Transactions of the American Fisheries Society 125:768-779. 1996© Copyright by the American Fisheries Society 1996

Spatial Variation in Spawning Habitat of Cutthroat Trout in aSediment-Rich Stream Basin

JAMES P. MAGEE' AND THOMAS E. McMAHON2

Biology Department, Fish and Wildlife ProgramMontana State University, Bozeman, Montana 59717, USA

RUSSELL E THUROWU.S. Forest Service, Intermountain Research Station

316 East Myrtle Street. Boise, Idaho 83702, USA

Abstract.—We examined distribution and habitat characteristics of spawning sites of cutthroattrout Oncorhynchus clarki at various spatial scales to assess effects of sedimentation within a largebasin in Montana. Redd density varied widely across the basin; nearly all (99%) of the 362 reddsobserved occurred in two high-elevation headwater tributaries. Redd density at the reach scalewas positively correlated (r2 = 0.72, P = 0.001) with abundance of spawning gravels. Otherhabitat variables, such as gradient, width, depth, embeddedness, bank stability, and percent riffle,were not significantly correlated to redd density. Taylor Fork redds contained some of the highestproportions of fine sediments (<6.35 mm, mean = 41.6%; <0.85 mm, 17.9%) observed in eggpockets of salmonid redds in the Rocky Mountain region. Cache Creek, a highly disturbed subbasin,had significantly greater proportions of fine sediments smaller than 0.85 mm in redds than theundisturbed Wapiti Creek subbasin. High fine-sediment levels in redds led to very low estimatedembryo survival (mean, 8.5%), but sedimentation did not appear to limit recruitment. Our datasuggest that compensatory juvenile survival and high embryo survival in the small proportion ofredds with good substrate quality may buffer the effects of the high sediment levels in the basin.

Predicting the effects of sedimentation resultingfrom land use activities has been a research pri-ority for fishery biologists for the past 60 years(Chapman 1988). Much of the research has fo-cused on effects of fine sediment on emergencesuccess of salmonid fry (e.g., Chapman 1988;Young et al. 1991; Weaver and Fraley 1993). Intheir review of sediment studies, Everest et al.(1987) concluded that studies of population-leveleffects are needed to examine how sedimentationaffects fish populations within entire drainage ba-sins. However, we found few published studies thathave adopted this approach. Exceptions are studiesby Shepard et al. (1984a), Cederholm and Reid(1987), and Scrivener and Brownlee (1989) thatassessed population effects of sedimentation bycombining data on numbers of recruits, relationsbetween fine sediments and egg-to-fry survival,and amount of fine sediments within redds con-tributed by land use activities.

Geologic characteristics of a drainage basin ex-ert a strong influence on degree of sedimentationand sensitivity to land disturbance (Everest et al.

1 Present address: Montana Fish, Wildlife, and Parks,730 North Montana Avenue, Dillon, Montana 59725,USA.

2 To whom correspondence should be addressed.

1987). Dominant geological formations in therange of cutthroat trout Oncorhynchus clarki inMontana include highly erosive granitic series ofthe Idaho Batholith, cobble-producing Belt series,and various fine silt-producing volcanic and sed-imentary series (Alt and Hyndman 1986). Muchof the research on sediment effects on salmonidsin the Rocky Mountain region has focused on gra-nitic soils and sand-sized particles of the IdahoBatholith (Bjornn et al. 1977; Platts et al. 1989).To evaluate the effects of smaller sediment parti-cles, we sampled streams draining sedimentary ge-ology.

Recent studies indicate that spawning habitat ofstream-dwelling salmonids may be more spatiallyvariable across drainage basins and may play alarger role in determining patterns of distributionof juveniles and adults than previously thought(Beard and Carline 1991; Bozek and Rahel 1991).To properly characterize such high spatial vari-ability in habitat use and document population ef-fects of habitat change, sampling designs that em-ploy a hierarchial series of spatial scales rangingfrom the entire watershed to microhabitats areneeded (Frissell et al. 1986).

The objectives of our study were to (1) char-acterize spawning habitats of cutthroat trout acrossa large sediment-rich drainage basin, (2) examine

768

This file was created by scanning the printed publication.Errors identified by the software have been corrected;

however, some errors may remain.

CUTTHROAT TROUT SPAWNING HABITAT 769

D5

LEGENDTrapsThermographsReach Boundaries

2km

FIGURE I.—Locations of spawning reaches (stippled and coded areas), weirs, and thermographs in Taylor Forkbasin, Montana.

factors influencing redd distribution at differentspatial scales (basin, subbasin, and reach), and (3)evaluate effects of sedimentation on redd substratecomposition and estimated emergence success. Weuse the data to address the question, Is fine sedi-ment limiting recruitment of the cutthroat troutpopulation in the Taylor Fork basin?

Study AreaTaylor Fork is a fourth-order tributary to the

upper Gallatin River in southwest Montana (Figure1). The drainage encompasses 161 km2 and con-tains about 100 stream kilometers. Peak flows oc-cur from snowmelt during late May to mid-June.Three major elevational land types exist in the

drainage. Headwater sections (above 2,300 m) ofthe major third-order tributaries (Cache, Wapiti,and Lightning creeks) meander through alpinemeadows. Midelevation sections (2,100-2,300 m)have steeper gradients, more constrained valleyfloors, and are bordered by coniferous forest. Low-er sections (below 2,100 m) and the main-stemTaylor Fork drain wide valley floors vegetated bysagebrush Artemisia spp. and willow Salix spp.

Taylor Fork drains sedimentary soils predomi-nantly consisting of fine silt (Snyder et al. 1978).This material is highly erosive and produces largequantities of fine sediment; many of the stream-banks in the basin were rated unstable by Snyderet al. (1978). Habitat conditions in the basin range

770 MAGEE ET AL.

TABLE 1.—Genetic status of Taylor Fork cutthroat troutas determined by protein electrophoresis (R. Leary, Uni-versity of Montana, personal communication).

Percent genetic composition of:

Location

Westslope Yellowslonecutthroat Rainbow cutthroat

trout trout trout

Cache CreekUpper Wapiti CreekDeadhorse CreekUpper Taylor Fork0

87578593

11

102

243

55

a Near Lightning Creek.

from pristine to highly altered. Splash dams wereoperated in the late 1800s on the main-stem TaylorFork to transport logs downstream (Snyder et al.1978). Livestock graze allotments in upper Cacheand middle Wapiti creeks, and areas of the CacheCreek drainage have been recently logged. In theCache Creek subbasin, there is a predominance ofbare, unstable banks and the incised channels andencroachment of sagebrush characteristic of heavi-ly grazed riparian zones (Snyder et al. 1978). Incontrast, upper Wapiti Creek, which is similar insize, elevation, and geomorphology to upperCache Creek, has no known history of logging,grazing, or road building.

The basin contains one of the few remainingpopulations of westslope cutthroat trout O. clarkilewisi in the Gallatin River drainage (Liknes 1984).Other fishes inhabiting the basin include mountainwhitefish Prosopium williamsoni, mottled sculpinCottus bairdi, nonnative rainbow trout O. mykiss,brown trout Salmo trutta, and Yellowstone cut-throat trout O. c. bouvieri. Rainbow trout werestocked in Taylor Fork from 1928 to the late 1980s.Yellowstone cutthroat trout were stocked in smallheadwater lakes and subsequently invaded down-stream reaches. As a result, westslope cutthroattrout are introgressed with rainbow trout and Yel-lowstone cutthroat trout to varying degreesthroughout the basin (Table 1). We assume a neg-ligible influence of this introgression on the effectsof sedimentation.

MethodsSpawner and redd distribution.—The entire ba-

sin was surveyed in summer 1991 to identify po-tential spawning sites by the presence of redds,suitable spawning gravel, and newly emerged fry.Potential spawning sites were then stratified into17 reaches (Figure 1), ranging in length from 324to 3,020 m (mean length, 914 m). Reach bound-aries were delineated by changes in gradient, geo-

morphology, or the presence of a tributary junction(Frissell et al. 1986). In 1992 we counted redds ineach reach every 3-4 d from 25 May to 1 July.We counted a redd only if we observed fish spawn-ing; cleaned gravel sites could not be reliably iden-tified as redds because of their small size, lack ofperiphyton, and the high fine-sediment load in thestream. We marked individual redds by placingnumbered stakes on the streambank and mappedprobable egg pocket locations by observing wherefish spawned. Redd density per reach was ex-pressed as number of redds/100 m2 of wettedstream area. Thermographs recorded hourly tem-perature at five locations in the basin (Figure 1).

We also employed weirs in the lower (LittleWapiti Creek), middle (Deadhorse Creek), and up-per (Cache Creek) portions of the basin (Figure 1)to assess when and where spawning occurred.Weirs spanned the stream channel and capturedupstream and downstream migrants. Weirs werechecked twice daily from 19 May to 1 July 1992.Fish were weighed, measured for fork length, andtagged with a visible implant tag behind the lefteye. Species, sex, and spawning condition werealso noted.

Spawning habitat.—Habitat features were mea-sured in 11 of the 17 spawning reaches to evaluatepotential physical factors associated with variationin redd density. We were unable to measure habitatfeatures in the other six spawning reaches as aresult of early ice cover. Sampled reaches repre-sented the full range of observed habitat conditionsand redd densities. We measured habitat accordingto the procedure of Overton et al. (in press), amodification of the systematic habitat-samplingprocedure described by Hankin and Reeves (1988).Proceeding upstream, we described each channelunit (pool, riffle, etc.) and measured its length,mean wetted width, and mean and maximumdepths. The area of potential spawning gravel wasmeasured with a meter stick and expressed as per-cent of total reach surface area. Based on obser-vations of 362 completed redds, we defined po-tential spawning gravel as patches of substrate atleast 0.25 m2 in area with particles 2-35 mm indiameter. The same observer made all estimates ofspawning gravel to reduce bias. At every 10thchannel unit, we measured gradient with a clinom-eter and visually estimated percent bank stabilityand substrate embeddedness, using rating criteriadescribed by Platts et al. (1983). In pool tailoutsand riffles, we measured substrate compositionwith the pebble count technique (Wolman 1954),using 100 data points at each site in streams wider

CUTTHROAT TROUT SPAWNING HABITAT 771

than 1 m and 50 data points in streams narrowerthan 1 m. We classified substrate size as sand-silt(<2 mm in diameter), pea gravel (2-16 mm), grav-el (17-75 mm), rubble (76-150 mm), cobble (151-300 mm), or boulder (>300 mm). We used Spear-man \s rank correlation and simple linear regression(Zar 1984) to assess associations between redddensity and reach habitat features. We were unableto statistically compare habitat features amongsubbasins because only one reach was sampled intwo of the three subbasins.

Redd substrate composition.—We used a McNeilhollow-core sampler (Platts et al. 1983) to exca-vate substrate from 21 redds in upper Cache Creek(five reaches or reach groups) and 15 redds in upperWapiti Creek (three reaches). In Cache Creek, wegrouped north fork (reaches 8-11) and south fork(reaches 13-17) spawning reaches together forease of sampling. Three to seven redds were ran-domly chosen for coring in each of these eightdesignated reaches or reach groups. To quantifysubstrate conditions near time of emergence, wecollected cores during a 2-week period (23 July to6 August) following our first sighting of emergentfry (22 July). We used the redd maps to positioncores over egg pockets. Cores were taken to adepth of 10 cm to mimic egg pocket depths ob-served in cutthroat trout redds (B. May, GallatinNational Forest, personal communication). We la-beled cores as egg pocket samples if we observedeggs or alevins in the core.

Oven-dried core samples were weighed after be-ing mechanically shaken through sieves of 50.8,25.4, 12.4, 9.5, 6.35, 2.36, 0.85, and 0.074 mm(see Magee 1993 for further detail). We collectedan average of 3.6 kg of substrate per core. Asrecommended by Chapman (1988) and Young etal. (1991), we expressed substrate composition bytwo methods: the percentage of fine substratesmaller than a given size (6.35, 2.36, and 0.85 mm)and by a measure of central tendency, the fredleindex F/ (F/ = DgIS0\ Dg is the geometric meanparticle diameter and S0 is a sorting coefficient;see Young et al. 1991). As in previous studies(Grost et al. 1991; Thurow and King 1994), wecalculated F/ by excluding particle sizes greaterthan 50.8 mm to minimize bias. (Particles greaterthan 50.8 mm ranged from 0% by weight in 17redds to 3.3-37.5% in 19 redds.) This procedurealso allowed us to compare our results with otherstudies that used truncated data.

We compared substrate composition amongspawning reaches with a Kruskal-Wallis nonpara-metric analysis of variance and Tukey's multiple

comparison test (Zar 1984). We used rank corre-lation to examine associations between redd den-sity and measures of substrate composition (F/;proportion of particle sizes less than 6.35, 2.36,0.85 mm). Stepwise multiple regression was usedto assess how the proportion of different-sized fineparticles influenced variation in F/. A Mann-Whit-ney IMest was used to compare redd substratecomposition among egg pocket and non-egg pock-et cores and to compare redd substrate compositionamong subbasins having low (Wapiti Creek) andhigh (Cache Creek) land disturbance.

Fry production.—We predicted fry emergencesuccess for each cored redd using the equationdeveloped by Weaver and Fraley (1993) for west-slope cutthroat trout:

% emergence = -0.7512 (arcsine % SP^js)+ 39.67;

% Sp6.35 = percentage of substrate particlessmaller than 6.35 mm.

Total fry production per reach was estimated bycombining data on total redd number, estimatedegg deposition, and average substrate compositionof cored redds. To account for effects of femalesize on egg deposition, we divided the length-fre-quency distribution of females captured in theCache Creek spawning trap into 25-mm intervalsand calculated egg deposition (£) for each size-class using a length-fecundity relation for west-slope cutthroat trout: E = 82.63e° 0007958L, whereL is fork length in millimeters (C. Downs and B.Shepard, Montana State University, personal com-munication). We assumed all mature females hadan equal likelihood of spawning and apportionedegg deposition to redds by the percentage of fe-males within each size-class (25%, 150-174 mm;50%, 175-199 mm; 16%, 200-224 mm; 6%, 225-249 mm; and 3%, >250 mm). Fry production ina reach was estimated by multiplying total esti-mated egg deposition by estimated emergence suc-cess.

ResultsSpawner and Redd Distribution

The majority (271, 94%) of trout trapped duringupstream spawning migration were captured in theCache Creek weir; only 10 trout were captured inDeadhorse Creek and 6 in Little Wapiti Creekweirs. Cutthroat trout accounted for 99.6% of fishcaught in the upper basin weir at Cache Creek;rainbow trout constituted a larger proportion of thecatch at the two lower-elevation weirs (10% in

772 MAGEE ET AL.

TABLE 2.—Number and density (redds/100 m2) of troutredds by stream order in Wapiti (W), Little Wapiti (LW),Deadhorse (D), and Cache (C) creeks.

Reach

WlC8C9CIOCl lC13C14C15C16C17Total or mean

W2W3LW4D5C7CI2Total or mean

C6Grand total or mean

Number ofLength (m) redds

First order2,262

500900843859585656464493

1,0258,587

Second order472

1.258575324833477

3.939Third order

3.02015,585

571

25341413933

39198

1732

13

3924

116

48362

Redddensity

1.7O.I3.33.41.42.01.10.60.63.81.8

2.40.80.10.73.32.81.7

0.51.7

Deadhorse Creek and 70% in Little Wapiti Creek).Mature male cutthroat trout averaged 189 mm (N= 282) and females 191 mm (N = 156) in forklength.

Ninety percent of upstream migrants enteredCache Creek from 19 May to 4 June at mean dailywater temperatures of 7-9°C. Fish were observedon redds in Cache Creek from 25 May to 19 Juneand in Wapiti Creek from 8 to 22 June when meandaily temperatures were near 8°C. Peak spawningin Cache Creek occurred from 3 to 5 June, when77 redds (40% of total) were observed. Peakspawning in Wapiti Creek occurred about 1 weeklater. Water temperatures in the main-stem TaylorFork were similar to those in Cache Creek. Meandaily temperatures in Lightning Creek did notreach 8°C until mid-July. Females remained inspawning tributaries an average of 11 d; malesremained for an average of 9 d.

Most of the 362 redds counted in the Taylor Forkbasin in 1992 were clustered in two areas (Table2). Nearly all redds were observed in the upperreaches of Cache Creek (252, 70%) and WapitiCreek (106, 29%). Despite large numbers of reddsin these areas, no redd superimposition was ob-served. In 1991, no redds or fry were observed in

the main stem of Taylor Fork or in Eldridge Creek,Meadow Creek, Left Fork Creek, lower CacheCreek, middle Wapiti Creek, or Alp Creek. Smallnumbers of newly emerged fry were observed inlower Wapiti Creek, Little Wapiti Creek, andLightning Creek.

Most redds were found in first- and second-ordertributaries, but redd density varied widely, evenamong adjacent reaches (Table 2). Cache Creekreaches 15 and 16 contained three redds each(0.6/100 m2), whereas adjacent reach 17 contained39 redds (3.8/100 m2). Similarly, reach 8 contained1 redd (0.01/100 m2), whereas adjoining reaches7 and 9 contained 39 and 25 redds (3.3/100 m2),respectively.

Spawning HabitatCutthroat trout spawned in reaches of small (1-

3-m-wide) tributaries with low to moderate gra-dients (0.5-3.8%) and suitable spawning gravels(Table 3). Redds were typically not observed atgradients greater than 4%. Streambank stability ofspawning reaches averaged less than 50% stable,and no reaches had streambanks classified as great-er than 75% stable. Substrate embeddedness washigh, averaging about 50%.

Measured habitat features generally varied littleamong reaches that supported redds. Features, in-cluding gradient, width, depth, embeddedness,bank stability, percent riffle, and substrate com-position, were not significantly correlated withredd density (Table 3). Only the proportion of po-tential spawning gravel was highly correlated withredd density (Figure 2; Table 3). Proportion ofpotential spawning gravel varied widely, rangingfrom 3 to 25%, even among nearby reaches. Wefound no association (P = 0.1-0.6) between abun-dance of spawning gravel and reach gradient, pro-portion of pea gravel, or proportion of gravel sub-strate.

Redd CharacteristicsWe found eggs in 11 of the 36 redds cored. We

found no significant differences (Latest, P > 0.47)in percentages of substrate smaller than 6.35, 2.36,and 0.85 mm or in F/ values between cores withand without eggs. We therefore used all 36 coresamples to characterize substrate composition ofredds.

Cutthroat trout redds consisted primarily ofsmall gravels (75% less than 25.4 mm in diameter)and contained a high proportion of particles lessthan 6.35 mm (Figure 3). Forty percent of the sub-strate was smaller than 6.35 mm and 18% was

CUTTHROAT TROUT SPAWNING HABITAT 773

TABLE 3.—Means of habitat features in spawning reaches of Wapiti (W), Deadhorse (D), and Cache (C) creeks.Spearman's rank correlation with redd density (Table 1) shown as rs. Asterisk denotes significance at P < 0.01.

Reachor

corre-lation

W3D5C5C7C8C9CIOCI2C13C14C17rs

Gra-dient Width(%) (m)

1.2 3.10.5 1.30.5 3.20.54.02.53.00.51.03.8

.4

.1

.1

.2

.8

.1

.21.5 1.00.05 -0.38

Percent by area

Depth(cm)

251527196679

12277

-0.36

Embed-dedness3

2.82.62.93.13.23.74.42.82.62.93.20.47

Bankstability5

1.92.53.42.72.22.73.42.12.93.42.50.15

Spawninggravel

14.19.98.5

17.33.2

16.213.918.013.38.5

25.4

0.80*

Riffle

3942343850424644433445

0.25

Silt

17201816232019222416250.14

Peagravel

249

142211141816248

18

0.48

Gravel

5259465844464050423754

0.06

Rubble

87

163

151115104

284

-0.393 Rated on a scale from 1 to 5 for percent of substrate embedded: 1

et al. 1983).b Rated on a scale from 1 to 4 for percent stable banks: 1 = <25%; 2

>75%; 2 = 50-75%; 3 = 25-49%; 4 = 5-24%; 5 = <5% (Plaits

25-49%; 3 = 50-75%; 4 = >75% (Ptatts et al. 1983).

smaller than 0.85 mm. When substrate sampleswere truncated to remove particles larger than 50.5mm, the percentage of substrate smaller than 6.35mm ranged from 26 to 73%, the percentage smallerthan 0.85 mm ranged from 8 to 40%, and Ff valuesranged from 0.48 to 5.14. The F/ was negativelycorrelated with percentage of substrate smallerthan 6.35, 2.36, and 0.85 mm (r, = -0.86 to-0.97, P < 0.001). However, when all three mea-sures of fines were regressed against F/ by multipleregression, only percentage of substrate smallerthan 0.85 mm accounted for a significant propor-tion of the variation in F/ (R2 = 0.75, P = 0.001).

The proportion of substrate smaller than 0.85mm was the only substrate measure that differedsignificantly (Kruskal-Wallis: H = 14.8, P = 0.04)

0 2 4 6 8 10 12 14 16 18 20 22 24Percent spawning gravel

26

FIGURE 2.—Relationship between redd density (num-ber/100 m2) and percentage of the substrate consistingof potential spawning gravel in 11 reaches of TaylorFork, 1992.

among the eight spawning reaches. Other redd sub-strate measures (f / , proportions smaller than 6.35and 2.63 mm) were not significantly different. Co-efficients of variation (SD/mean) in redd substratevariables differed widely both among and withinreaches, being low (less than 15%) in some reaches(e.g., C6 and C8-11) and high (up to 106%) inothers (e.g., W3 and C7). We found no significantassociations (P > 0.16) between redd density andvarious measures of redd substrate composition.

Both the low-disturbance subbasin (WapitiCreek) and the high-disturbance subbasin (CacheCreek) had high percentages of fines smaller than6.35 mm (means, 42.6 and 44.6%). However, thepercentage of substrate smaller than 0.85 mm was

100.0N = 36

0.074 0.85 2.36 6.35 9.5 12.4 25.4 50.8Particle size (mm)

FIGURE 3.—Mean ( + 2 SE) particle size distributionof cutthroat trout redds sampled in Cache and Wapiticreeks, 1992.

774 MAGEE ET AL.

TABLE 4.—Number of redds observed and cored, meanfredle index (F/) of substrate composition of cores (rangein parentheses), estimated egg deposition, predicted meanemergence success, and estimated fry production perspawning reach in Cache (C) and Wapiti (W) creeks.

Number ofredds

Reach orsubbasin

ReachWIW2W3C6C7C8-11C12C13-17

SubbasinWapiiiCacheTotals

Ob-served

5717323939743367

106252358

Cored F,

73543644

1521

36

2.8 (1.4-4.8)1.3 (0.7-1.8)2.3(0.5-5.1)1.0(0.5-1.6)1.6 (0.8-2.7)1.0 (0.7-1.4)2.0(1.3-2.5)1.0(0.5-1.6)

2.3(0.7-5.1)1.3 (0.5-2.7)

1.4

Eggdeposition

20,6066,679

11,56513,92713,92726,53611,90724,174

38,85090,471

129,321

Emer-gencesuc-cess

11.66.87.96.88.37.8

10.57.9

8.29.48.5

Numberof fry

2.392455916940

1.1492,0801,2541,931

3,7637,354

11,117

significantly higher in Cache Creek (21.6%, versus17.1% in Wapiti Creek; U = 2.57, P = 0.01) andthe FJ was significantly lower in Cache Creek (Ta-ble 4; U = 2.32, P = 0.02).

Fry ProductionPredicted emergence success was low, averag-

ing 8.5% for all redds (Table 4). Estimated eggdeposition ranged from 292 to 699 eggs/redd. Ofthe total estimated deposition of 129,321 eggs, itwas estimated that 11,117 fry were produced, av-eraging 24-42 fry/redd. Estimated numbers of fryproduced per redd were similar in Cache and Wa-piti creeks, but because more fish spawned inCache Creek, total estimated fry production therewas about twice that of Wapiti Creek.

DiscussionSpawner and Redd Distribution

Previous studies have documented low-ordertributaries as important spawning habitat for cut-throat trout (Rieman and Apperson 1989). We alsofound most redds (88%) in first- and second-ordertributaries. However, we found high variation inuse of these sites. Nearly all redds (99%) wereobserved in the upper reaches of the high-elevationsubbasins of Cache and Wapiti creeks even thoughthese areas represent a relatively small proportion(15%) of the basin. Substantially fewer redds wereobserved in lower-elevation tributaries in the ba-sin, although these areas contained some spawninggravels and had suitable water temperatures (forexample, Deadhorse and Little Wapiti creeks).

Trout species composition and density variedwith elevation and appeared to contribute to dif-ferences observed in number of redds and spawn-ers within the basin. Segregation by elevation be-tween native cutthroat trout and nonnative rainbowtrout and brown trout is common where the speciesco-occur (Bozek and Hubert 1992). Spawning cut-throat trout used high-elevation tributaries almostexclusively. At lower elevations, cutthroat troutnumbers declined and the proportion of rainbowtrout increased. Ireland (1993) found that juvenileand adult cutthroat trout were abundant (20fish/100 m2) in reaches immediately downstreamof Cache Creek and Wapiti Creek spawning trib-utaries but were rare in the rest of the basin. Theclose association between redd distribution anddensity of juvenile and adults has been reportedby other investigators (Beard and Carline 1991).Conversely, rainbow and brown trout were foundonly in the main-stem Taylor Fork and lower-el-evation tributaries at low densities (2 fish/100 m2).Ireland (1993) attributed the low densities of rain-bow trout and brown trout in the main-stem TaylorFork to poor habitat conditions.

Some higher-elevation subbasins, such as Light-ning Creek, may have been too cold for extensivespawning and fry recruitment. Peak spawning tem-perature of 8°C occurred there 1 month later thanin Cache and Wapiti creeks. We observed newlyemerged fry in Lightning Creek in early Septemberas compared with late July to early August inCache and Wapiti creeks. Such late-emerging frymay have poor overwinter survival (Smith andGriffith 1994).

Cutthroat trout in the Taylor Fork basin appearto conform to a resident form of life history (Lik-nes and Graham 1988; Rieman and Apperson1989), because we captured very few (3%) largespawners (>250 mm) characteristic of fluvial pop-ulations (Shepard et al. 1984b; Rieman and Ap-person 1989). Ireland (1993) found that afterspawning, most cutthroat trout in Cache Creek andWapiti Creek moved relatively short distances(<2,000 m) to summer and winter habitats in sec-ond- and third-order tributaries. Historically,spawners following a fluvial life history wereprobably common in the basin, but negative in-teractions with nonnative trouts, habitat disrup-tion, and angler exploitation in lower portions ofthe basin probably selected against this life historyform (Liknes and Graham 1988).Spawning Habitat

Abundance of potential spawning gravels was akey factor influencing redd density at the reach

CUTTHROAT TROUT SPAWNING HABITAT 775

TABLE 5.—Mean proportion of fines smaller than 6.35 mm and 0.85 mm and the mean fredle index (F/) of eggpockets in salmonid redds from the Rocky Mountain region, USA.

Troutspecies

CutthroatCutthroatCutthroat

Steelhead

BrookBrown

Percent finessmaller than:

Location

MontanaMontanaIdaho

Idaho

WyomingWyoming

Time of sampling

Near emergence

Near emergence

Near emergence

Near spawningNear spawning

6.35 mm

41.627.424.4

15.4

12.115.0

0.85 mm F-t

17.9 2.05.96.5 7.6

3.7 10.5

6.43.0

N

111313

9

3169

Reference

This studyWeaver and Fraley ( 1993)R. Thurow and J. King,

unpublished dataR. Thurow and J. King,

unpublished dataYoung et al. (1989)Grost etal. (1991)

scale (Figure 3). Similar relations between redddensity and availability of spawning gravel werenoted by Cope (1957) for Yellowstone cutthroattrout and by Beard and Carline (1991) for browntrout. The lack of strong association betweenspawning gravel abundance and several reachcharacteristics (gradient, proportion of gravel andpea gravel) suggests that local hydrologic featuresinfluenced spawning gravel availability. Althoughthe distribution of spawning substrate was patchywithin the Taylor Fork drainage, there appeared tobe adequate habitat to support resident spawnersbecause redd superimposition was minimal. Thedistribution of redds apparently was not stronglyinfluenced by the levels of fine sediment in spawn-ing gravel, because we found no negative relationsbetween redd density and various measures of reddsubstrate composition.

Studies of spawning habitat selection by salmo-nids have focused on the microhabitat features ofredds (e.g., Thurow and King 1994). The high vari-ation we observed in redd density across the basinand even among nearby study reaches suggests thatbiotic and abiotic factors operating at larger scalesare also important in describing spawning habitatsuitability (Beard and Carline 1991; Bozek andRahel 1991). Our data suggest that temperature,fish density, and availability of spawning gravelsinteracted to influence redd density across the en-tire basin.

Redd CharacteristicsTaylor Fork redds contained some of the highest

proportions of fine sediments observed in eggpockets of salmonid redds in the Rocky Mountainregion (Table 5). Several factors may account fordifferences among sites, including sampling meth-od, geology, and sediment transport. Other inves-tigators, using freeze core samplers, have docu-mented the unique qualities of egg pockets and

their significantly lower concentrations of fines(Chapman 1988; Young et al. 1989). The lack ofdifferences in our egg pocket and non-egg-pocketsamples may have been influenced by the inabilityof hollow core samplers to preserve the verticalstratification of redd samples (Everest et al. 1980)or by rapid infiltration of fines into redds soon afterconstruction, as Grost et al. (1991) observed. Timeof sampling also may result in redd particle sizedifferences (Grost et al. 1991). We sampled reddsat the time of emergence, whereas some other au-thors sampled egg pockets near the time of spawn-ing (Table 5). Everest et al. (1987) and Lisle andLewis (1992) recommended sampling redds nearthe time of emergence to document actual incu-bation conditions.

We believe that the erosive geology and asso-ciated sediment loading in Taylor Fork contributedto the high percentages of fines in redds comparedwith other sites in the region. The proportion ofsmall fines (<0.85 mm) in our study redds (17.9%)averaged two to six times higher than in reddselsewhere. These small fines are readily removedduring redd excavation (Young et al. 1989; Grostet al. 1991), yet apparently reaccumulate to highlevels before emergence, attesting to the high fine-sediment load in spawning reaches. High propor-tions of fines within the undisturbed Wapiti Creeksubbasin further illustrate the highly erosive natureof the siltstone geology of the basin.

Proportions of small fines in redds were highlyvariable, accounting for a significant proportion ofvariation (R2 = 0.75, P = 0.001) in the F,. Small-scale differences in sediment transport and infil-tration, particularly in streams with high loads offine sediment, commonly cause high variation insmaller particle sizes among redds (Lisle and Lew-is 1992). Everest et al. (1987) and Chapman (1988)emphasized that predicting the effects of fine sed-iments on salmonid embryo survival requires ac-

776 MAGEE ET AL.

curate sampling of egg pockets and greater atten-tion to measuring temporal changes in substrateconditions during incubation. Our findings suggestthat accurate predictions of effects of fines alsorequire knowledge of spatial variability of reddsubstrate conditions and knowledge of the influ-ence of basin geology on substrate size compo-sition (Everest et al. 1987; Lisle and Lewis 1992).

Our data further suggest that land disturbanceexacerbated the high levels of fine sediments inTaylor Fork redds. Previous reports have noted thatsmall fines (<0.85 mm) are the most sensitive in-dicator of the level of disturbance in watersheds(McNeil and Ahnell 1964; Cederholm and Reid1987; Young et al. 1991; see also Scrivener andBrownlee 1989). The high-disturbance CacheCreek subbasin had significantly greater propor-tions of small fines in redds than the relativelyundisturbed Wapiti Creek subbasin. Cederholmand Reid (1987) found that road surface erosionwas the major contributor to higher proportions ofsmall fines in redds in disturbed subbasins of theClearwater River, Washington. We did not quantifythe source of small fines in our study; however,road surface erosion was not responsible for in-creased sedimentation of Cache Creek becausemost surveyed spawning areas were located up-stream from roads. Streambanks in Cache Creekwere degraded (Snyder et al. 1978) and we spec-ulate that increased erosion resulting from live-stock disturbance may be a primary source of el-evated small fines; cattle are common on stream-banks and within the stream channel during in-cubation (Magee 1993).

Is Fine Sediment Limiting Recruitment?High levels of fine sediment in Taylor Fork redds

led to very low predictions of overall embryo sur-vival (mean, 8.5%) and numbers of fry producedper redd (mean, 30). Our estimates of emergencesuccess were made by using proportions of finesediment less than 6.35 mm in diameter, based onthe experiments of Weaver and Fraley (1993) withartificial egg pockets planted in stream substrate.Actual emergence success may have been evenlower than predicted due to high proportions ofsmall fines (<0.85 mm). Small fines are more det-rimental to survival of incubating eggs (Reiser andWhite 1988) and intrude deeper into spawninggravels (Beschta and Jackson 1979) than coarserfines. The proportions of both small (<0.85 mm;mean, 17.9%) and coarse (<6.36 mm; mean,44.3%) fines in Taylor Fork are considerably abovethe levels for these particle sizes (10 and 30%,

respectively) shown to adversely affect salmonidemergence success (McNeil and Ahnell 1964;Shepard et al. 1984a; Cederholm and Reid 1987;Reiser and White 1988).

Previous studies have demonstrated that reducedemergence success from high sedimentation canresult in low juvenile densities and low adult re-cruitment (Cederholm and Reid 1987; Scrivenerand Brownlee 1989). Despite high fine-sedimentlevels and low predicted emergence success of fryfrom Taylor Fork redds, there was little supportingevidence for low recruitment. Juvenile and adultdensities in pools of Cache Creek (43.5 fish/100m2) and Wapiti Creek (32.5/100 m2) are notablyhigher than both the average (9.2/100 m2) andmaximum (26.1/100 m2) density in pools reportedfor seven other westslope cutthroat trout popula-tions in Idaho and Montana (Shepard et al. 1984b;Ireland 1993). Beard and Carline (1991) found ev-idence for recruitment limitation of a residentbrown trout population as reflected by low redddensity, high ratios of mature females to redds,high rates of redd superimposition, low embryosurvival, and low juvenile and adult density. Wepredicted low embryo survival in our study; how-ever we observed high numbers of redds, low reddsuperimposition, low ratio of females to redds, andhigh juvenile and adult density.

Several factors may buffer the effects of highsedimentation of redds. High fry mortality is com-mon in resident salmonids, but low fry densitiesmay result in decreased competition, increasedgrowth, and compensatory survival (McFadden1969). High fry survival is suggested for TaylorFork cutthroat trout. A population estimate of ju-venile and adult cutthroat trout in Cache Creek infall 1994 (our unpublished data; N = 3,726, 95%confidence interval = 2,894-4,588) was 50% ofthe predicted numbers of fry produced in CacheCreek in 1992 (Table 3; N = 7,354). Typical annualmortality rates for cutthroat trout are estimated at95% for fry and 30-50% for older age-classes(Rieman and Apperson 1989).

Variation in substrate quality among redds (F/= 0.48-5.14) also could lead to underestimates offry recruitment. With a large number of redds, thesmall proportion with high substrate quality mayproduce enough fry to compensate for low emer-gence success in remaining redds (Lisle and Lewis1992). Reductions of fines during redd buildingmay counteract the effects of high sediment loads(Everest et al. 1987; Chapman 1988), but this isprobably not an important factor for Taylor Forkcutthroat trout because of the small size of spawn-

CUTTHROAT TROUT SPAWNING HABITAT 777

ers and the very high proportion of fines. It ispossible that the population has evolved other re-productive adaptations to compensate for the nat-urally high sediment load in the basin (e.g., dif-ferences in egg size, fecundity, egg depth; Holtbyand Healey 1986). For example, van den Bergheand Gross (1989) reported that small salmonideggs survive better than large eggs in highly sedi-mented substrates. Small, resident cutthroat troutin Taylor Fork therefore may have adapted to theinherently high fines and have better emergencesuccess than the values we applied from Weaverand Fraley (1993). Additional measurements ofegg deposition, fecundity, and emergence successare needed to evaluate compensatory mechanisms.

Our results support the theory that resident sal-monid populations are typically not limited by re-duced spawning success and that recruitment isprobably limited by available rearing habitat, ex-cept when seeding is very low (Everest et al.1987). However, the high levels of fine sedimentsin Taylor Fork redds, among the highest reportedin the literature for salmonids (Table 5; Chapman1988; Scrivener and Brownlee 1989; Beard andCarline 1991), suggest that fine sediments may beapproaching the level that would limit recruitment.The loss of larger, more fecund fluvial cutthroattrout together with degradation of remaining hab-itats may threaten the persistence of cutthroat troutin the basin. Populations with reduced genetic di-versity and those in degraded habitats will be lessresilient and at increased risk of local extinctionfrom both stochastic and deterministic processes(Rieman and Mclntyre 1993).

A potential confounding factor in our study wasthe genetic introgression of Taylor Fork cutthroattrout. Little is known about how phenotypic traitsare influenced by the level of introgression (Learyet al. 1995). Comparative studies of habitat useand survival of cutthroat trout having varying lev-els of introgression are needed, given that purepopulations represent only a small proportion ofremaining populations (Liknes and Graham 1988:Leary et al. 1995).

Spatial VariationOur results concur with studies illustrating that

different spatial scales are important in character-izing fish habitat requirements and assessing landuse effects (Frissell et al. 1986; Bozek and Rahel1991). Bozek and Rahel (1991) noted that manyfish habitat studies are conducted at the micro-habitat or reach spatial scales. In our study, reachmeasurements helped explain why certain spawn-

ing reaches had higher redd densities than others.However, knowledge of basin scale variables (e.g.,temperature, fish species composition and density)was also important. For example, extrapolation ofdata on redd density obtained from a subset ofTaylor Fork reaches without more general knowl-edge of redd distribution across the entire basinwould result in a high degree of error. High vari-ation in fish population and habitat characteristicsacross watersheds illustrates the importance of us-ing multiple scales to define suitable habitat andto design appropriate monitoring programs (Fris-sell et al. 1986; Hankin and Reeves 1988; Bozekand Rahel 1991).

AcknowledgmentsThis study was supported by funds provided by

the U.S. Forest Service Intermountain ResearchStation. Jack Mclntyre was instrumental in itsfunding and development. We thank Bruce Mayfor facilitating substrate analyses; Sue Ireland,Mike Jones, and Jennifer Staples for field assis-tance; and Robb Leary for providing data on ge-netic status. Reviews by Alan Barta, Mason Bry-ant, Brad Shepard, Tom Weaver, Bob White, andanonymous reviewers significantly improved themanuscript.

ReferencesAlt, D., and D. W. Hyndman. 1986. Roadside geology

of Montana. Mountain Press. Missoula, Montana.Beard, T. D., and R. F. Carline. 1991. Influence of

spawning and other stream habitat features on spa-tial variability of wild brown trout. Transactions ofthe American Fisheries Society 120:711-722.

Beschta, R. L., and W. L. Jackson. 1979. The intrusionof fine sediments into a stable gravel bed. Journalof the Fisheries Research Board of Canada 36:204-210.

Bjornn, T. C, and six coauthors. 1977. Transport ofgranitic sediment in streams and its effects on in-sects and fish. University of Idaho, Forest, Wildlife,and Range Experiment Station, Bulletin 17, Mos-cow.

Bozek, M. A., and W. A. Hubert. 1992. Segregation ofresident trout in streams as predicted by three hab-itat dimensions. Canadian Journal of Zoology 70:866-890.

Bozek, M. A., and F. J. Rahel. 1991. Assessing habitatrequirements of young Colorado River cutthroattrout by use of macrohabitat and microhabitat anal-yses. Transactions of the American Fisheries So-ciety 120:571-581.

Cederholm, C. J., and L. M. Reid. 1987. Impact of forestmanagement on coho salmon (Oncorhynchus ki-sutch) populations of the Clearwater River, Wash-ington: a project summary. Pages 373-398 in E.

778 MAGEE ET AL.

Salo and T. Gundy, editors. Streamside manage-ment: forestry and fishery interactions. Universityof Washington, College of Forest Resources, Con-tribution 57, Seattle.

Chapman, D. W. 1988. Critical review of variables usedto define effects of fines in redds of large salmonids.Transactions of the American Fisheries Society 117:1-21.

Cope, O. B. 1957. The choice of spawning sites bycutthroat trout. Proceedings of the Utah Academyof Sciences, Arts, and Letters 34:73-79.

Everest, F. H., C. E. McLemore, and J. F. Ward. 1980.An improved tri-tube cryogenic gravel sampler.U.S. Forest Service Research Note PNW-350.

Everest, F. H., and five coauthors. 1987. Fine sedimentand salmonid production—a paradox. Pages 98-142in E. Salo and T. Cundy, editors. Streamside man-agement: forestry and fishery interactions. Univer-sity of Washington, College of Forest Resources,Contribution 57, Seattle.

Frissell, C. A., W. J. Liss, C. E. Warren, and M. D.Hurley. 1986. A hierarchical framework for streamhabitat classification: viewing streams in a water-shed context. Environmental Management 10:199-214.

Grost, R. T., W. A. Hubert, and T. A. Wesche. 1991.Description of brown trout redds in a mountainstream. Transactions of the American Fisheries So-ciety 120:582-588.

Hankin, D. G., and G. H. Reeves. 1988. Estimating totalfish abundance and total habitat area in smallstreams based on visual estimation methods. Ca-nadian Journal of Fisheries and Aquatic Sciences45:834-844.

Holtby, L. B., and M. C. Healey. 1986. Selection foradult size in female coho salmon (Oncorhynchuskisutch). Canadian Journal of Fisheries and AquaticSciences 43:1946-1959.

Ireland, S. C. 1993. Seasonal distribution and habitatuse of westslope cutthroat trout in a sediment-richbasin in Montana. Master's thesis. Montana StateUniversity, Bozeman.

Leary, R. F, F. W. Allendorf, and G. K. Sage. 1995.Hybridization and introgression between introducedand native fish. Pages 91-101 in H. L. Schramm,Jr. and R. G. Piper, editors. Uses and effects ofcultured fishes in aquatic ecosystems. AmericanFisheries Society, Symposium 15, Bethesda, Mary-land.

Liknes, G. A. 1984. The present status and distributionof the westslope cutthroat trout (Salmo clarki lewisi)east and west of the Continental Divide in Montana.Montana Department of Fish, Wildlife, and Parks,Helena.

Liknes, G. A., and P. J. Graham. 1988. Westslope cut-throat trout in Montana: life history, status, andmanagement. Pages 53-60 in R. E. Gresswell, ed-itor. Status and management of interior stocks ofcutthroat trout. American Fisheries Society, Sym-posium 4, Bethesda, Maryland.

Lisle, T. E., and J. Lewis. 1992. Effects of sedimenttransport on survival of salmonid embryos in a nat-

ural stream: a simulation approach. Canadian Jour-nal of Fisheries and Aquatic Sciences 49:2337-2344.

Magee, J. P. 1993. A basin approach to characterizingspawning and fry rearing habitats for westslope cut-throat trout in a sediment-rich basin, Montana. Mas-ter's thesis. Montana State University, Bozeman.

McFadden, J. T 1969. Dynamics and regulation of sal-monid populations in streams. Pages 313-329 in T.G. Northcote, editor. Symposium on salmon andtrout in streams. University of British Columbia,Vancouver.

McNeil, W. J., and W. H. Ahnell. 1964. Success of pinksalmon spawning relative to size of spawning bedmaterials. U.S. Fish and Wildlife Service SpecialScientific Report Fisheries 469.

Overton, C. K., S. P. Wollrab, B. C. Roberts, and M. A.Radko. In press. R1/R4 (Northern/IntermountainRegion) fish and fish habitat standard inventory pro-cedures. U.S. Forest Service General Technical Re-port INT.

Plans, W. S., W. F. Megahan, and G. W. Minshall. 1983.Methods for evaluating stream, riparian, and bioticconditions. U.S. Forest Service General TechnicalReport INT-138.

Platts, W. S., R. J. Torquemada, M. L. McHenry, and C.K. Graham. 1989. Changes in salmon spawningand rearing habitat from increased delivery of finesediment to the South Fork Salmon River, Idaho.Transactions of the American Fisheries Society 118:274-283.

Reiser, D. W, and R. G. White. 1988. Effects of twosediment size-classes on survival of steelhead andchinook salmon eggs. North American Journal ofFisheries Management 8:432-437.

Rieman, B. E., and K. A. Apperson. 1989. Status andanalysis of salmonid fisheries: westslope cutthroattrout synopsis and analysis of fishery information.Idaho Department of Fish and Game, Federal Aidin Sport Fish Restoration, Project F-73-R-11, Boise,Idaho.

Rieman, B. E., and J. D. Mclntyre. 1993. Demographicand habitat requirements for conservation of bulltrout. U.S. Forest Service General Technical ReportINT-302.

Scrivener, J. C., and M. J. Brownlee. 1989. Effects offorest harvesting on spawning gravel and incubationsurvival of chum (Oncorhynchus keta) and cohosalmon (O. kisutch) in Carnation Creek, British Co-lumbia. Canadian Journal of Fisheries and AquaticSciences 46:681-696.

Shepard, B. B., S. A. Leathe, T. M. Weaver, and M. D.Enk. 1984a. Monitoring levels of fine sedimentwithin tributaries to Flathead Lake, and impacts offine sediment on bull trout recruitment. Pages 146-156 in E Richardson and F. H. Hamre, editors. Wildtrout III. Trout Unlimited, Vienna, Virginia.

Shepard, B. B., K. L. Pratt, and P. J. Graham. 1984b.Life histories of westslope cutthroat trout and bulltrout in the upper Flathead River basin, Montana.Montana Department of Fish, Wildlife, and Parks,Kalispell.

CUTTHROAT TROUT SPAWNING HABITAT 779

Smith, R. W, and J. S. Griffith. 1994. Survival of rain-bow trout during their first winter in the HenrysFork of the Snake River, Idaho. Transactions of theAmerican Fisheries Society 123:747-756.

Snyder, G., J. Black, P. Swetik, and H. F. Haupt. 1978.Modeling forest water quality of the upper Gallatinand Madison watersheds, using the benchmark sys-tem. U.S. Forest Service, Intermountain Forest andRange Experiment Station, Forestry Sciences Lab-oratory, Moscow, Idaho.

Thurow, R. F, and J. G. King. 1994. Attributes of Yel-lowstone cutthroat trout redds in a tributary of theSnake River, Idaho. Transactions of the AmericanFisheries Society 123:37-50.

van den Berghe, E. R, and M. R. Gross. 1989. Naturalselection resulting from female breeding competi-tion in a Pacific salmon (coho: Oncorhynchus ki-sutch). Evolution 43:125-140.

Weaver, T. M., and J. J. Fraley. 1993. A method tomeasure emergence success of westslope cutthroat

trout fry from varying substrate compositions in anatural stream channel. North American Journal ofFisheries Management 13:817-822.

Wolman, G. M. 1954. A method of sampling coarseriver-bed material. Transactions, American Geo-physical Union 35:951-956.

Young, M. K., W. A. Hubert, and T. A. Wesche. 1989.Substrate alteration by spawning brook trout in asoutheastern Wyoming stream. Transactions of theAmerican Fisheries Society 118:379-385.

Young, M. K., W. A. Hubert, and T. A. Wesche. 1991.Selection of measures of substrate composition toestimate survival to emergence of salmonids and todetect changes in stream substrates. North AmericanJournal of Fisheries Management 11:339-346.

Zar, J. H. 1984. Biostatistical analysis, 2nd edition.Prentice-Hall, Englewood Cliffs, New Jersey.

Received March 31, 1995Accepted April 15, 1996