Polymorphic mountain whitefish (Prosopiumwilliamsoni) in a coastal riverscape: size classassemblages, distribution, and habitatassociationsJames C. Starr1,2, Christian E. Torgersen31School of Environmental and Forest Sciences, University of Washington, Seattle, WA, USA2U.S. Geological Survey, Washington Water Science Center, Tacoma, WA, USA3U.S. Geological Survey, Forest and Rangeland Ecosystem Science Center, Cascadia Field Station, School of Environmental and Forest Sciences, University ofWashington, Seattle, WA, USA

Accepted for publication June 10, 2014

Abstract – We compared the assemblage structure, spatial distributions, and habitat associations of mountainwhitefish (Prosopium williamsoni) morphotypes and size classes. We hypothesised that morphotypes would havedifferent spatial distributions and would be associated with different habitat features based on feeding behaviour anddiet. Spatially continuous sampling was conducted over a broad extent (29 km) in the Calawah River, WA (USA).Whitefish were enumerated via snorkelling in three size classes: small (10–29 cm), medium (30–49 cm), and large(≥50 cm). We identified morphotypes based on head and snout morphology: a pinocchio form that had anelongated snout and a normal form with a blunted snout. Large size classes of both morphotypes were distributeddownstream of small and medium size classes, and normal whitefish were distributed downstream of pinocchiowhitefish. Ordination of whitefish assemblages with nonmetric multidimensional scaling revealed that normalwhitefish size classes were associated with higher gradient and depth, whereas pinocchio whitefish size classes werepositively associated with pool area, distance upstream, and depth. Reach-scale generalised additive modelsindicated that normal whitefish relative density was associated with larger substrate size in downstream reaches(R2 = 0.64), and pinocchio whitefish were associated with greater stream depth in the reaches farther upstream(R2 = 0.87). These results suggest broad-scale spatial segregation (1–10 km), particularly between larger and morephenotypically extreme individuals. These results provide the first perspective on spatial distributions and habitatrelationships of polymorphic mountain whitefish.

Key words: Mountain whitefish; resource polymorphism; river habitat; spatial patterns; generalised additive model

Introduction

Fluvial mountain whitefish (Prosopium williamsoni)offer an example of trophic polymorphism amongstream dwelling salmonids in temperate North Ameri-can river systems. Trophic polymorphism is a formof resource-based phenotypic diversification thatoccurs when exploitation of under-utilised resourcesnecessitates specific morphological characteristics.Trophic polymorphisms are most common amonganimals that subdue, handle, and capture their prey

with their mouth, such as birds, amphibians, andfishes (Wimberger 1994). In fishes, morphologicalvariation typically occurs in the head and mouth(Smith & Skulason 1996). Fishes exhibit many exam-ples of polymorphisms across taxa, particularlyamong species of cichlids in low latitudes (Klingen-berg et al. 2003), and in recently glaciated lakes inhigh latitudes (Robinson & Wilson 1994). The spatialstructure of lacustrine environments (e.g. littoral, lim-netic, and profundal habitats) provides a template fordiversification and has led to alternate phenotypes of

Correspondence: J. Starr, 338 Hunt Road, Port Angeles, WA 98363, USA. E-mail: jstarr1031@gmail.comJ. Starr is not currently employed by the University of Washington but was affiliated at the time of this research.

doi: 10.1111/eff.12163 1

Ecology of Freshwater Fish 2014 Published 2014. This article is a U.S.Government work and is in the public domain in the USA.

ECOLOGY OFFRESHWATER FISH

various species. Examples include, but are not limitedto, lake whitefish (Coregonus clupeaformis), brookchar (Salvelinus fontinalis), three-spinned stickleback(Gasterosteus aculeatus), and Arctic char (Salvelinusalpinus; Smith & Skulason 1996). The occurrence ofsympatric morphotypes may be maintained throughspatial segregation of discrete foraging habitats(Ostberg et al. 2009).Mountain whitefish exhibit resource-based poly-

morphism in rivers, with one form that develops anelongated and slightly upturned snout (McPhail &Troffe 1998), coined the ‘pinocchio’ form by Troffe(2000), which is distinguished from the normal formthat does not develop an elongated snout and has anevenly sloped head and blunt snout (Fig. 1). Moun-tain whitefish are widely distributed and abundantamong temperate rivers in western North America(Northcote & Ennis 1994), but polymorphic moun-tain whitefish have not been widely reported in theliterature. Morphological variation in mountain white-fish was historically thought to be the result of sexualdimorphism, with males developing an elongatedsnout (Evermann 1893), but research has identifiedthat both sexes can have elongated snouts (Troffe

2000). Observations on foraging tactics reported byTroffe (2000) from the upper Fraser River, BritishColumbia (Canada), indicated that the pinocchio formexpended approximately half of its time foraging forbenthic invertebrates using the elongated snout tooverturn substrate, whereas the normal form fed pri-marily on drifting invertebrates. Whiteley (2007)observed the greatest variation in snout morphologyin the largest fish (>47 cm), and the least amount ofvariation in the smallest fish (<25 cm), and con-cluded that a stage-specific ontogenetic shift likelyoccurs between the ages 2 and 3. This age range cor-responds to shifts in diet (Pontius & Parker 1973)and patterns of habitat use (Northcote & Ennis 1994)from juvenile to adult life stages. As an individualmoves from shallow to deeper water, divergent devel-opment of the snout occurs and leads to further spe-cialisation as the individual matures. However, thebiotic interactions and environmental factors influ-encing variation in snout development have not beendescribed.Quantifying the spatial distribution of alternate

morphotypes in stream fishes in relation to streamhabitat characteristics is important for understandingpolymorphic populations (Davis & Pusey 2010). Nostudies have examined polymorphic fluvial whitefishwithin the context of habitat heterogeneity and sizeclass assemblage structure. In addition, informationon coastal populations of mountain whitefish is lack-ing in the published literature. The objectives of ourstudy were to (i) evaluate differences in the spatialdistribution of whitefish morphotypes and size clas-ses, (ii) investigate the assemblage structure of mor-photypes and size classes and their relationships withaquatic habitat, and (iii) quantify the associationsbetween the relative density of morphotypes and lon-gitudinal variation in aquatic habitat. We hypothes-ised that morphotypes would be associated withdistinctly different habitat features based on previousstudies of feeding behaviour and diet and that thesehabitat differences would be associated with segrega-tion of morphotypes at the reach scale. We definespatial segregation here as broad-scale differences inthe overall distribution of morphotypes across theentire stream length, although overlap at finer spatialscales may occur.

Methods

Study area

The South Fork (SF) Calawah and mainstem Cala-wah River flow westward for 34 and 18 km, respec-tively, on the western side of the Olympic Peninsulain Washington State (USA; Fig. 2). They have acombined drainage area of 187 km2 and range in

(a)

(b)

(c)

1 cm

1 cm

Fig. 1. Mountain whitefish morphotypes in the Quileute basin,WA (USA): pinocchio (a), intermediate (b), and normal (c). Thethin white lines drawn over the snouts highlight differencesbetween morphotypes.

2

Starr & Torgersen

elevation from sea level to 1143 m (De Cillis 1998).The river channel is geomorphically stable and mod-erately confined by fluvial terraces in the lower main-stem Calawah River but is confined by steep valleywalls in the upper mainstem Calawah and SF Cala-wah rivers. The hydrograph is strongly influenced byrainfall, with peak flows occurring in November andDecember (Hook 2004). The upper 25 km of the SFCalawah River are located within the OlympicNational Park (ONP), and the downstream portion ofthe river, along with most of the Sitkum River, lieson federally owned forestland (U.S. Department ofAgriculture Forest Service), which is managed aslate-successional forest reserve (De Cillis 1998; Hook2004). The mainstem Calawah River flows throughprivately managed forests that are actively harvestedon a 35–40-year rotation (Hook 2004). The riparianforest is dominated by Sitka spruce (Picea sitchen-sis), red alder (Alnus rubra), bigleaf maple (Acermacrophyllum), western redcedar (Thuja plicata), andDouglas fir (Pseudotsuga mensiesii) in the lowlands,with western hemlock (Tsuga heterophylla) andsilver fir (Abies amabilis) at higher elevations (Smith2000).

In addition to mountain whitefish, the CalawahRiver and its tributaries support populations of sum-mer and winter steelhead (Oncorhynchus mykiss),coastal cutthroat trout (O. clarkii clarkii), fall cohosalmon (O. kisutch), and fall and summer chinooksalmon (O. tshawytscha). A small population ofriver-type sockeye salmon (O. nerka) spawns annu-ally in the SF Calawah. Non-salmonids includePacific lamprey (Entosphenus tridentata), westernbrook lamprey (Lampetra richardsonii), speckleddace (Rhinichthys osculus), longnose dace (R. cata-ractae), and sculpin (Cottus spp.).

Sampling

We conducted spatially continuous snorkelling andhabitat surveys on 16–19 August 2010 during sum-mer base flow (1.84 m3�s�1, USGS gauging station12043000) in the mainstem Calawah and SF CalawahRiver. Sampling began within the boundaries of theONP, 4 km upstream from the confluence of the SFCalawah and Sitkum rivers (47°560N, 124°120W) andended approximately 2 km upstream from the conflu-ence of the mainstem Calawah River and Bogachiel

0 4 8 12 16 km

N

WA

2

Calawah RiverN.F. C

alawah River

Upper S.F. Calawah River

Sitkum River

Hyas Creek

Elk Creek

Fig. 2. Location of the Calawah River on the Olympic Peninsula in Washington State, USA. The thick grey line between black hash marksrepresents the sampled stream length in the mainstem Calawah River and South Fork Calawah River.

3

Polymorphic mountain whitefish

River (47°560N, 124°260W, Fig. 2). The total extentof the survey was approximately 29 km, and 225channel units were sampled; 8 units were <0.25 mdeep and could not be sampled. Two kilometres ofthe lower mainstem Calawah River were not sampledbecause the wetted channel was too wide (>40 m)and deep (>5 m) for a single diver.We used well-established methods for snorkelling

in streams (Thurow 1994; Torgersen et al. 2006;Brenkman et al. 2012). Snorkelling offered an alter-native to electrofishing and provided a relativelyaccurate and efficient method for enumerating salmo-nids in our study section, in which the channel wastoo small to sample with a boat-mounted electrofish-er, and yet too large to sample with a backpack elec-trofisher (Cunjack et al. 1988; Joyce & Hubert2003). All surveys were conducted by the diver mov-ing downstream, typically in a single pass adjacent tothe thalweg. In locations with high salmonid densityand instream cover, the diver made a second passand averaged the two counts. Habitat size and com-plexity, fish species and size, density, and the inten-sity of sampling effort may affect the efficiency ofsnorkelling estimates of fish abundance (Rodgerset al. 1992; Bayley & Dowling 1993). Our snorkel-ling surveys of whitefish distribution and abundancedid not account for sampling efficiency using alterna-tive methods to verify whitefish counts and sizes.Using one highly experienced diver, we maintained aconsistent sampling bias that allowed us to evaluatepatterns of distribution and relative abundance butdid not provide accurate estimates of total fish abun-dance (Hankin & Reeves 1988).Fish and habitat surveys began at 09:00 and were

completed each day by 16:00 to avoid diurnal andcrepuscular movement patterns of salmonids (Woot-ton 1998) and to maximise underwater visibility.During the fish surveys, the water was very clear,and the diver was able to observe fish across theentire wetted width. Counts of whitefish morphotypeswere recorded in each channel unit (i.e. pool andnonpool). In addition to head shape, pinocchio white-fish were distinguished from normal whitefish by thepresence of highly visible white scar tissue on the tipof the snout (Fig. 1). This white scar tissue and elon-gated snout was visible but less pronounced in thesmallest size class of whitefish, and fish <20 cmwere rare among both morphotypes. The white scartissue and differences in snout morphology on themedium and large size classes of the pinocchio mor-photype were easily distinguishable by the diver. Fishthat were intermediate between pinocchio and normalin terms of snout length also displayed white scar tis-sue and an elongated snout (Fig. 1). Because theseintermediate morphotype fish showed signs of devel-oping into the pinocchio form, they were included in

the count of pinocchio morphotypes. To avoid biasassociated with estimating fish size underwater (Edgaret al. 2004), three size classes were used that wereeasily distinguishable underwater: small (10–29 cm),medium (30–49 cm), and large (≥50 cm).In each sampled channel unit, a technician on the

bank measured data on length, wetted width, maxi-mum depth, and channel slope, and the diver esti-mated substrate composition and instream covervisually. Maximum unit depth was measured with astadia rod, and gradient was measured with a stadiarod and clinometer. Length and width were measuredwith a laser rangefinder (Impulse 200 LR). Substratecomposition was visually estimated as a per cent ofpebble, cobble, boulder, and bedrock. The dominantsubstrate of each channel unit was assigned a rankbased on median size: pebble (16–64 mm) = 1; cob-ble (64–256 mm) = 2; boulder (>256 mm) = 3; andsimilar estimates were averaged (Allan 1995).Bedrock substrate was converted to a binary variable(i.e. present or absent) because it did not constitute asignificant portion of the streambed surface area(i.e. <10%) and typically was covered by a lens ofcobble or pebble. Divisions between channel unitswere mapped with a global positioning system (GPS;Garmin III). Field-measured length was correctedwith map distances using GPS point data in ArcViewGIS (version 9.1, ESRI 2004) and high-resolutiondigital orthographic imagery (USDA 2009). Channelunits were classified as either a pool or nonpool. Apool was defined as a habitat unit with smooth sur-face flow, a maximum depth that was at least 25% ofbankfull depth, and length, or width that was at least10% of bankfull width (Montgomery et al. 1995).Channel type was identified as pool-riffle, forcedpool-riffle, or plane bed (Montgomery & Buffington1997).

Data analysis

Data collected at the unit scale were binned intoapproximately 1-km reaches (n = 29) for analysis ofdistribution, assemblage structure, and habitat associ-ations. Bin numbers started at the downstream end ofthe Calawah River and continued to the upstream endof the SF Calawah River. The length of each bin wasdetermined according to changes in channel morphol-ogy, including major tributary junctions, and naturalbreaks in channel and unit type. Channel type wascalculated as a percentage of total bin length, andpool area was calculated as a percentage of total binarea. Floodprone and bankfull width were measuredon high-resolution orthophotographs (USDA 2009).Bankfull width, floodprone width, valley widthindex, substrate score, and maximum unit depth wereaveraged for each bin. Bin gradient was derived by

4

Starr & Torgersen

multiplying unit-scale gradient by length, summingacross all units within the bin, and dividing by totalbin length. Counts of mountain whitefish weresummed for each bin and standardised in two differentways. First, data were standardised by columns (per-centage cumulative abundance) and plotted versus dis-tance upstream for comparison of spatial distributionsin SigmaPlot, version 10.0 (SYSTAT 2006). Second,abundance data were standardised as relative density(Dr) to account for variation in bin length and to reducevariance in the datasets (Brenkman et al. 2012). Wedefine relative density as the ratio between fish densitywithin a sample unit (approximately 1-km bin) to theoverall density across the entire sample area:

Dr ¼fili

� �

ftlt

� � (1)

where fi = number of fish per 1-km custom bin,li = length (km) of custom bin, ft = total number offish, and lt = total length (km) sampled. Positive andnegative values of relative density indicated densitiesof mountain whitefish that were above and below theaverage density for the entire length of stream sam-pled, respectively (Brenkman et al. 2012). Relativedensity and mean reach-scale channel and floodpronewidths were transformed using the natural logarithm(Zar 1984).Assemblage structure of normal and pinocchio

whitefish size classes was assessed with nonmetricmultidimensional scaling (NMS) in the statisticalsoftware R (R Core Development Team 2009), usingthe ‘vegan’ package. NMS is a powerful nonparamet-ric ordination technique for analysis of communitydata that violates the assumption of multivariate nor-mality (McCune & Grace 2002). NMS uses the rankorder of distances to represent objects in multivariatespace (Digby & Kempton 1987). We used the Bray-Curtis distance coefficient, a distance measure com-monly applied to count data with zero values, and10,000 iterations to avoid reaching a local minimumin stress (Legendre & Legendre 1998). Completeabsence of all ‘species’ (i.e. row sum of observa-tions = 0) cannot be calculated as a distance coeffi-cient in multivariate analysis. Therefore, we removedseveral rows (n = 5) from the multivariate data set,which corresponded to bins 26–29 in the upper SFCalawah River, and bin 20 in the lower SF CalawahRiver. A two-dimensional ordination was used in theanalysis to simplify interpretation. Although addi-tional dimensionality did result in a lower stressvalue, this did not change the dominant gradients inassemblage structure. Centroids of size classes andmorphotypes in ordination space were plotted in con-

junction with vectors representing correlationsbetween axis scores and a second matrix of environ-mental variables. We used a cut-off value ofP ≤ 0.10 for plotting vectors for environmental vari-ables in the ordination (Legendre & Legendre 1998).Associations between the relative density of mor-

photypes with respect to environmental variableswere quantified using a series of generalised additivemodels (GAMs) in R (R Core Development Team2009), with the ‘mgcv’ package (Wood 2006). Thedefault smoothing function, thin-plate penalisedregression splines (Wood 2003), was used to modelnon-linear relationships with a Gaussian error distri-bution and identity link function:

dðliÞ ¼ b0 þ s1ðx1iÞ þ s2ðx2iÞ þ � � � (2)

where E(yi) � li, yi are the independent observations,d is a ‘link function’ (identity in this case), b0 is theintercept, and s1 is a smoothing function for the linearpredictor x1 (Wood 2001).Generalised additive models have several advanta-

ges over the commonly used generalised linear modelwhen testing for species habitat relationships, particu-larly when data are collected continuously over alarge extent (Hastie & Tibshirani 1990). A GAM pro-vides a less restrictive non-linear approach to model-ling in which the data determine the shape of therelationship (vs. the limited response shapes in aparametric model) through a series of smoothingsplines (Hastie & Tibshirani 1990). This approach isparticularly effective when the response of an organ-ism to a habitat variable is not a simple linear orcurve–linear relationship. Smoothed estimates arebased on a weighted average of neighbouring obser-vations using a back-fitting algorithm (Hastie & Tib-shirani 1987). This approach is particularly usefulwhen the form of the relationship between theresponse and predictor variable is unknown. Theform of the relationship may range from a straightline to increasingly complex nonparametric curves.Prior to fitting models for each morphotype, we

examined multicollinearity among independent vari-ables. Independent variables that were correlated(│r│ > 0.3) were excluded from analysis. We thenvisually examined spatial patterns of physical habitatfeatures in relation to patterns in the relative densityof whitefish. Physical habitat was modelled as aresponse to distance upstream using the thin-platepenalised regression spline smoother in the ‘mgcv’package (Wood 2006) in R (R Core DevelopmentTeam 2009). Resulting spatial patterns of physicalhabitat provided a context for understanding the rela-tionship between the relative density of whitefish anddistance upstream. Substrate score was modelled as a

5

Polymorphic mountain whitefish

function of distance upstream and displayed a spatialpattern of peaks and troughs (i.e. larger and smallersubstrate size). However, the relationship in theGAM was only significant when the effect degrees offreedom (i.e. ‘knots’) was >9, and the linear correla-tion was not significant (R2 = 0.003, P > 0.1). Meanmaximum depth did not display a strong spatial pat-tern (R2 = 0.05, P > 0.1) in the GAM. Therefore, weassumed that including either substrate or depth in anadditive model with distance upstream wouldnot result in inflated variance due to the effects ofco-linearity.We also examined spatial autocorrelation in the

model residuals. Spatial autocorrelation is the ten-dency for samples close together in space to be moresimilar than those far apart, which violates theassumption of independence in statistical analyses(Legendre 1993). We used GAMs to demonstrate thatdistance upstream was significantly associated withboth normal and pinocchio whitefish relative density(P < 0.01). We then extracted the model residuals toevaluate spatial autocorrelation using variograms(Palmer 2002) with the ‘variog’ function in the‘geoR’ package in R (R Core Development Team2009). We assessed semi-variance at multiple lagintervals and distances and determined that the resid-uals did not display strong spatial dependence. Weincluded distance upstream as a covariate when quan-tifying associations between whitefish abundance andother physical habitat characteristics.We analysed associations of whitefish relative den-

sity with each physical habitat variable individually,including distance upstream as the first covariate ineach model to de-trend spatial patterns. The variablesthat explained the most variance in each normal andpinocchio whitefish GAM were used in a model ofthe alternate morphotype to determine how relation-ships with the same variables differed between mor-photypes. We then used stepwise forward-variableselection and analysis of deviance (ANODEV) with anapproximate v2 test, (Hastie 1991) to select a suite ofphysical habitat variables that were associated witheach morphotype. Fitted values and standard errors

from the best GAMs for normal and pinocchio white-fish were back-transformed and plotted versus dis-tance upstream (x-axis) in SigmaPlot (SYSTAT2006) using locally weighted scatterplot smoothing(LOWESS; Trexler & Travis 1993) with a second-degree polynomial smoothing parameter. Observedvalues were included in plots to evaluate how welleach GAM predicted the locations and magnitudes ofpeaks and troughs in the relative density of normaland pinocchio whitefish.

Results

Longitudinal distribution of whitefish size classes

Mountain whitefish abundance was dominated bypinocchio whitefish (75% of all individuals), andnearly 50% of all whitefish counted were medium-sized pinocchio whitefish (Tables 1 and 2). Nowhitefish were observed in the SF Calawah Riverupstream of the confluence with the Sitkum River.Spatial distributions differed between pinocchio andnormal whitefish. Normal whitefish were distributedfarther downstream than pinocchio whitefish, andlarge size classes of both morphotypes were distrib-uted farther downstream than small and mediumsize classes (Fig. 3). Spatial patterns of small andmedium size classes of normal whitefish were simi-lar; 50% of small and medium normal whitefishwere in the lower 9.6 and 10.7 km of the studyarea, respectively (Fig. 3). Fifty per cent of normalwhitefish in the large size class were in the lower6.6 km of the study area. The cumulative abun-dance of large, normal whitefish increased from50% to 80% at 7.8 km. The upper extent of smalland medium size classes (24.1 and 25.1, respec-tively) was much farther upstream than the largesize class (16.4 km).Pinocchio whitefish displayed size class distribu-

tion patterns similar to those of normal whitefishsuch that small- and medium-sized fish were distrib-uted upstream and larger fish were downstream.However, small and medium size classes of pinocchio

Table 1. Abundance and relative density of mountain whitefish (Prosopium williamsoni) by morphotype and size class in the Calawah River. See Equation 1 forthe calculation of relative density.

Morphotype Size class Length (cm) Raw abundance Per cent abundance Mean relative density (�SD)

Normal mountain whitefish Small 10–29 283 9 0.968 � 1.396Medium 30–49 469 14 0.975 � 1.167Large ≥50 70 2 0.986 � 1.935All size classes 822 25 0.974 � 1.192

Pinocchio mountain whitefish Small 10–29 566 17 0.965 � 1.420Medium 30–49 1542 47 0.980 � 0.979Large ≥50 348 11 0.989 � 0.907All size classes 2456 75 0.978 � 1.004

6

Starr & Torgersen

whitefish were distributed farther upstream than thesesize classes of normal whitefish. Approximately 50%of the small and medium pinocchio whitefish weredistributed in the lower 13.1 and 12.8 km, respec-tively. Small pinocchio whitefish reached 78% incumulative abundance at 14.9 km, and mediumpinocchio whitefish reached 76% in cumulativeabundance at 15.2 km. Pinocchio whitefish in bothsmall and medium size classes extended 25 kmupstream to the confluence with the SF CalawahRiver and Sitkum River. The distribution of the largepinocchio whitefish was similar to the distribution ofmedium-sized normal whitefish. Large pinocchiowhitefish reached 51%, 75% and 100% in cumula-tive abundance at 9.0, 14.1 and 25 km upstream,respectively.

Size class assemblage structure

Multivariate ordination with NMS provided evidenceof segregation among size classes in both morpho-types (Fig. 4). Centroids of normal whitefish sizeclasses were grouped in the lower left quadrant of theordination plot. Small and medium size classes ofnormal whitefish were associated with both reachgradient and mean maximum depth, whereas thelarge size class was associated with reach gradient.All three size classes of normal whitefish were inver-sely associated with distance upstream, and the largesize class had the strongest inverse relationship.Pinocchio whitefish size classes were grouped in theupper two quadrants of the ordination plot. Smalland medium size classes of pinocchio whitefish were

Distance upstream (km)

Cum

ulat

ive

abun

danc

e (%

)

Fig. 3. Cumulative abundance of mountainwhitefish morphotypes and size classesversus distance upstream.

Table 2. Summary statistics of physical habitat characteristics in the lower and upper mainstem Calawah River, and the lower and upper South Fork (SF)Calawah River.

Habitat characteristics

Lower mainstem (n = 10) Upper mainstem (n = 6) Lower SF (n = 9) Upper SF (n = 4)

Mean Range SD Mean Range SD Mean Range SD Mean Range SD

Length (km) 1.0 0.86–1.19 0.12 1.0 0.88–1.09 0.09 1.02 0.87–1.26 0.11 0.94 0.75–1.06 0.15Wetted width (m) 33 26–42 4.2 29 22–38 6.3 20 16–25 2.7 9.8 8–12 1.7Bankfull width (m) 47 38–59 5.5 40 31–45 4.9 39 22–53 10.1 17.5 15–21 2.9Channel type (%)Pool-riffle 52 0–100 39 44 0–100 50 37 0–100 39 27 0–54 25Forced pool-riffle 30 0–100 33 17 0–100 41 41 0–100 40 57 13–100 37Plane bed 17 0–54 23 40 0–100 49 22 0–100 44 16 0–44 21

Gradient (%) 1.2 0.02–1.5 0.04 1.2 0.09–1.5 0.03 1.2 0.08–2.2 0.05 1.1 1–1.4 0.02Pool area (%) 35 0–76 23 37 24–54 12 29 0–66 29 23 11–44 16Mean max. depth (m) 1.45 0.94–2.05 0.35 1.5 0.77–2.12 0.49 1.25 0.54–1.80 0.46 1.05 0.74–1.33 0.26Substrate score 2.3 1.7–2.7 0.33 2.2 1.5–2.75 0.41 2.3 1.8–2.9 0.40 2.2 2.0–2.4 0.16Valley width index 1.54 1.32–1.78 0.13 1.53 1.30–1.84 0.19 1.56 1.31–2.05 0.22 1.52 1.49–1.57 0.03

Sections were sampled at the unit scale and data were binned into approximately 1-km reaches (N = 29) using major tributary junctions, and natural breaks inchannel and unit type. Substrate score corresponds to pebble, cobble, and boulder (i.e. scores of 1, 2, and 3, respectively). Valley width index is the ratio offlood-prone width to channel width.

7

Polymorphic mountain whitefish

positively associated with per cent pool area andmean maximum depth. Large pinocchio whitefishwere positively associated with distance upstream.All three size classes of pinocchio whitefish weremore closely associated with distance upstream thannormal whitefish. The grouping of size classes bymorphotype indicated that combining size classeswas appropriate for analyses of habitat associations.

Habitat associations with the relative density ofmorphotypes

Generalised additive models indicated that variationin the relative density of normal whitefish was bestexplained by distance upstream, substrate size (anincrease in substrate score is synonymous with anincrease in size), and a linear coefficient for theoccurrence of bedrock substrate (Fig. 5a,b;R2 = 0.64). The association between normal white-fish relative density and substrate, given the effectsof distance upstream and bedrock occurrence, waspositive up to a substrate score of 2.4 (analogous toreaches dominated by boulder and boulder/cobblemix); at substrate scores >2.4, there was an inverserelationship between substrate size and normal white-fish relative density. Distance upstream (d.f. = 4.9,P < 0.001), substrate size (d.f. = 4.1, P = 0.02), andbedrock occurrence (d.f. = 1, P = 0.02) resulted in46%, 14% and 18% reductions in deviance, respec-tively. No other variables were significantly associ-ated with the relative density of normal whitefish.The relative density of pinocchio whitefish was

positively associated with distance upstream toapproximately 13 km and was negatively associatedwith distance upstream to 29 km (Fig. 5c). However,there was no significant association with substrate

size (Fig. 5d; R2 = 0.59). Pinocchio whitefish rela-tive density was best explained by distance upstreamand mean maximum depth (Fig. 5g,h; R2 = 0.87).Pinocchio whitefish displayed a strong positivecurve–linear relationship with mean maximum depth,with higher densities in reaches with a mean maxi-mum depth ≥1.2 m. In contrast, there was a weakpositive linear relationship (R2 = 0.47) between nor-mal whitefish relative density and mean maximumdepth (Fig. 5f), after accounting for the effect of dis-tance upstream (Fig. 5e). Mean maximum depth(d.f. = 2.9, P < 0.001) and distance upstream(d.f. = 7.3, P < 0.001) in the GAM for pinocchiowhitefish resulted in 67% and 24% reductions indeviance, respectively. Outliers were omitted fromFig. 5c (x = 13, y = 8.8, and x = 15, y = 12.7),Fig. 5d (x = 3.2, y = 4.9), Fig. 5g (x = 15, y = 5.8),and Fig. 5h (x = 2, y = 7.7). No other variables weresignificantly associated with pinocchio whitefish rela-tive density.Plots of fitted and observed whitefish relative den-

sities versus distance upstream indicated that theGAM for normal whitefish accurately predicted thelocations of peaks and troughs in whitefish relativedensity (Fig. 6a). However, this model was not ableto predict the magnitudes of peaks and troughs in rel-ative density. For example, the three highest peaks inobserved normal whitefish relative density exceededthe standard error interval. In contrast, the pinocchiowhitefish GAM accurately predicted both the loca-tions and the magnitudes of peaks and troughs in rel-ative density (Fig. 6b). Observed values forpinocchio whitefish relative density were generallywithin the standard error intervals, and the secondhighest observed peak in pinocchio whitefish relativedensity matched the fitted value.

Fig. 4. Non-metric multidimensionalscaling (NMS) ordination of mountainwhitefish morphotypes and size classes.Black diamonds represent 1-km bins inspecies space. Small, medium, and largeopen circles (normal) and open squares(pinocchio) indicate the centroids of sizeclasses. Vectors indicate the direction andrelative correlation of environmentalvariables with respect to ordination axes.Only vectors with P ≤ 0.10 and R2 > 0.30are plotted.

8

Starr & Torgersen

Distance upstream (km) Mean max. depth (m)

P = 0.011P = 0.032

P < 0.001

P = 0.002

Rel

ativ

e de

nsity

Rel

ativ

e de

nsity

Substrate score

P = 0.008 P = 0.059

P = 0.001

Rel

ativ

e de

nsity

Rel

ativ

e de

nsity

P = 0.505

Substrate score

Mean max. depth (m)

(a) (b)

(c) (d)

(e) (f)

(g) (h)

Fig. 5. Thin-plate regression splines of modelled relative density of normal (a, b, e, f) and pinocchio (c, d, g, h) whitefish with respect tothe additive effects of distance upstream and substrate score, and distance upstream and mean maximum depth. Substrate score correspondsto pebble, cobble, and boulder (i.e. scores of 1, 2, and 3, respectively). Solid circles represent partial raw residuals, dashed lines indicate�2 standard errors, and P-values refer to the significance of the smoothed parameter.

9

Polymorphic mountain whitefish

Discussion

Our results showed that pinocchio and normal moun-tain whitefish morphotypes had different spatial dis-tributions that were associated with differences inaquatic habitat. Overall distribution patterns sug-gested that there was segregation of whitefish mor-photypes at a broad spatial scale (1–10 km). It isimportant to note that previous studies of segregationtypically focus on finer spatial scales (Fausch &White 1981; Gibson & Erkinaro 2009). Our studyfocused on overall patterns of segregation across andentire stream length.

Longitudinal distribution

Normal whitefish were more abundant in the lowerCalawah River, and pinocchio morphotype fish weremore abundant in the upper Calawah River and thelower SF Calawah River (Fig. 3). No whitefish werefound in the upper SF Calawah River where mean wet-ted width was <10 m (Table 2). Although there are no

previous studies on the spatial distribution of polymor-phic whitefish in rivers, our results corroborate otherreports that mountain whitefish are generally foundlower in the watershed and in mainstem habitats (Platts1979; Meyer et al. 2009). Smaller tributaries may notprovide suitable habitat conditions, such as adequatedepth and cover for mountain whitefish (Sigler 1951).However, McPhail & Troffe (1998) and Meyer et al.(2009) found that mountain whitefish occupied smallerstreams (5–10 m wetted width). The species also hasbeen observed in small tributaries (mean wetted width<10 m) within the Calawah River and Hoh Riverdrainages during late winter and spring (J.C. Starr &J.R. McMillan, unpublished data). Meyer et al. (2009)speculated that whitefish in the southern portion oftheir range typically use larger streams, whereas white-fish in the northern portion of their range generally arefound in smaller streams. Coastal mountain whitefishmay move into tributaries after they spawn in the fall toseek refuge during periods of higher flow.We observed a longitudinal pattern of larger fish

downstream and smaller fish upstream in both mor-photypes (Fig. 3). Similar patterns with large individ-uals downstream have been documented for otherspecies (Power 1984; Welcomme 1985). Our resultssuggest that large mountain whitefish, regardless ofmorphotype, may require a greater volume of habitatduring summer base flow conditions, perhaps to bal-ance the trade-offs between predation risk and forag-ing opportunities.Small and medium size classes of both morpho-

types had similar spatial distributions that were dis-tinctly different from the large size classes of normaland pinocchio whitefish, which were distributed far-ther downstream. We observed differences betweensmall and medium versus large size classes that weresimilar to the observations by Whiteley (2007), whofound that morphological and dietary variation wasgreatest in larger individuals (>47 cm), and least insmaller individuals (<25 cm). Although the patternsin cumulative abundance among size classes corre-sponded with this spectrum of size-related morpho-logical variation, the relative abundances did not. Ifdietary variation is less in smaller individuals, wewould expect that the abundance of smaller normalwhitefish would be greater in this size class. How-ever, the abundance of small (10–29 cm) pinocchiowhitefish was twice that of small normal whitefish. Itis possible that juvenile mountain whitefish in low-elevation coastal river systems, specifically on theOlympic Peninsula, undergo more rapid growth asjuveniles and make a shift in feeding habits earlier intheir ontogeny (McHugh 1942). Currently, there islimited information on growth and maturation ofcoastal mountain whitefish populations, and moreresearch in this area is needed.

Rel

ativ

e de

nsity

Distance upstream (km)

Rel

ativ

e de

nsity

(a)

(b)

Fig. 6. Longitudinal variation in modelled and observed relativedensity of normal (a) and pinocchio (b) mountain whitefish. Thesolid lines are locally weighted scatterplot smoothing (LOWESS)of the fitted values, dashed lines represent a LOWESS smooth of�2 standard errors, and open circles are observed values.

10

Starr & Torgersen

Size class assemblage structure

We found that whitefish morphotypes occurred indistinct assemblages based on size class. For exam-ple, small and medium size classes occurred together,whereas the large size class was distinct from othersize classes in ordination space (Fig. 4). The positiveassociation between depth and the small and mediumsize classes of both morphotypes provided furtherevidence of fine-scale spatial overlap between white-fish morphotypes in deep-water habitat (i.e. pools).These results suggest that habitat use becomes morespecialised in larger individuals that are more pheno-typically extreme, i.e. have more pronounced, elon-gated snouts. These results are similar to the findingsby Whiteley (2007), in which larger individuals(>47 cm) displayed greater morphological variationthat was associated with significant differences indiet. Whiteley (2007) concluded that this stage-spe-cific ontogenetic shift occurs at approximately3 years of age (i.e. 20–25 cm in length). There maybe a significant dietary shift between juvenile andadult life stages (Pontius & Parker 1973) that corre-sponds with a change in habitat use from shallow,marginal habitat to deeper and faster habitat in themain channel (Northcote & Ennis 1994). In ourstudy, the medium size class (30–49 cm) was themost abundant size class. The small size class ofwhitefish (10–29 cm) was dominated by individualsthat were 25–29 cm in length and had the head shapeand white scar tissue on the tip of the snout charac-teristic of the pinocchio morphotype. Morphologicalvariation appeared less extreme in fish <25 cm, andwhitefish <20 cm were extremely rare. Benjaminet al. (2014) showed that the majority of juvenilewhitefish move downstream in their first year of life.Thus, our study area may not encompass the down-stream extent of juvenile whitefish in the Quileutebasin. Our results support the findings by Pontius &Parker (1973) that mountain whitefish undergo aniche shift later in ontogeny (20–24 cm) and the con-clusions by Whiteley (2007) that this shift leads tophenotypic diversification. Although the positiveassociation between gradient and large normal white-fish may be a departure from the typical relationshipbetween whitefish and depth, this association doessuggest that normal whitefish may be using high-gra-dient reaches for their high rates of invertebrate preydelivery (Troffe 2000).

Associations between whitefish relative density andaquatic habitat

Generalised additive models enabled us to quantita-tively assess spatial associations between mountainwhitefish relative density and aquatic habitat. When all

size classes were combined in a model of relative den-sity, normal whitefish were associated with larger sub-strate (Fig. 5b). In our study, it is possible that deeper,fast-water habitat, combined with large boulder sub-strate in the lower Calawah River provided better for-aging opportunities for drifting invertebrates preferredby normal whitefish (sensu Troffe 2000; Whiteley2007). However, distance upstream, substrate, andbedrock occurrence explained only 64% of the varia-tion in relative density, suggesting that other variables,which we did not consider (e.g. temperature, primaryproduction, and aquatic invertebrate communities),may explain additional longitudinal variation in theabundance of normal whitefish. The weak linear rela-tionship between normal whitefish relative density anddepth (Fig. 5f) suggests that the normal morphotype isassociated with large substrate in deep-water habitatsthat provides refuge from both flow and predation foroptimal feeding on drifting aquatic invertebrates.Pinocchio whitefish relative density was strongly asso-ciated with mean maximum depth (Fig. 5h). Mountainwhitefish have been shown to be associated withgreater depth (DosSantos 1985; Torgersen et al. 2006),and it has been speculated in other studies that deeppools provide adequate cover for the species (Sigler1951). The pinocchio whitefish GAM fitted peaks inthe relative density located in the upper Calawah Riverand a second in the SF Calawah River. Bedrock out-croppings and forced pool-riffle channel morphologydominate these river segments (Table 2), creating deepscour pools that provide slower velocities, cover frompredators, and small-diameter substrate. Deep poolswith small-diameter substrate in low-gradient reachesmay provide the habitat conditions that pinocchiowhitefish require to overturn substrate and forage forbenthic invertebrates (Troffe 2000). We observed indi-vidual pinocchio whitefish using their snout to over-turn pebbles and small cobbles, but we did notquantify the extent of this behaviour.Feeding habits of mountain whitefish morphotypes

may partially explain their patterns of habitat associa-tion (Northcote & Ennis 1994). The species is oftenassociated with greater stream width, lower gradient,deep pool habitat (Sigler 1951; Meyer et al. 2009),and small substrate size (DosSantos 1985). They alsohave been associated with higher-gradient shallowhabitats (e.g. runs and riffles; Torgersen et al. 2006).Thompson & Davies (1976) found that Sheep Rivermountain whitefish occupied positions in the watercolumn that were 2 to 10 cm from the stream bedwhile they were feeding on drifting invertebrates.None of these whitefish were observed feeding onbenthic invertebrates, although gravel or sandoccurred in the stomach contents of 54% of the fishexamined. In contrast, DosSantos (1985) found thatmountain whitefish in the Kootenai River fed

11

Polymorphic mountain whitefish

disproportionately on chironomids, and underwaterobservations revealed that these fish were using theirsnout to overturn smaller substrate. Differences inhabitat relationships and feeding behaviour describedin the aforementioned studies may be explained inpart by specific polymorphic associations with habitatunique to a given basin or subbasin.

Ecological importance and implications of polymorphismin mountain whitefish

Across their range, mountain whitefish play animportant role in stream ecosystems, but their ecol-ogy is still not fully understood. Fluvial mountainwhitefish spend their entire lives in the freshwaterenvironment and have been reported to live up to29 years (McHugh 1942). Consequently, mountainwhitefish have been identified as a potential indicatorspecies of riverine habitat conditions and water qual-ity (McPhail & Troffe 1998). Studies indicate thatmountain whitefish exhibit homing behaviour associ-ated with summer rearing and fall spawning locationsand that they are capable of extensive movementwithin a stream network (Baxter 2002; Benjaminet al. 2014) and, thus, may contribute substantially tonutrient transport within a watershed (Lance & Baxter2011). Movement by mountain whitefish throughoutstream networks during different life stages suggeststhat they may require a watershed-scale approach tohabitat conservation that has been applied to managedstocks of anadromous salmonids (Roni et al. 2002).Mountain whitefish have been studied extensively

in interior rivers (e.g. east of the Cascade Mountainsand in the Rocky Mountains) in the USA and Can-ada, yet comparatively little is known about theirecology in coastal watersheds of the USA. TheOlympic Peninsula in western Washington offers aunique example of coastal mountain whitefish popu-lations. Whiteley et al. (2006) found that whitefishfrom the Hoh River basin, which flows into thePacific ocean on the western side of the OlympicPeninsula, were genetically related to the coastal pop-ulation of mountain whitefish in the lower FraserRiver in British Columbia, Canada. However, white-fish from the North Fork Skokomish River that flowsinto the Hood Canal on the eastern side of the Olym-pic Peninsula were more closely related to popula-tions west of the Cascade Mountains in Oregon,Washington, and British Columbia. Beyond range-wide genetic assessments, there have been no otherstudies of mountain whitefish in the coastal regionsof Washington, Oregon, and northern California.Our results provide an important perspective on

coastal mountain whitefish ecology, including thespatial distributions of morphotypes, size classassemblages, and associations with aquatic habitat at

multiple spatial scales. Broad-scale spatial segrega-tion of phenotypically extreme individuals (i.e. thosewith more pronounced, elongated snouts) and the dif-fering habitat associations of the two morphotypesmay have the ecological function of reducing intra-specific competition and maintaining phenotypicdiversity. The degree of reproductive isolation (spa-tial or temporal) between the two morphotypes isunknown. Because mountain whitefish are broadcastspawners, reproductive isolation between morpho-types may be very limited (Whiteley 2007). Addi-tional investigation on reproductive behaviour isneeded to determine the degree of spatial and tempo-ral isolation between morphotypes. Condition factor(i.e. lipid levels), growth, movement, and survival ofthe two morphotypes at juvenile and adult life stagesalso require further examination to elucidate thephysiological trade-offs of phenotypic diversification.Spatially explicit data on basin-wide abundance col-lected in conjunction with detailed information onmovement, growth, and survival may be required toidentify the mechanisms that maintain the phenotypicplasticity found in mountain whitefish. Coastal moun-tain whitefish constitute an important component ofthe biomass and ecological complexity of river sys-tems in the Pacific Northwest. Polymorphism enablesmountain whitefish to diversify its food base, and thisprovides a potential explanation for why mountainwhitefish are often more abundant than sympatricsalmonids in rivers across their range.

Acknowledgements

We thank Walter A. (Peter) Starr for generous moral andfinancial support to J.C.S. throughout this project. The U.S.Geological Survey (USGS) Western Fisheries Research Centeralso provided funding. We are grateful to Aaron Ruesch [pre-viously with the University of Washington (UW), School ofEnvironmental and Forest Sciences] for his assistance withdata collection. Samuel Brenkman (Olympic National Park,Chief Fisheries Biologist), and the USGS Forest and Range-land Ecosystem Science Center (FRESC) Cascadia Field Sta-tion provided essential survey equipment and field support.We also thank Susan Bolton (UW School of Environmentaland Forest Sciences), Thomas Quinn (UW School of Aquaticand Fishery Sciences), and the Quinn Lab group for essentialfeedback and constructive criticism. We thank Jeffrey Duda(Western Fisheries Research Center, Research Ecologist) andtwo anonymous reviewers for their comments that greatlyimproved the manuscript. Any use of trade, product, or firmnames is for descriptive purposes only and does not implyendorsement by the U.S. Government.

References

Allan, J.D. 1995. Stream ecology: structure and function ofrunning waters. New York: Chapman and Hall.

12

Starr & Torgersen

Baxter, C.V. 2002. Fish movement and assemblage dynamicsin a Pacific Northwest riverscape. Corvallis: Doctoral disser-tation, Oregon State University.

Bayley, P.B. & Dowling, D.C. 1993. The effects of habitat inbiasing fish abundance and species richness estimates whenusing various sampling methods in streams. Polskie Archi-wum Hydrobiologii 40: 5–14.

Benjamin, J.R., Wetzel, L.A., Martens, K.D., Larsen, K. &Connolly, P.J. 2014. Spatio-temporal variability in move-ment, age, and growth of mountain whitefish (Prosopiumwilliamsoni) in a river network based upon PIT tagging andotolith chemistry. Canadian Journal of Fisheries and AquaticSciences 71: 131–140.

Brenkman, S.J., Duda, J.J., Torgersen, C.E., Welty, E., Pess,G.R., Peters, R. & McHenry, M.L. 2012. A riverscape per-spective of pacific salmonids and aquatic habitats prior tolarge-scale dam removal in the Elwha River, Washington,USA. Fisheries Management and Ecology 19: 36–53.

Cunjack, R.A., Randall, R.G. & Chadwick, E.M.P. 1988.Snorkeling versus electrofishing: a comparison of censustechniques in Atlantic salmon rivers. Canadian Naturalist115: 89–93.

Davis, A.M. & Pusey, B.J. 2010. Trophic polymorphism andwater clarity in northern Australian Scortum (Pisces: Tera-pontidae). Ecology of Freshwater Fish 19: 638–643.

De Cillis, P. 1998. Fish habitat. Sitkum and South Fork Cala-wah watershed analysis. Olympia, WA: Olympic NationalForest.

Digby, P.G.N. & Kempton, R.A. 1987. Multivariate analysisof ecological communities. In: Usher, M.B. & Rosenzweig,M.L., eds. Population and community biology. New York:Chapman and Hall, pp. 149–175.

DosSantos, J.M. 1985. Comparative food habits and habitatselection of mountain whitefish and rainbow trout in theKootenai River, Montana. Bozeman: Master’s of Sciencethesis, Montana State University.

Edgar, G.J., Barret, N.S. & Morton, A.J. 2004. Biases associ-ated with the use of underwater visual census techniques toquantify the density and size-structure of fish populations.Journal of Experimental Marine Biology and Ecology 308:269–290.

ESRI. 2004. ArcGIS desktop: release 9.1. Redlands, CA:Environmental Systems Research Institute.

Evermann, B.W. 1893. A reconnaissance of the streams andlakes of western Montana and northwestern Wyoming. Bul-letin of the United States Fish Commission XI: 3–60.

Fausch, K.D. & White, R.J. 1981. Competition between brooktrout (Salvelinus fontinalis) and brown trout (Salmo trutta)for positions in a Michigan stream. Canadian Journal ofFisheries and Aquatic Sciences 38: 1220–1227.

Gibson, R.J. & Erkinaro, J. 2009. The influence of waterdepths and inter-specific interactions on cover responses ofjuvenile Atlantic salmon. Ecology of Freshwater Fish 18:629–639.

Hankin, D.G. & Reeves, G.H. 1988. Estimating total fishabundance and total habitat area in small streams based onvisual estimation methods. Canadian Journal of Fisheriesand Aquatic Sciences 45: 834–844.

Hastie, T.J. 1991. Generalized additive models. In: Chambers,J.M. & Hastie, T.J., eds. Statistical models in S. New York:Chapman and Hall, pp. 249–307.

Hastie, T.J. & Tibshirani, R. 1987. Generalized additive mod-els: some applications. Journal of the American StatisticalAssociation 82: 371–386.

Hastie, T.J. & Tibshirani, R. 1990. Generalized additive mod-els. Monographs on Statistics and Applied Probability 43.New York: Chapman and Hall.

Hook, A. 2004. WRIA 20: technical assessment level I waterquality and habitat. Forks: University of Washington Olym-pic Natural Resources Center.

Joyce, M.P. & Hubert, W.A. 2003. Snorkeling as an alterna-tive to depletion electrofishing for assessing cutthroat troutand brown trout in stream pools. Journal of Freshwater Ecol-ogy 18: 215–222.

Klingenberg, C.P., Barluenga, M. & Meyer, A. 2003. Bodyshape variation in cichlid fishes of the Amphilophus citrinel-lus species complex. Biological Journal of the Linnean Soci-ety 80: 397–408.

Lance, M.J. & Baxter, C.V. 2011. Abundance, production,and tissue composition of mountain whitefish (Prosopiumwilliamsoni) in a central Idaho wilderness stream. NorthwestScience 85: 445–454.

Legendre, P. 1993. Spatial autocorrelation: trouble or new par-adigm? Ecology 74: 1659–1673.

Legendre, P. & Legendre, L. 1998. Numerical ecology, 2ndEnglish edition. Amsterdam: Elsevier.

McCune, B. & Grace, J.B. 2002. Analysis of ecological com-munities. Gleneden Beach, OR: MJM Software Design.

McHugh, J.L. 1942. Growth of the Rocky Mountain white-fish. Journal of the Fisheries Resources Board of Canada 5:337–343.

McPhail, J.D. & Troffe, P.M. 1998. The mountain whitefish(Prosopium williamsoni): a potential indicator species forthe Fraser system. Vancouver: Report DOE FRAP 1998-16prepared for Environment Canada, Aquatic and AtmosphericSciences Division.

Meyer, K.A., Elle, F.S. & Lamansky, J.A. Jr 2009.Environmental factors related to the distribution, abundance,and life history characteristics of mountain whitefish inIdaho. North American Journal of Fisheries Management29: 753–767.

Montgomery, D.R. & Buffington, J.M. 1997. Channel-reachmorphology in mountain drainage basins. Geological Soci-ety of America Bulletin 109: 596–611.

Montgomery, D.R., Buffington, J.M., Smith, R.D., Schimdt,K.M. & Pess, G.R. 1995. Pool spacing in forest channels.Water Resources Research 31: 1097–1105.

Northcote, T.G. & Ennis, G.L. 1994. Mountain whitefish biol-ogy and habitat use in relation to compensation andimprovement possibilities. Reviews in Fisheries and Science2: 347–371.

Ostberg, C.O., Pavlov, S.D. & Hauser, L. 2009. Evolution-ary relationships among sympatric life history forms ofDally Varden inhabiting the landlocked Kronotsky Lake,Kamchatka, and the neighboring anadromous population.Transactions of the American Fisheries Society 138:1–14.

Palmer, M.W. 2002. Scale detection using semivariogramsand autocorrelograms. In: Gergel, S.E. & Turner, M.G.,eds. Learning landscape ecology: a practical guide toconcepts and techniques. New York: Springer, pp. 129–144.

13

Polymorphic mountain whitefish

Platts, W.S. 1979. Relationships among stream order, fishpopulations and aquatic geomorphology in an Idaho riverdrainage. Fisheries 4: 5–9.

Pontius, R.W. & Parker, M. 1973. Food habits of the moun-tain whitefish, Prosopium williamsoni (Girard). Transactionsof the American Fisheries Society 102: 764–773.

Power, M.E. 1984. Depth distribution of armored catfish:predator induced resource avoidance? Ecology 65: 523–528.

R Core Development Team. 2009. R: A language and envi-ronment for statistical computing. R Foundation for Statisti-cal Computing, Vienna, Austria. Available at: http://www.R-project.org/.

Robinson, B.W. & Wilson, D.S. 1994. Character release anddisplacement in fishes: a neglected literature. American Nat-uralist 144: 596–627.

Rodgers, J.D., Solazzi, M.F., Johnson, S.L. & Buckman, M.A.1992. Comparison of three techniques to estimate juvenilecoho salmon populations in small streams. North AmericanJournal of Fisheries Management 12: 79–86.

Roni, P., Beechie, T.J., Bilby, R.E., Leonettie, F.E., Pollock,M.M. & Pess, G.R. 2002. A review of stream restorationtechniques and a hierarchical strategy for prioritizing restora-tion in Pacific Northwest watersheds. North American Jour-nal of Fisheries Management 22: 1–20.

Sigler, W.F. 1951. The life history and management of Proso-pium williamsoni, in the Logan River, Utah. Utah StateAgricultural College, Bulletin 347, Logan. Also availableonline at: http://digitalcommons.usa.edu/uaes_bulletins/308.

Smith, C.J. 2000. Salmon and steelhead habitat limiting fac-tors in the north coastal streams of WRIA 20. Lacey, WA:Washington State Conservation Commission.

Smith, T.B. & Skulason, S. 1996. Evolutionary significance ofresource polymorphism in fishes, amphibians, and birds.Annual Review of Ecology and Systematics 27: 111–113.

SYSTAT. 2006. SigmaPlot, version 10.0. Richmond, CA:SYSTAT Software Inc.

Thompson, G.E. & Davies, R.W. 1976. Observations on theage, growth, reproduction, and feeding of mountain white-fish (Prosopium williamsoni) in the Sheep River, Alberta.Transactions of the American Fisheries Society 105: 209–219.

Thurow, R.F. 1994. Underwater methods for study of salmo-nids in the Intermountain West United States Forest Service.General Technical Report INT-GTR-307.

Torgersen, C.E., Baxter, C.V., Li, H.W. & McIntosh, B.A.2006. Landscape influences on longitudinal patterns of riverfishes: spatially continuous analysis of fish-habitat relation-ships. In: Hughes, R., Wang, L. & Wofford, J.E., eds. Influ-ences of landscapes on stream habitats and biologicalassemblages. Bethesda, MD: American Fisheries Society,pp. 473–492.

Trexler, J.C. & Travis, J. 1993. Nontraditional regressionanalysis. Ecology 74: 1629–1637.

Troffe, P.M. 2000. Fluvial mountain whitefish (Prosopiumwilliamsoni) in the Upper Fraser River: a morphological,behavioural, and genetic comparison of foraging forms.Vancouver: Masters of Sciences, The University of BritishColumbia. 68 pp.

USDA. 2009. Compressed county mosaic 1-m color ortho-photo for Clallam County. Salt Lake City, UT: USDA-FSAAerial Photography Field Office, FSA National AgricultureImagery Program (NAIP).

Welcomme, R.L. 1985. River fishes. FAO technical paper262. Rome, Italy: Food and Agriculture Organization of theUnited Nations.

Whiteley, A.R. 2007. Trophic polymorphism in a riverine fish:morphological, dietary and genetic analysis of mountainwhitefish. Biological Journal of the Linnean Society 92:253–267.

Whiteley, A.R., Spruel, P. & Allendorf, F.W. 2006. Can com-mon species provide valuable information for conservation?Molecular Ecology 15: 2767–2786.

Wimberger, P.H. 1994. Trophic polymorphisms, plasticity,and speciation in vertebrates. In: Stouder, D.J., Fresh, K.L.& Feller, R.J., eds. Advances in fish foraging theory andecology. Columbia, SC: University of South Carolina Press,pp. 19–43.

Wood, S.N. 2001. mgcv: GAMs and generalized ridge regres-sion for R. R-NEWS 1: 20–25.

Wood, S.N. 2003. Thin-plate regression splines. Journal of theRoyal Statistical Society: Series B (Statistical Methodology)65: 95–114.

Wood, S.N. 2006. Generalized additive models: an introduc-tion with R. New York: Chapman and Hall/CRC.

Wootton, R.J. 1998. Ecology of teleost fishes. London: Klu-wer Academic Publications.

Zar, J.H. 1984. Biostatistical analysis, 2nd edn. EnglewoodCliffs, NJ: Prentice Hall.

14

Starr & Torgersen