L,:J

1061

Stream-bed scour, egg burial depths, and theinfluence of salmonid spawning on bed surfacemobility and embryo survivalDavid R. Montgomery, John M. Buffington, N. Phil Peterson,David Schuett-Hames, and Thomas P. Quinn

Abstract: Bed scour, egg pocket depths , and alteration of stream-bed suraces by spawning chum salmon (Onchorhynchusketa) were measured in two Pacific Northwest gravel-bedded streams. Close correspondence between egg bural depths andscour depths durg the incubation period suggests an adaptation to tyical depths of bed scour and indicates that even minorincreases in the depth of scour could significantly reduce embryo surival. Where egg bural depths are known, expressingscour depth in terms of bed-load transport rate provides a means for predicting embryo mortality resulting from changes inwatershed processes that alter shear stress or sediment supply. Stream-bed alteration caused by mass spawng also mayinfluence embryo surival. Theoretical calculations indicate that spawning-related bed surface coarsening, sorting, and formdrag reduce grain mobility and lessen the probability of stream-bed scour and excavation of buried salmon embryos. Thispotential feedback between salmon spawng and bed mobility implies that it could become increasingly difficult to reversedeclines in mass-spawning populations because decreased spawning activity would increase the potential for bed scourfavorig higher embryo mortality. Furher analysis of this effect is warranted, however, as the degree to whichspawnig-related bed loosening counteracts reduced grain mobility caused by surace coarsenig, sorting, and redd formdrag remains uncertain.

Resume: Des chercheurs ont mesure l'affouilement du lit , la profondeur de la ponte et l'alteration de la surace du lit degravier de deux cours d' eau causee par Ie saumon keta (Onchorhyncus keta) sur la cote nord-ouest du Pacifique. L' etroitrapprochement entre la profondeur d' enfouissement des oeufs et celle de l'affouillement pendant la periode d' incubationpourait pointer vers une adaptation it des profondeurs caracreristiques d' affouilement et pourait signifier que meme unleger approfondissement de l'affouilement pourait nuire sensiblement it la surie des embryons. Lorsqu on connait laprofondeur d' enfouissement des oeufs , l' expression de la profondeur d' affouillement en termes de charge de fond constitueun moyen de prevoir la mortalite des embryons attbuable it des changements dans les conditions dynamiques du bassinhydrographique it l'origine d' une modification de la force de cisailement exercee par l'eau ou de l'apport en sediments.Alteration de la surace du lit de gravier causee par l' activite de fraye en masse inuence peut-etre la surie des embryons.Des calculs theoriques indiquent que la rugosite du lit, que Ie tr des mareriaux it sa surface ainsi que la microtopographie desnids, attbuables it l' activite de fraye des saumons, devraient diminuer la mobilite des grains et abaisser la probabilite d'affouillement du lit ainsi que de l' excavation des embryons de saumon enfouis. L' existence possible d'une telle boucle deretroaction entre la fraye du saumon et la mobilite du lit signfie qu il pourait etre de plus en plus diffcile de renverser latendance chez des populations en declin qui fraient en masse puisque toute diminution de l' activite de fraye devraitcorrespondre it une hausse de la possibilite pour l'affouillement des nids , ce qui favorise une hausse du taux de mortalitechez les embryons. L' analyse plus poussee de cet effet paraitjustifiee puisqu on ignore dans quelle mesure l'ameublissementdu lit associe it la fraye emaye la perte de mobilite des grains attbuable au durcissement de la surace du lit, au tri et it latrainee causee par la forme du nid.(Traduit par la Redaction J

Introduction of dam constrction, overfishing, habitat degradation, and

other factors, but it has proven extremely difficult to isolatespecific influences, in part because of difficulties in linknghabitat change to population-level response (Holtby and Scriv-ener 1989; Bisson et aI. 1992). We hypothesize that an impor-tant factor in salmon population biology is the active role thatsalmon play in the creation and maintenance oftheir own habi-tat. Beaver dams are probably the most widely recognized

Dramatic declines of Pacific salmon (Oncorhynchus spp.) inthe northwestern United States led to the application of theEndangered Species Act to certain populations (Federal Reg-ister 1991 , 1992) and concern that many other populations areeither in jeopardy or already extinct (Nehlsen et aI. 1991). De-clines in salmon populations reflect the combined influences

Received November 11 , 1994. Accepted November 20 , 1995.Jl2628

R. Montgomery and J.M. Buffngton. Deparent of Geological Sciences , University of Washington, Seattle, WAP. Peterson. Simpson Timber Company, P.O. Box 460 , Shelton, WA 98584 , U.S.A

D. Schuett-Hames. Northwest Indian Fisheries Commission, 6730 Marin Way East, Olympia, W A 98506 , U.P. Quinn. School of Fisheries , University of Washington, Seattle, WA 98195 , U.

98195 , U.S.A

Can. 1. Fish. Aquat. Sci. 53: 1061-1070 (1996). 1996 NRC Canada

1062

Fig. 1. Annual hydrographs of mean monthly flows for KennedyCreek (1960-1971) and Montana Creek (1965-1987).

, "", "

O--Q..

,/ " ..

Kennedy Creek

.. ..

0- .. Montana Creek

:c E

o C)

::

c.rcr (.CD .::0

January June

10 11 12December

example of animal-habitat interactions (e. , Ives 1942; White1979; Naiman et aI. 1988), but burowing activity (e. , Grin-nell1923; Cox 1987; Black and Montgomery 1991) and graz-ing (e. , McNaughton et aI. 1988) provide additionalexamples. It is well known that spawning salmonids constrctredds for their eggs and that in the process they sort and coars-en stream-bed gravels (e. , Everest et aI. 1987; Chapman1988; Young et aI. 1989; Kondolf et aI. 1993) and influencethe invertebrate community (Field-Dodgson 1987), but neitherfisheries biologists nor geomorphologists have explored howspawning activity influences bed mobility and reproductivesuccess.

Anadromous salmon are spawned in fresh water, migrate tosea after a varable period of freshwater residence , spend sev-eral years at sea, then retu to their natal stream and spawn.Each female constrcts a redd consisting of several egg pock-ets dug in the channel bed by turning on her side and rapidlyundulating her tail and body. This action sweeps finer particlesinto the water colum where they are transported downstream(e. , Kondolf et aI. 1993). After eggs are released and fertil-ized, the female buries them while simultaneously excavatinganother egg pocket immediately upstream (Groot and Mar-golis 1991). So long as they are alive, females defend thesenests from encroachment, but high spawning densities mayresult in superiposition of nests, which is widely accepted asa major cause of density-dependent mortality (Hunter 1959;McNeil 1964). In the absence of substantial superimpositionmost embryo mortality is thought to occur from poor watercirculation associated with the filling of intergravel porespaces by fine sediment (Tappel and Bjorn 1983; Chapman1988; Lisle and Lewis 1992) and scour durng high-flowevents (McNeil 1964; Holtby and Healey 1986; Lisle 1989).Although high salmon densities can reduce surival, we reportevidence that mass spawning activity alters channel morphol-ogy in ways that may reduce the vulnerability of eggs to scour.We fuher show that nested embryos are quite vulnerable toincreases in the average depth of bed scour.

Study areasThe study sites are low-gradient, gravel-bedded channels usedby spawning chum salmon (Oncorhynchus keta). KennedyCreek is located at the south end ofPuget Sound, Washingtonand Montana Creek is near Juneau, Alaska. The KennedyCreek study reach is a pool-riffle tye channel with a bank-fullwidth and depth of approximately 10.8 and 0.42 m, and areach-average slope of 0.0047. Flow convergence and diver-

Can. J. Fish. Aquat. Sci. Vol. 53 , 1996

gence associated with accumulations of large woody debris(LWD) locally influence the bed morphology. The MontanaCreek study reach is a plane-bed tye channel (Montgomeryand Buffington 1993) with a bank-full width and depth of18.4and 0.72 m, and a reach-average slope of 0.0037. The MontanaCreek reach has little cross-channel topography and a very lowLWD loading (i. , 0.01 pieces/m ). The annual hydro graph ofKennedy Creek is rainfall dominated with the greatest averagedischarges occurng from November through March (Fig. 1),a period coinciding with incubation of salmonid embryos inthe stream-bed gravel. In contrast, summer high flows domi-nate the annual hydrograph of Montana Creek, with a rela-tively uniform high discharge period from May throughOctober (Fig. 1). Chum salmon tyically spawn in MontanaCreek during July in the middle of this high discharge period.

MethodsPaired pebble counts (Wolman 1954) of 100 grains were conductedin each reach in unspawned locations proximal to redds and in tailspilor tailout portions of redds immediately downstream of egg pockets.Redds (spawned sites) were readily discerned by a change in thetextue and patina of the bed surace. Pebble counts were conducted

durng spawnng activity in the fall of 1991 in Kennedy Creek and inthe summer of 1993 in Montaa Creek. A pebble count (Wolman1954) records the intermediate axis length of clasts randomly selecteddurng a traverse across the channel bed. Characterization of the bedsurface material is preferable to bulk samples of the channel substratefor examining bed stability because the size of material in the surfacelayer governs bed mobility.

Channel cross sections and slopes were sureyed using a digitaltheodolite in Kennedy Creek and a hand level and stadia rod inMontana Creek. Redd dimensions and frequencies estimated at eachstudy site characterized alteration of bed topography durng spawningactivity. Repeated topographic mapping of the entire study reachimmediately prior to and after spawnng fuher quantified effects onchanel morphology at Kennedy Creek.

A total of 104 scour chains placed throughout the Kennedy Creeksite gauged the depth of bed scour and fill durg the winter seasonfollowing spawnig. Scour monitors consisted of an anchored chainwith a series of 4 cm diameter perforated plastic balls, a variant onthe sliding bead scour monitors developed by Nawa and Frissell(1993). Scour monitors were distrbuted throughout the study reachboth at redd locations and at sites not used by spawning fish. Eggpocket depths in 40 redds were measured after visual identificationimmediately following spawning (see Peterson and Quin (1996) forfuher details). Field observations of discharge and channel scour atKennedy Creek were made from November 1991 to February 1992during the time of both peak flows and incubation of salmonid em-bryos within stream gravels.

ResultsSpawning occured throughout the entire study reach atKennedy Creek durg fall 1990 , coarsening the bed surfacepartially filling pools, and excavating bar margins. Similarspawning effects on channel morphology were documentedthe following year through repeated topographic sureyingbefore and after spawning activity. Redds formed a distinctivemicrotopography on the channel bed with an amplitude of

10-20 cm and a wavelength of about 2.0 m. These bed formspersisted until the bed surface was reworked durg highwinter flows. At Montana Creek, half of the bed surface hadbeen disturbed by spawning at the time of the survey. Redd

~ 1996 NRC Canada

Montgomery et al. 1063

Table L Median surace grain sizes and sorting coeffcients of paired unspawned and spawned samples.

(mm) OJ ((j)

Sample Unspawned Spawned % difference Unspawned Spawned % difference

Kennedy Creek

+39 1.8+33

Montana Creek

+67 1.0+56 1.5 1.8+73 1.48+61 1.1

Note: aj is the inclusive graphic standard deviation defmed as (!j9S 16)/4 + (!j9S - !js)/6.6 (Folk 1974), where!j is the standardlog base 2 unit of grain size measurement and the subscripts represent percentiles of a given grain size distrbution in !j unts.Lower values of aj indicate better sorting.

Fig. 2. (A) Composite cumulative size distrbutions of bed surfacesediment for non-redd and redd locations sampled on October 281991 , at Kennedy Creek. (B) Composite cumulative sizedistributions of bed surface sediment for non-redd and reddlocations sampled on July 21 and 22 , 1993 , at Montaa Creek.

100

non-reddredd

Q; 40

100

non-redd

redd

Grain Size (mm)

100

100

microtopography was comparable to that observed in KennedyCreek.

Bed-surface grain sizeSpawning activity visibly dtarsened the channel bed in both

Table 2. tests of all possible groupings of spawned(S) and unspawned (U) samples from Kennedy Creek.

Sample grouping

lU/2UlUllSlU/2S2U/IS2U/2S1S/2S

Probability of similarity

0.528000016000002083

Note: Unpaired, two- tailed tests were used assumingunequal and unown varances.

study reaches, as demonstrated by grain size distributions ofunspawned and postspawning samples (Table 1). Spawningcaused a 33-39% increase in median surace grain size Kennedy Creek, and a 56-73% increase in in MontanaCreek. While the coarse end of each composite grain size dis-trbution was similar, the percentage of fine particles on theredd surfaces decreased (Fig. 2). Body undulations during reddconstrction cause fine particles to be suspended and trans-ported downstream (e. , Kondolf et al. 1993), producingcoarser and better sorted gravels. Fine sediment from portionsof the bed surface distubed by spawning in Kennedy Creekcollected downstream on channel margins and in pools. Bedsurface sorting, defined by the inclusive graphic standard de-viation (Folk 1974), aI, increased by 20-48% in KennedyCreek and by 11-53% in Montana Creek as a result of spawn-ing (Table 1).

Statistical similarty among the mean grain sizes ofspawned and unspawned portions of Kennedy Creek wasevaluated through tests (Table 2). Differences among un-spawned samples were not significant at a 0.05 probabilitylevel. The same was tre of posts pawning samples. Howeverdifferences between spawned and unspawned grain size distr-butions were significant (P 05). Composite grain size dis-trbutions indicate that spawning activity locally increased

from 22 to 30 mm in Kennedy Creek (Fig. 2).Similarity among the postspawning and unspawned loca-

tions in Montana Creek also was evaluated using tests(Table 3). While each of the paired samples differed signifi-cantly, one of the postspawning samples (lS) was not differentfrom the unspawned samples of the other paired pebble counts(2U, 3U, and 4U). Except for sample 1 , all spawned sampleswere similar, as were all unspawned samples. These observations

~ 1996 NRC Canada

1064

Table 3. tests of all possible groupings of spawned(S) and unspawned (U) samples from Montaa Creek.

Sample grouping

1 S/2S

1 S/3S

1 S!4S

2S/3S2S/4S3S/4SIU/:2UlU/3UU/4U

2U/3U2U/4U3U/4UIS/lU1 S/2U

IS/3UIS/4U2S/lU2S/2U2S/3U2S/4U3S/lU3S/2U3S/3U3S/4U4S/lU4S/2U4S/3U4S/4U

Probabilty of similarty

016003000988316238015051015682888770000290144196000001

000000000000000000000000000000

- Note: Unpaired, two-tailed tests were used assumingunequal and unown varances. .

are consistent with Kondolf et aI.' s (1993) finding that the bestpredictor of the postspawning concentration of fme sedimentin redds is the prespawning concentration. Composite grainsize distrbutions from Montana Creek showed that in-creased from 15 mm before spawning to 25 mm after spawn-ing (Fig. 2).

In addition to coarsening and sorting bed surace grainsspawning activity also changed bed material packing. Al-though we did not quantify bed packig, we noted that in bothKennedy Creek and Montana Creek, gravels distubed byspawning were more loosely packed than unspawned portionsof the bed. Postspawning gravels were tyically overloose inthe terminology introduced by Church (1978), whereasprespawning gravels were underloose (imbricated) to nor-mally loose. Application of these qualitative terms , howeveris highly subjective.

Bed scour and egg pocket depthsThe peak discharge observed in Kennedy Creek from Novem-ber 1991 to February 1992 slightly exceeded bank-full stage.During this approximately bank-full event, scour depths re-corded by 104 scour chains distrbuted throughout the reachexhibited a roughly exponential distrbution ranging from 0 to60 cm, with a mean of 13.4 cm (Fig. 3A). Fort measured egg-pocket depths ranged from 9.8 to 48.9 cm, with a mean of22.6 cm (Fig. 3B). Comparison of bed scour and egg pocket

Can. J. Fish. Aquat. Sci. Vol. 53 , 1996

Fig. 3. Normalized histographs of (A) scour and (B) egg buraldepths for Kennedy Creek in 1991-1992. Scour depths (n = 104)

were determined from scour chains monitored through the winterseason. Egg bural depths (n = 40) were measured durg spawningactivity.

B 30

20 Depth (em)

Fig. 4. Percent surival versus scour depth for Kennedy Creekderived from Fig. 3B.

100

scour depth (em)

distrbutions revealed that only a small proportion of the chan-

nel bed scoured to egg-pocket depths durng this approxi-mately ban-full event. Sixty-five percent of the scour chainsrecorded scour of less than 10 cm, while less than 5% of theegg pockets were less than 10 cm deep. Distrbutions of scourand egg-pocket bural depths, however, indicated that even aslight increase in scour depths would significantly affect theintegrty of many egg pockets. A doubling of the mean scourdepth to 26.8 cm, for example, would jeopardize the majorityof egg pockets (Fig. 4).

~ 1996 NRC Canada

Montgomery et al.

DiscussionSalmonid reproductive success depends in par on egg buraldepths exceeding the depth of scour durng the incubation pe-riod. Successful reproduction thus requires adjustment to thetyical timing and depth of bed scour. The observed distrbu-tion of egg bural depths below scour depths during a bank-fullflood both suggests such an adjustment and indicates that anincrease in the average depth of bed scour would significantlyaffect salmon populations. A strong correlation between mor-tality and peak seasonal discharge (Holtby and Healey 1986)suggests a direct link to bed scour. Therefore , conceptual mod-els that relate bed scour to discharge and bed-load transportrates provide a framework for predicting embryo mortlitycaused by changes in watershed processes and grain size.

Bed scourThe depth of scour in gravel-bedded channels reflects the rateof bed-load transport, which is a function of boundary shearstress (and thus flow stage) relative to the critical shear stressesof both the surface and subsurface material (e. , Emmett andLeopold 1963; Leopold and Emmett 1984; Carling 1987; Has-san et al. 1991). In armored gravel-bedded streams, mobility ofthe surface layer is followed by a rapid rise in bed-load trans-port (Jackson and Beschta 1982) resulting from exposure andscour of the finer and more easily mobilized subsurface mate-rial. The rate of bed- load transport is also a fuction of sedi-ment supply. Increased sediment loads cause bed suracefining and thus higher transport rates for a given shear stress(e. , Dietrich et al. 1989). Finer bed suraces likely causemore frequent and greater depths of scour.

The depth of scour can be determined from the bed-loadtransport rate as

(1) = QlI(ubPs(1 -

y))

where is the mean depth of scour, Qb is the bed-load trans-port rate per unit channel width Ub is the average bed-loadvelocity, Ps is sediment density, andy is bed porosity (Carling1987). Bed- load transport rates vary substantially betweenchannels and thus primarily control average scour depths.

Numerous expressions describe bed-load transport as anonlinear fuction of shear stress, discharge, velocity, orstream power (see review by Gomez and Church 1989). Oneconceptually simple equation expresses the bed-load transportrate per unit channel width as a function of the difference

between the effective basal shear stress (1: ) and the criticalshear stress (1:

):

(2) Qb = k(1: 1:y.

where is an empirical constant. Together eqs. 1 and 2 providea simple framework for quantifying scour depth and potentialembryo mortality owing to changes in watershed processes.Changes in either the effective or critical shear stresses canalter bed-load flux and should influence the depth of scour andthereby embryo surival. Changes in sediment supply can alter

c (e. , Dietrch et al. 1989), while changes in discharge andLWD loading can affect 1:' (e. , Buffngton 1995). Furher-more, the observed stream-bed modifications caused byspawning chum salmon also have the potential to alter both 1:and 1:c and thus influence bed scour and embryo mortality.

1065

Influence of spawning on stream-bed mobiltySalmonid-induced modifications of grain size, sorting, pack-ing, and bed topography all potentially affect bed surface mo-bility, and thereby the vulnerability of embryos to fluvialexcavation. In gravel-bedded channels of mixed grain sizesthe mobility of the median surface grain size provides an indi-cator of general bed mobility. Many gravel-bedded rivers ex-hibit equal grain mobility at incipient motion (Parker et al.1982), owing in par to surface armoring (Jackson and Beschta1982), friction angle distrbutions (Buffngton et al. 1992), andhiding effects (Einstein 1950), which together cause a thresh-old condition of general bed surace mobility. Consequently,the critical boundary shear stress required to mobilize the sur-face grains of a gravel-bedded channel can be estimated fromShields ' (1936) equation wrtten in terms of the surface

(3) 50 = 1: (Ps - )gd

where 50 and 50 are the dimensional and dimensionlesscritical shear stresses for the median surface grain size , P s andPw are the densities of sediment and water, and g is gravita-tional acceleration. Numerous and disparate values reported inthe literatue complicate determining 50 (see reviews byChurch 1978 and Buffngton 1995). Here we use Parker andKlingeman s (1982) value of 0.035 , as it represents bed suracecharacteristics of mixed grain sizes typical of gravel-beddedstreams and is derived from a study reach with subdued bed-form roughness (Milhous 1973). Regardless of the specific

50 value used, inspection of eq. 3 shows that 50 is linearly

proportional to implying that surace coarsening increasesthe shear stress necessary to initiate general bed mobility.

Incorporating sediment and water densities of 2650 and1000 kg/m3 into eq. 3 predicts that spawning-related bedcoarsening (i. , 22-30 mm) increased the critical shear stressfrom 12.5 to 17.0 Pa in Kennedy Creek. Comparison of theseshear stresses with the bank-full shear stress allows the signifi-cance of this increase in 50 to be estiated. The reach-average

bank-full shear stress calculated from the depth-slope product(1:0 = PwgDS where 1:0 is the total boundary shear stress andDand S are the flow depth and slope, respectively) is approxi- .mately 19.4 Pa. This predicts that spawning-related surfacecoarsening therefore increased the critical shear stress fromabout 64 to 88% of the bank-full shear stress. However, thesetheoretical (i. , calculated) critical shear stresses underesti-mate the effects of bar form and L WD roughness that dissipateshear stress, effectively lowering the basal shear stress avail-able for sediment transport (i. , 1:) and increasing the totalshear stress required for grain mobility (see Buffngton 1995for fuher discussion). Hence, 1:50 for redd surfaces is likely

to exceed 88% of the bank-full shear stress.Similar critical shear stress calculations for Montana Creek

indicate that spawning-related bed coarsening increased from approximately 33 to 54% of the bank-full shear stress.Roughness effects associated with L WD and bar forms areminimal in Montana Creek. Equation 3 thus predicts thatspawning-related bed surface coarsening significantly in-creases the critical shear stress in both Kennedy and Montanacreeks.

Sorting of bed surface material durng spawning likely fu-ther reduces grain mobility. Increased sorting raises frictionangles ( ) between the bed surface and particles moving overthe bed (Fig. 5), causing greater resistance to motion (Miller

~ 1996 NRC Canada

1066

Fig. 5. Diagram ilustrating the increase in frction angles, cD

owing to increased sorting. Grains composing the bed surace areshaded, while particles traversing the bed are unshaded. Removalof fIner grains durg spawng activity produces better sorted bedsuraces that present deeper intergranular pockets and higherfrction angles. Protrsions of parcles equivalent in size to the of the bed surace are also shown. Protrsion is defmed by apartcle s projection

p,

above the local datu, z = 0 , and by itsexposure above upstream grains (Kichner et al. 1990).

and Byrne 1966; Buffmgton et al. 1992). The vared arrange-ment of grain sizes composing a heterogeneous bed suraceresults in a distrbution of frction angles for a given parclesize on that surace (Kiclmer et al. 1990; Buffmgton et aI.1992). Distrbutions of cD for unspawned and spawned valuesof were determined empircally as functions of grain sizeand sorting (O'r) (Buffmgton et al. 1992). In addition to frctionangle, grain mobility depends on projection (P) and exposure(e) (Fig. 5) (Fenton and Abbott 1977). As with cD, distrbutionsof and characterize each paricle size on a bed surace andcan be determined empircally as fuctions of grain size andcD (Kiclmer et al. 1990).

The combined effects of bed surace coarsenig and sortingon grain mobility are examined by calculating critical shearstresses of un spawned and postspawning gravels using Kich-ner et al.' s (1990) theoretical critical shear stress formulation.In this model, the critical shear stress for incipient grain motionis expressed as

(4) = (Ps - Pw) grJ3K2 (tancDcosa - sina) /

f2(z) cf (2z (2p d))2/12 dzp-e

f2(z) tancD)

where is grain size, von Karman s constant K equals 0.407(Wiberg and Smith 1987), is the inclination of the bed, C(",0.45) is the drag coefficient for a sphere at high-particleReynolds numbers (i. , 10 1 OS) (Rouse 1946), is the heightabove the bed .fz) is a velocity profie fuction of the law ofthe wall, and C (""0.2) is the lift coefficient for a sphere(Wiberg and Smith 1987) (see Kiclmer et al. 1990 for fuerdetails). Inspection of eq. 4 shows that 't is inversely propor-

Can. J. Fish. Aquat. Sci. Vol. 53, 1996

Fig. 6. Calculated critical shear stress distrbutions of grainsizes on unpawned and spawned bed suraces in (A) MontanaCreek and (B) Kennedy Creek.

100

if 60

CD 40

Unspawned d50

Spawned d50

100 200 300

100

t 40

Unspawned d50

Spawned d50

100 200 300 400

Critical Shear Stress (Pa)

tional to grain protrsion

(p

e), and directly proportional to

and cD (and thus O'J.Results from ths analysis predict a shift in the 'tcso distrbu-

tions for unpawned and postspawning surfaces (Fig. 6). InMontaa Creek the median value of't cso increases by 73% asa result of surace coarsening and sorting. Moreover, there is athreefold increase in the overall range of stresses required toentrain grain sizes in Montana Creek. In Kennedy Creekthe median critical shear stress for increases by 111 %.

Additionally, there is a severe expansion in the range of criticalshear stresses , with about 10% of the distrbution characterizedby frction angles in excess of 90 , indicating viral immobi-lity of ths percentage of the sizes. Because these calcula-tions employ frction angle and protrsion data derived fromgrains placed on bed suraces, they provide conservative esti-mates ofthe stresses required to mobilize the actual stream-bedsurace; grains composing the bed surace likely wil be char-acteried by lower protrsions and higher friction angles

(Bufgton et al. 1992; Jiang and Haff 1993), resulting in criti-cal shear stresses greater than those estimated here.

Grain mobility and susceptibility to scour may be fuherreduced by redd microtopography that introduces form dragthat decreases bed shear stresses (i. , lowers 't ). Bar-form

~ 1996 NRC Canada

Montgomery et al. 1067

Table 4. Effects of spawng activity on grin mobility: percent change in 50 relative to the unspawned value and percentage of bank-fullshear stress represented by ths change.

Surace coarsenig (eq. 3)

% change in'lC50 % of ban-fu 'I

Surace coarsenig and sortingcombined (eq. 4)a

% change in'lC50 % of ban-full 'I

Redd form drag (eq. 5)

% change in'lc50 % of ban-full 'I

Kennedy CreekMontana Creek

111 17-26-

11-

Evaluated using median critical shear stresses of calculated stress distrbutions (Fig. 6). 't values calculated from eq. 4 may represent instataeous tubulentsweeps and so may not be comparable to the tie-averged values ofunpawned 't 50 and ban-full 't used here (Bufgton et al. 1992).

resistance in gravel-bedded rivers varies with stage and bedmorphology but may comprise 10-75% of the total chanelroughness (Parker and Peterson 1980; Prestegaad 1983; Hey1988). Nelson and Smith (1989) presented a theoretical modelthat describes the shear stress dissipated by flow over regularlyspaced channel-spanning bedforms as

(5) 'l = 't In

g 21(2 A Z

where 'tbf and 'tg are the shear stresses dissipated by bed-formand grain roughness, respectively, C is approxiately equalto 0.21 for fully separated flow (Smith and McLean 1977), and

and A are the bed-form amplitude and wavelength, respec-tively. Shear stress dissipated by redd topography decreasesthe basal shear stress available for sediment transport, effec-tively increasing the total shear stress requied for grain mobil-ity. The additional shear stress necessar to overcome reddform drag and initiate particle motion can be calculated fromeq. 5 , recognizing that at incipient motion 't = 't 50 as defiedby eq. 3.

Using 'tg equal to the postspawng 't 50 values calculatedfrom eq. 3 together with A = 2.0 m = 0. 10-0.20 m, and

= 0. (of posts pawning gravels), eq. 5 predicts that atincipient grain motion, redd topography dissipates 2. 8 Paof channel shear stress in Kennedy Creek and 2. 6 Pa inMontana Creek. In Kennedy Creek, these 'tbf values increasethe critical shear stress of unspawned gravels (12.5 Pa)by17-62% and constitute 11-40% of the ban-full shear stress(19.4 Pa). Similarly in Montana Creek, redd form drag in-creases the unspawned 'tcso value (8. 5 Pa) by 26-89% and rep-resents 8-29% of the bank-full shear stress (26.1 Pa). Ouranalyses indicate that the combined inuence of redd formdrag, surface coarsening, and sorting more than doubles 't 50 in

Kennedy and Montana creeks (Table 4).Spawning-related changes in packig of bed material, how-

ever, may also affect grain mobility. Redd constrction loos-ens the stream bed, likely causing it to be less imbricated (i.have lower frction angles) and thus more easily mobilized(e. , Church 1978;' Reid et al. 1985; Li and Komar 1986).However, loose packing of the surace layer is unely to per-sist for long, as its deep intergranular pockets provide highfrction angle locations that wil trap fie material (e. , Kich-ner et al. 1990) that is normally in motion at flows less thancritical (e. , Jackson and Beschta 1982). These fier grainspartially bury the larger loosely packed ones, liely decreasingtheir mobility by lowering grain protrsion and increasing frc-tion angles (Buffington et al. 1992).

While we observed that redd gravels were tyically moreloosely packed than non-redd suraces , the signficance of bed

loosenig relative to that of surace coarsening and sorting isunown. As we are unaware of any studies that have thor-oughly quantified the influence of packig on grain mobility,we cannot estimate the importance of bed loosening by sal-monids. Nevertheless, the influence of changes in packing onbed surace mobilty would liely counteract the other spawng-related effects to some degree. Although critical shear stresscalculations for Kennedy Creek indicated that postspawningbed mobility, and thus scour, should occur at stages signfi-cantly in excess of bank-full, field observations show that thepostspawng bed surace mobilized durg an approximatelybank-full event. Ths bed mobility and scour at roughly thebank-full stage implies that loosenig of the bed surace offsetsapproximately 32-44% of the increase in critical shear stressindicated by our calculations, implying that the net effect ofspawning is reduced bed mobility, despite loosening of the bedsurace.

Another issue remainng to be investigated is the longevityof salmonid modifications of the channel. Well-sorted bed sur-faces tend to be hydraulically rough and have deep intergranu-lar pockets that readily trap fie material. Salmonid-alteredgravels may progressively decrease in size and become morepoorly sorted in channels with high sedient supply that trans-port fme sediment at low to moderate stages. P .E. Porter (citedby Everest et al. 1987), for example , found that the percentageof fme paricles (0:1 mm) in stream-bed gravels of MontanaCreek retued to prespawning levels withi 1-2 months. Reddtopography, however, may discourage fme sediment deposi-tion above egg pockets by forcing both upstream depositionwith redd pits and downstream suspension over tailspils(e. , Everest et aI. 1987). Nevertheless, persistence of reddforms is unlikely in streams that normally lack such micro-topography as part of their bed morphology.

ImplicationsThe close correspondence between egg bural depths and scourdepths durg tyical annual high flows implies a fmely tuedadaptation to long-term rates of sediment transport and implieshigh salmonid population sensitivity to variations in scourdepths. The greater than 60% embryo surival for a bank-fullflood inerred from Fig. 3 may define a reasonable minimumestimate oflong-term mortality, as ban-full flows represent apredictable mium annual peak discharge. The lower sur-vival rate at higher discharges noted by Holtby and Healey(1986) may simply reflect scourng deeper into the populationof embryos bured in the stream bed. Our data imply thatchanges in processes influencing scour depths could have dra-matic impacts on salmonid population dynamics. Higher bed-load transport rates resulting from increased sediment supply

~ 1996 NRC Canada

1068

and consequent bed surface fining (Dietrch et al. 1989), forexample, should directly increase scour depths. Many humanactivities (e. , logging, road constrction, urbanization) mayincrease bed scour by increasing the discharge and (or) thesediment supply to streams. Although numerous processes.contribute to the ongoing decline of salmonid populationschanges in the depth of bed scour may be a key factor in thedecline of certain species.

Our calculations indicate that spawning-related bed modi-fications may also influence scour and embryo survival bydecreasing bed mobility. Bank-full flows for many rivers haverecurence intervals of roughly 1 or 2 years (Wolman andLeopold 1957; Wolman and Miler 1960; Dur 1973; Wiliams1978). Hence , salmonid modifications that increase the criticalshear stress may help protect embryos from scour during tyi-cal annual to biennial flows. The potential for spawning activ-ity to reduce bed surface mobility may be particularlyimportant at our study sites because of the coincidence ofspawning timing and the highest monthly average discharges.Spawning activity in both channels may decrease bed mobilitydurg periods when embryos are most susceptible to disrup-tion and scour by high flows, implying that spawning alters thechannel in a way that may enhance the probability of offspringsurival. Spawning-related bed modifications wil decrease as

a mass-spawning salmonid population declines, potentially re-sulting in more frequent scour to egg-pocket depths , decreasedembryo survival, and hence fuher population decline. Suchfeedback would make it increasingly diffcult to reverse a de-clining population trend, especially if the decline coincidedwith changes in land use that lead to more frequent or higherdischarges (e. , Harr 1986; Booth 1991) or to bed surfacefining and thus greater bed mobility (e. , Dietrch et al. 1989)and scour. The influence of spawning-related bed modifica-

tions on scour and embryo surival warrants further investiga-tion, however, because of the unquantified effect of bedloosening.

Fishing patterns that decrease fish size (e. , Ricker et al.1978; Ricker 1980) could also increase embryo mortality frombed scour. Smaller fish bur their eggs less deeply, leavingthem more vulnerable to scour durng bed-mobilizing events(e. , van den Berghe and Gross 1984). Consequently, man-agement practices that tend to reduce fish size might increaseembryo mortality and would exacerbate any positive feedbackbetween a declining population and bed mobility.

Examples of biological influences (other than the pervasiveimpact of human activity (Turer et al. 1990)) on physical en-vironmental processes are relatively uncommon, and manyecological models assume that animals passively respond toenvironmental constraints. Our observations, however, sug-gest a potential feedback mechanism between spawning-related habitat modification, bed mobility, and embryo sur-vival. This feedback is likely to be significant only for mass-spawning species using low-gradient gravel-bedded channelsin which bed scour tyically involves a fraction of the fullthickness of the alluvium composing the stream bed. Thismechansm is unikely to significantly affect dispersed-spawngspecies inhabiting steep cobble and boulder-bedded channelsin which frequent scour incorporates most of the availablespawning gravel. In such channels the timing of spawning mayrepresent a fundamental adaptation to channel-bed mobility(Montgomery 1994).

Can. J. Fish. Aquat. Sci. Vol. 53 , 1996

AcknowledgmentsThis research was supported by grants TfW FY92-0 1 0 andFY94-004 from the Washington State Timber-fish-Wildlifeagreement. T. Q. also acknowledges support ofthe H. MasonKeeler endowment and the U.S. forest Service. We thankAne Bikle and Kevin Schmidt for help with fieldwork. Themanuscript was improved by an insightful discussion withMatt Kondolf and by comments from Rick Smith and two

anonymous reviewers.

ReferencesBisson, P. , Quin, T. , Reeves , G. , and Gregory, S.V. 1992.

Best management practices, cumulative effects , and long-termtrends in fish abundance in Pacific Nortwest River Systems.

Watershed management: balancing sustainabilty and environ-mental change. Edited by R.J. Naiman. Springer-Verlag, NewYork. pp. 189-232.

Black, T. , and Montgomery, DR 1991. Sediment transport byburrowing mammals, Mar County, California. Earth Sur. Pro-cesses Landforms , 16: 163-172.

Booth, D.B. 1991. Urbanization and the natual drainage system:impacts, solutions, and prognoses. Nortwest Environ. J.7: 93-118.

Buffmgton, J.M. 1995. Effects of hydraulic roughness and sedimentsupply on surface textues of gravel-bedded rivers. M. S. thesisUniversity of Washington, Seattle, Wash.

Buffmgton, J. , Dietrch, W. , and Kichner, J.W. 1992. Frictionangle measurements on a natually formed gravel streambed: im-plications for critical boundary shear stress. Water Resour. Res.28: 411-425.

Carling, P.A. 1987. Bed stability in gravel streams , with reference tostream regulation and ecology. In River channels: environmentand process. Edited by K.S. Richards. Blackwell , Ltd. , Oxford.pp. 321-347.

Chapman, D.W. 1988. Critical review of varables used to defmeeffects of fines in redds of large salmonids. Trans. Am. Fish. Soc

, -

117: 1-21.Church, M. 1978. Paleohydrological reconstrctions from a Holo-

cene valley fill. In Fluvial sedimentology. Edited by A.D. Miall.Canadian Society of Petroleum Geologists, Calgary, Alta.pp. 743-772.

Cox, G.W. 1987. Soil translocation by pocket gophers in a Mimamoundfield. Oecologia, 72: 207-210.

Dietrch, W. , Kichner, J. , Ikeda, H. , and Iseya, F. 1989. Sedi-ment supply and the development of the coarse surace layer ingravel-bedded rivers. Natue (London), 340: 215-217.

Dur, G.H. 1973. Magntude-frequency analysis and chanel mor-phology. In Fluvial geomorphology. Edited by M. Morisawa.

Publications in geomorphology. State University of New YorkBinghampton, N.Y. pp. 91-121.

Einstein , H.A. 1950. The bed-load fuction for sediment transporta-tion in open channel flows. USDA Tech. Bull. No. 1026.

Emmett, W. , and Leopold, L.B. 1963. Downstream pattern ofriverbed scour and fill. In Proceedings of the Federal InteragencySedimentation Conference. Misc. Publ. U. S. Dep. Agrc. No. 970.pp. 399-409.

Everest, F. , Beschta, R. , Scrivener, J. , Koski, K.V., Sedell, and Cederholm, C.J. 1987. Fine sediment and salmonid pro-

duction: a paradox. In Streamside management: forestr and fish-ery interactions. Edited by O. Salo and T.W. Cundy. Institute ofForest Resources, University of Washington, Seattle, Wash.

pp.

98-142.Federal Register. 1991. 56: 58 619- 58 624. Available from the Of-

fice of the Federal Register, Washington, D.

~ 1996 NRC Canada

Montgomery et al.

Federal Register. 1992. 57:.14653 - 14 663. Available from the Of-fice of the Federal Register, Washington, D.

Fenton, J. , and Abbott, lE. 1977. Initial movement of grains in astream bed: the effects of relative protrsion. Proc. R. Soc. Lon-don Ser. A , 352: 532-537.

Field-Dodgson, M. S. 1987. The effect of salmon redd excavation onstream substrate and benthic community of two salmon spawningstreams in Canterbur, New Zealand. Hydrobiologia, 154: 3-11.

Folk, R.L. 1974. Petrology of sedimentary rocks. Hemphill Publish-ing Co. , Austin, Tex.

Gomez, B. , and Church, M. 1989. An assessment of bed load sedi-ment transport formulae for gravel bed rivers. Water Resour. Res.25: 1161-1186.

Grinnell, J. 1923. The burowing rodents of California as agents ofsoil formation. J. Mammal. 4: 137-149.

Groot, C. , and Margolis , L. 1991. Pacific salmon life histories. Uni-versity of British Columbia Press , Vancouver, B.C.

Harr, R.D. 1986. Effects of clearcutting on rain-on-snow runoff inwestern Oregon: a new look at old studies. Water Resour. Res.22: 1095-1100.

Hassan, M. , Church, M. , and Schick, A.P. 1991. Distance of move-ment of coarse particles in gravel bed streams. Water Resour. Res.27: 503-511.

Hey, R.D. 1988. Bar form resistance in gravel-bed rivers. J. Hydraul.Div. Am. Soc. Civ. Eng. 114: 1498-1508.

Holtby, L.B. , and Healey, M.C. 1986. Selection for adult size infemale coho salmon (Oncorhynchus kisutch). Can. J. Fish. Aquat.Sci. 43: 1946-1959.

Holtby, L.B. , and Scrivener, lC. 1989. Observed and simulated ef-fects of climatic varability, clear-cut logging, and fishing on thenumbers of chum salmon (Oncorhynchus keta) and coho salmon(0. kisutch) retuing to Carnation Creek, British Columbia. Can.Spec. Publ. Fish. Aquat. Sci. No. 105. pp. 62-81.

Hunter, J.G. 1959. Surival and production of pin and chum salmonin a coastal stream. J. Fish. Res. Board Can. 16: 835-886.

Ives, RL. 1942. The beaver-meadow complex. J. Geomorphol.5: 191-203.

Jackson, W. , and Beschta, R.L. 1982. A model of two-phase bed-load transport in an Oregon Coast Range stream. Earh Sur. Pro-cesses Landforms , 7: 517-527.

Jiang, Z. , and Haff, P. 1993. Multiparticle simulation methodsapplied to the micromechanics of bedload transport. Water Re-sour. Res. 29: 399-412.

Kirchner, J. , Dietrch, W. , Iseya, F. , and Ikeda, H. 1990. Thevariability of critical shear stress, friction angle, and grain protr-sion in water-worked sediments. Sedimentology, 37: 647-672.

Kondolf, G. , Sale , M. , and Wolman, M.G. 1993. Modification offluvial gravel size by spawning salmonids. Water Resour. Res.29: 2265-2274.

Leopold, L.B. , and Emmett, W.W. 1984. Bedload movement and itsrelation to scour. In River meandering. Edited by M. Elliot.American Society of Civil Engineers, New York. pp. 640-649.

, Z. , and Komar, P.D. 1986. Laboratory measurements of pivotingangles for applications to selective entrainment of gravel in acurent. Sedimentology, 33: 413-423.

Lisle, T.E. 1989. Sediment transport and resulting deposition inspawnig gravels, north coastal California. Water Resour. Res.25: 1303-1319.

Lisle, T. , and Lewis, J. 1992. Effects of sediment transport onsurival of salmonid embryos in a natual stream: a simulation

approach. Can. J. Fish. Aquat. Sci. 49: 2337-2344.McNaughton, S. , Ruess, RW. , and Seagle, S.W. 1988. Large mamls

and procs dynamcs in Afca ecsystems. BioScience, 38: 794-800.McNeil, W.l 1964. Redd superimposition and egg capacity of pin

salmon spawning beds. J. Fish. Res. Board Can. 21: 1385-1396.Milhous, R.T. 1973. Sediment transport in a gravel-bottomed stream.

Ph.D. thesis , Oregon State University, Corvallis , Oreg.

1069

Miller, R. , and Byre, R.l 1966. The angle of repose for a singlegrain on a fixed rough bed. Sedimentology, 6: 303-314.

Montgomery, D. 1994. Geomorphological influences on salmonidspawning distributions. In Abstracts with programs. Vol. 26.

No. 7. Geological Society of America, Boulder, Colo. A-439.(Abstr.

Montgomery, D. , and Buffngton, lM. 1993. Chanel classifica-tion, prediction of channel response, and assessment of channelcondition. Rep. No. TFW-SHlO-93-002. Washington State De-partent of Natual Resources , Olympia, Wash.

Naiman, R. , Johnston, e.A. , and Kelley, J.C. 1988. Alteration ofNort American streams by beaver. BioScience, 38: 753-763.

Nawa, RK , and Frissell, e.A. 1993. Measurg scour and fill ofgravel streambeds with scour chains and sliding-bead monitors.N. Am. J. Fish. Manage. 13: 634-639.

Nehlsen, W. , Williams, J. , and Lichatowich, J.A. 1991. Pacificsalmon at the crossroads: stocks at risk from Californa, OregonIdaho, and Washigton. Fisheries (Bethesda), 16(2): 4-21.

Nelson, lM. , and Smith, J.D. 1989. Flow in meandering channels

with natual topography. In River meandering. Edited by S. Ikedaand G. Parker. Water Resour. Monogr. No. 12. American Geo-physical Union, Washigton, D.C. pp. 69-102.

Parker, G. , and Klingeman, P.C. 1982. On why gravel bed streamsare paved. Water Resour. Res. 18: 1409-1423.

Parker, G. , and Peterson, A.W. 1980. Bar resistance of gravel-bedstreams. J. Hydraul. Div. Am. Soc. Civil Eng. 106: 1559-1575.

Parker, G. , Klingeman, P. , and McLean, D.G. 1982. Bedload andsize distrbution in paved gravel-bed streams. J. Hydraul. Div.Am. Soc. Civil Eng. 108: 544-571.

Peterson, N. , and Quin, T.P. 1996. Spatial and temporal variationin dissolved oxygen in natual egg pockets of chum salmon On-corhynchus keta (Walbaum), in Kennedy Creek, Washigton. J.Fish BioI. In press.

Prestegaard, KL. 1983. Bar resistance in gravel bed streams at bank-full stage. Water Resour. Res. 19: 472-476.

Reid, 1., Frostick, L.E. , and Layman, J.T. 1985. The incidence andnatue of bedload transport durg flood flows in coarse-grainedalluvial channels. Ear Surf. Processes Landforms , 10: 33-44.

Ricker, W.E. 1980. Changes in the age and size of chum salmon(Oncorhynchus keta). Can. Tech. Rep. Fish. Aquat. Sci. No. 930.

Ricker, W. , Bilon, H. , and lro, KV. 1978. Causes of decreasein size of pink salmon (Oncorhynchus gorbuscha). Tech. Rep.

Fish. Mar. Servo Can. No. 820.Rouse, H. 1946. Elementary mechanics of fluids. John Wiley & Sons

New York. Reissued by Dover Publications Inc. , New York.Shields, I.A. 1936. Anwendung der Aehnlichkeitsmechanik und der

Turbulenzforschung auf die Geschiebebewegug. Mitteilungender Preussischen Versuchsanstalt fu Wasserbau und SchiffbauBerlin. (Translated from German by W.P. Ott and J.e. vanUchelen, USDA Soil Conservation Service Cooperative Labora-tory, California Institute of Technology, Pasadena, Calif.

Smith, J. , and McLean, S.R. 1977. Spatially averaged flow over awavy boundary. J. Geophys. Res. 82: 1735-1747.

Tappel, P. , and Bjorn, T.C. 1983. A new method of relating sizeof spawning gravel to salmonid embryo surival. N. Am. J. Fish.Manage. 3: 123-135.

Turner, B. , Clark, W. , Kates, RW. , Richards, J. , Mathews, and Meyer, W.B. (Editors). 1990. The Earh as transformed

by human action. Cambridge University Press, Cambridge, U.Kvan den Berghe , E. , and Gross, M.R. 1984. Female size and nest

depth in coho salmon (Oncorhynchus kisutch). Can. J. Fish.Aquat. Sci. 41: 204-206.

White, P. S. 1979. Pattern, process, and natual distubance in vege-tation. Bot. Rev. 45: 229-299.

Wiberg, P. , and Smith, J.D. 1987. Calculations of the critical shearstress for motion of uniform and heterogeneous sediment. WaterResour. Res. 23: 1471-1480.

~ 1996 NRC Canada

1070

Williams, G.P. 1978. Bankll discharge of rivers. Water Resour.Res. 14: 1141-1154.

Wolman, M.G. 1954. A method of sampling coarse river bed mate-rial. Trans. Am. Geophys. Union, 35: 951-956.

Wolman, M. , and Leopold, L.B. 1957. River flood plains: someobservations on their formation. U.S. Geol. Sur. Prof. Pap.

No. 282C. pp. 87-107.

Can. J. Fish. Aquat. Sci. Vol. 53, 1996

Wolman, M. , and Miler, J.P. 1960. Magnitude and frequency offorces in geomorphic processes. J. Geol. 68: 54-74.

Young, M. , Hubert, W. , and Wesche, T.A. 1989. Substrate al-teration by spawng brook trout in a southeastern Wyomingstream. Trans. Am. Fish. Soc. 118: 379-385.

~ 1996 NRC Canada