 3: 1352–1362.

May CL, Gresswell RE. 2003b. Processes and rates of sediment andwood accumulation in headwater streams of the Oregon CoastRange, USA. Earth Surface Processes and Landforms 28: 409–424.

Mazzorana B, Hübl J, Zischg A, Largiader A. 2011. Modelling woodymaterial transport and deposition in alpine rivers. Natural Hazards56: 425–449.

Meentemeyer RK, Butler DR. 1999. Hydrogeomorphic effects of beaverdams in Glacier National Park, Montana. Physical Geography 20:436–446.

Megahan WF. 1982. Channel sediment storage behind obstructions inforested drainage basins draining the granitic bedrock of the IdahoBatholith. In Sediment Budgets and Routing in Forested DrainageBasins, Swanson FJ, Janda RJ, Dunne T, Swanston DN (eds), USDAForest Service General Technical Report PNW-141. USDA ForestService: Portland, OR; 114–121.



Mikus P, Wyzga B, Kaczka RJ, Walusiak E, Zawiejska J. 2013. Islands ina European mountain river: linkages with large wood deposition,flood flows and plant diversity. Geomorphology 202: 115–127.

Millington CE, Sear DA. 2007. Impacts of river restoration on small-wood dynamics in a low-gradient headwater stream. Earth SurfaceProcesses and Landforms 32: 1204–1218.

Mills EA. 1913. In Beaver World. Houghton Mifflin: Boston, MA.Montgomery DR, Abbe TB. 2006. Influence of logjam-formed hardpoints on the formation of valley-bottom landforms in an old-growth forest valley, Queets River, Washington, USA. QuaternaryResearch 65: 147–155.

Montgomery DR, Abbe TB, Peterson NP, Buffington JM, Schmidt KM,Stock JD. 1996. Distribution of bedrock and alluvial channels inforested mountain drainage basins. Nature 381: 587–589.

Montgomery DR, Collins BD, Buffington JM, Abbe TB. 2003a. Geomor-phic effects of wood in rivers. In The Ecology and Management ofWood in World Rivers, Gregory SV, Boyer KL, Gurnell AM (eds).American Fisheries Society: Bethesda, MD; 21–47.

Montgomery DR, Massong TM, Hawley SCS. 2003b. Influence ofdebris flows and log jams on the location of pools and alluvialchannel reaches, Oregon Coast Range. Geological Society ofAmerica Bulletin 115: 78–88.

Morgan LH. 1868. The American Beaver and His Works. J.B. Lippincott& Co.: Philadelphia, PA.

Naiman RJ, Melillo JM. 1984. Nitrogen budget of a subarctic streamaltered by beaver (Castor canadensis). Oecologia 62: 150–155.

Naiman RJ, Melillo JM, Hobbie JE. 1986. Ecosystem alteration ofboreal forest streams by beaver (Castor canadensis). Ecology 67:1254–1269.

Naiman RJ, Johnston CA, Kelley JC. 1988. Alteration of North Americanstreams by beaver. BioScience 38: 753–762.

Naiman RJ, Pinay G, Johnston CA, Pastor J. 1994. Beaver influences onthe long-term biogeochemical characteristics of boreal forestdrainage networks. Ecology 75: 905–921.

Naiman RJ, Sedell JR. 1979. Benthic organic matter as a function ofstream order in Oregon. Arch. Hydrobiol. 87: 404–422.

Nakamura F, Swanson FJ. 1993. Effects of coarse woody debris onmorphology and sediment storage of a mountain stream system inwestern Oregon. Earth Surface Processes and Landforms 18: 43–61.

Nanson GC, Barbetti M, Taylor G. 1995. River stabilization due tochanging climate and vegetation during the late Quaternary inwestern Tasmania, Australia. Geomorphology 13: 145–158.

O’Connor JE, Jones MA, Haluska TL. 2003. Flood plain and channeldynamics of the Quinault and Queets Rivers, Washington, USA.Geomorphology 51: 31–59.

Oswald EB, Wohl E. 2008. Wood-mediated geomorphic effects of ajökulhlaup in the Wind River Mountains, Wyoming. Geomorphology100: 549–562.

Patterson LJ, Muhammad Z, Bentley SJ, et al. 2003. 210Pb and 137Csgeochronology of the Lake Fausse Pointe region of the lower Atchaf-alaya Basin. Gulf Coast Association of Geological Societies Transac-tions 53: 668–675.

Phillips JD. 2012. Log-jams and avulsions in the San Antonio Riverdelta, Texas. Earth Surface Processes and Landforms 37: 936–950.

Piégay H 2003. Dynamics of wood in large rivers. In The Ecology andManagement of Wood in World Rivers, Gregory SV, Boyer KL,Gurnell AM (eds). American Fisheries Society: Bethesda, MD;109–133.

Piégay H, Thevenet A, Kondolf GM, Landon N. 2000. Physical andhuman factors influencing potential fish habitat distribution along amountain river, France. Geografiska Annaler 82A: 121–136.

Pollock MM, Beechie TJ, Jordan CE. 2007. Geomorphic changesupstream of beaver dams in Bridge Creek, an incised stream channelin the interior Columbia River basin, eastern Oregon. Earth SurfaceProcesses and Landforms 32: 1174–1185.

Pollock MM, Beechie TJ, Wheaton JM, Jordan CE, Bouwes N, Weber N,Volk C. 2014. Using beaver dams to restore incised stream ecosys-tems. BioScience 64: 279–290.

Pollock MM, Heim M, Werner D. 2003. Hydrologic and geomorphiceffects of beaver dams and their influence on fishes. In TheEcology and Management of Wood in World Rivers, Gregory SV,Boyer KL, Gurnell AM (eds). American Fisheries Society: Bethesda,MD; 213–233.


Polvi LE, Wohl E. 2013. Biotic drivers of stream planform: implicationsfor understanding the past and restoring the future. BioScience 63:439–452.

Raikow DF, Grubbs SA, Cummins KW. 1995. Debris dam dynamicsand coarse particulate organic matter retention in an AppalachianMountain stream. Journal of the North American BenthologicalSociety 14: 535–546.

R Core Team. 2014. R: A Language and Environment for StatisticalComputing. R Foundation for Statistical Computing: Vienna.

Richmond AD. 1994. Characteristics and Function of Large WoodyDebris in Mountain Streams of Northern Colorado, MS Thesis,Colorado State University, Fort Collins, CO.

Ruiz-Villanueva V, Bodoque JM, Diez-Herrero A, Bladé E. 2014. Largewood transport as significant influence on flood risk in a mountainvillage. Natural Hazards 74(2): 967–987. DOI:10.1007/s11069-014-122-4.

Russell HAJ, Cummings DI, Ponomarenko D. 2009. GeomorphicImpact of a Beaver Dam Outbreak Flood, Gatineau Hills, Quebec:A Pictorial Record. Geological Survey of Canada: Ottawa; 71 pp.

Ryan SE, Bishop EL, Daniels JM. 2014. Influence of large wood onchannel morphology and sediment storage in headwater mountainstreams, Fraser Experimental Forest, Colorado. Geomorphology217: 73–88.

Sawyer AH, Bayani Cardenas M, Buttles J 2012. Hyporheic temperaturedynamics and heat exchange near channel-spanning logs. WaterResources Research 48 W01529.

Schama, S. 1995. Landscape and Memory. Alfred A. Knopf: New York.Scheffer PM. 1938. The beaver as an upstream engineer. Soil Conserva-tion 3: 178–181.

Sear DA, Millington CE, Kitts DR, Jeffries R. 2010. Logjam controls onchannel: floodplain interactions in wooded catchments and their rolein the formation of multi-channel patterns. Geomorphology 116:305–319.

Scott DN, Montgomery DR, Wohl EE. 2014. Log step and clast interac-tions in mountain streams in the central Cascade Range of Washing-ton State, USA. Geomorphology 216: 180–186.

Shields FD, Smith RH. 1991. Large woody debris effects on channelfriction factor. In Proceedings of the 1991 National Conference onHydraulic Engineering, Shane RM (ed). American Society of CivilEngineers: New York; 757–762.

Shields FD, Morin N, Cooper CM. 2004. Large woody debris structuresfor sand-bed channels. Journal of Hydraulic Engineering 130:208–217.

Skalak K, Pizzuto J. 2010. The distribution and residence time ofsuspended sediment stored within the channel margins of agravel-bed bedrock river. Earth Surface Processes and Landforms35: 435–446.

Smith RD, Sidle RC, Porter PE. 1993. Effects on bedload transport ofexperimental removal of woody debris from a forest gravel-bedstream. Earth Surface Processes and Landforms 18: 455–468.

Tank JL, Rosi-Marshall EJ, Griffiths NA, Entrekin SA, Stephen ML. 2010.A review of allochthonous organic matter dynamics and metabolismin streams. Journal of the North American Benthological Society 29:118–146.

Thompson DM. 1995. The effects of large organic debris on sedimentprocesses and stream morphology in Vermont. Geomorphology 11:235–244.

Triska FJ. 1984. Role of wood debris in modifying channel geomorphol-ogy and riparian areas of a large lowland river under pristine condi-tions: a historical case study. Verhandlung Internationale VereinigungLimnologie 22: 1876–1892.

Umazano AM, Melchor RN, Bedatou E, Bellosi ES, Krause JM. 2014.Fluvial response to sudden input of pyroclastic sediments duringthe 2008–2009 eruption of the Chaitén Volcano (Chile): the role oflogjams. Journal of South American Earth Sciences 54: 140–157.

Wagenmakers E-J, Farrell S. 2004. AIC model selection using Akaikeweights. Psychonomic Bulletin and Review 11: 192–196.

Wallace JB, Webster JR, Meyer JL. 1995. Influence of log additions onphysical and biotic characteristics of a mountain stream. CanadianJournal of Fisheries and Aquatic Sciences 52: 2120–2137.

Wallerstein NP. 2003. Dynamic model for constriction scour caused bylarge woody debris. Earth Surface Processes and Landforms 28:49–68.



Wallerstein NP, Thorne CR. 2004. Influence of large woody debris onmorphological evolution of incised, sand-bed channels. Geomor-phology 57: 53–73.

Ward GM, Aumen NG. 1986. Woody debris as a source of fineparticulate organic matter in coniferous forest stream ecosystems.Canadian Journal of Fisheries and Aquatic Sciences 43: 1635–1642.

Webb AA, Dragovich D. 2004. Episodic sediment pulses generated byforested flood plain stripping: Bruces Creek, Nadgee State Forest,southeastern Australia. In Sediment Transfer Through the Fluvial Sys-tem, Golosov V, Belyaev V, Walling DE (eds), IAHS Publication 288.IAHS Press: Wallingford; 13–20.

Webb AA, Erskine WD. 2003. Distribution, recruitment, and geomor-phic significance of large woody debris in an alluvial forest stream:Tonghi Creek, southeastern Australia. Geomorphology 51: 109–126.

Webster JR, Benfield EF, Ehrman TP, Schaeffer MA, Tank JL, Hutchens JJ,D’Angelo DJ. 1999. What happens to allochthonous material thatfalls into streams? A synthesis of new and published information fromCoweeta. Freshwater Biology 41: 687–705.

Webster JR, Stanford JA, Chaffin JL. 2002. Large wood jam in a fourth-order Rocky Mountain stream. Verhandlungen InternationaleVereinigung Limnologie 28: 1–4.

Welber M, Bertoldi W, Tubino M. 2013. Wood dispersal in braidedstreams: results from physical modeling. Water Resources Research49: 7388–7400.

Westbrook CJ, Cooper DJ, Baker BW. 2011. Beaver assisted river valleyformation. River Research and Applications 27: 247–256.

Wohl E 2011. Threshold-induced complex behavior of wood inmountain streams. Geology 39: 587–590.


Wohl E 2013. Floodplains and wood. Earth-Science Reviews 123:194–212.

Wohl E 2014. A legacy of absence: wood removal in US rivers. Progressin Physical Geography 38: 637–663.

Wohl E, Beckman ND. 2014. Leaky rivers: implications of the loss oflongitudinal fluvial disconnectivity in headwater streams. Geomor-phology 205: 27–35.

Wohl E, Jaeger K. 2009. A conceptual model for the longitudinaldistribution of wood in mountain streams. Earth Surface Processesand Landforms 34: 329–344.

Wohl E, Madsen S, MacDonald L. 1997. Characteristics of log and clastbed-steps in step-pool streams of northwestern Montana, USA.Geomorphology 20: 1–10.

Wohl E, Ogden FL, Goode J. 2009. Episodic wood loading in a moun-tainous neotropical watershed. Geomorphology 111: 149–159.

Yarnell SM, Mount JF, Larsen EW. 2006. The influence of relativesediment supply on riverine habitat heterogeneity. Geomorphology80: 310–324.

Yochum SE, Bledsoe BP, David GCL, Wohl E. 2012. Velocity predic-tion in high gradient channels. Journal of Hydrology 424–425:84–98.

Zelt RB. 2002. Channel Characteristics and Large Organic Debris inAdjacent Burned and Unburned Watersheds a Decade after Wildfire,Park County, Wyoming, Unpublished MS Thesis. Colorado StateUniversity, Fort Collins, CO.

Zelt RB, Wohl EE. 2004. Channel and woody debris characteristics inadjacent burned and unburned watersheds a decade after wildfire,Park County, Wyoming. Geomorphology 57: 217–233.


wood and sediment storage and dynamics in river corridors · wood and sediment storage and dynamics...

Documents