Landscape Influences on Longitudinal Patterns of River ...faculty.washington.edu/cet6/pub/Torgersen_etal_2006.pdf · Analysis of FishŒHabitat Relationships Christian E. Torgersen*,1

473

Landscape Influences on Longitudinal Patternsof River Fishes: Spatially ContinuousAnalysis of Fish�Habitat Relationships

Christian E. Torgersen*,1 and Colden V. Baxter2

Oregon Cooperative Fish and Wildlife Research Unit, Department of Fisheries and WildlifeOregon State University, Corvallis, Oregon 97331, USA

Hiram W. LiOregon Cooperative Fish and Wildlife Research Unit (U.S. Geological Survey)

Department of Fisheries and Wildlife, 104 Nash HallOregon State University, Corvallis, Oregon 97331, USA

Bruce A. McIntoshOregon Department of Fish and Wildlife, Corvallis Research Laboratory

28655 Highway 34, Corvallis, Oregon 97333, USA

Abstract.—Longitudinal analysis of the distribution and abundance of river fishes provides acontext-specific characterization of species responses to riverscape heterogeneity. We exam-ined spatially continuous longitudinal profiles (35–70 km) of fish distribution and aquatichabitat (channel gradient, depth, temperature, and water velocity) for three northeastern Ore-gon rivers. We evaluated spatial patterns of river fishes and habitat using multivariate analysisto compare gradients in fish assemblage structure among rivers and at multiple spatial scales.Spatial structuring of fish assemblages exhibited a generalized pattern of cold- and coolwaterfish assemblage zones but was variable within thermal zones, particularly in the warmest river.Landscape context (geographic setting and thermal condition) influenced the observed rela-tionship between species distribution and channel gradient. To evaluate the effect of spatialextent and geographical context on observed assemblage patterns and fish–habitat relation-ships, we performed multiple ordinations on subsets of our data from varying lengths of eachriver and compared gradients in assemblage structure within and among rivers. The relativeassociations of water temperature increased and channel morphology decreased as the spatialscale of analysis increased. The crossover point where both variables explained equal amountsof variation was useful for identifying transitions between cool- and coldwater fish assem-blages. Spatially continuous analysis of river fishes and their habitats revealed unexpected eco-logical patterns and provided a unique perspective on fish distribution that emphasized theimportance of habitat heterogeneity and spatial variability in fish–habitat relationships.

INTRODUCTION

Studies of river fish assemblages often focus ondescribing and understanding patterns in speciescomposition that occur along the length of riversystems. In general, fish assemblage structure isthought to change predictably from headwaters

*Corresponding author: [email protected] Present address: USGS-FRESC Cascadia Field Station,344 Bloedel Hall, University of Washington, Seattle,Washington 98195-2100, USA.2 Present address: Department of Biological Sciences,Idaho State University, Pocatello, Idaho 83209, USA.

American Fisheries Society Symposium 48:473–492, 2006© 2006 by the American Fisheries Society

22torgersen.p65 7/28/2006, 10:00 AM473

474 Torgersen et al.

to downstream reaches, with biotic zones (e.g.,cold- and warmwater assemblages) occurring inwhich species are added or replaced in responseto continuous gradients in temperature, chan-nel morphology, and water velocity (Huet 1959;Sheldon 1968; Horwitz 1978; Hughes andGammon 1987; Li et al. 1987; Rahel and Hubert1991; Paller 1994; Belliard et al. 1997). Bioticzonation and species addition are dominant,coarse-scale patterns that have been describedby numerous studies during the last century(Matthews 1998). Beyond these patterns, how-ever, there is little understanding of spatial het-erogeneity in fish distribution and habitatrelationships within biotic zones or the effectsof spatial scale and context on observed fish as-semblage patterns in rivers (Collares-Pereira etal. 1995; Duncan and Kubecka 1996; Poizat andPont 1996; Bult et al. 1998; Fausch et al. 2002).Consequently, there are likely many more pat-terns and spatial relationships that have yet tobe described, and these may be essential to un-derstanding river fish assemblages.

Discovery of new patterns and spatial rela-tionships may be constrained by repeated use oftraditional study approaches. Relatively shortsampling reaches (<500 m) spaced at wide in-tervals (>10 km) along the longitudinal profileor throughout a channel network may providethe information necessary to detect coarse gra-dients in fish assemblage structure associatedwith factors such as temperature and stream or-der (Vannote et al. 1980). However, such site-based studies lack the spatial resolution necessaryfor detecting patterns in fish–habitat relation-ships across a range of spatial scales. Conse-quently, the perception that river fishassemblages change gradually with respect tolongitudinal habitat gradients may be drivenlargely by the resolution and extent of data col-lection and analysis (Naiman et al. 1988; Wiens1989; Poole 2002). As a consequence of the dis-continuous and spatially limited manner inwhich river fishes are traditionally sampled, fun-damental questions about the nature and extent

of spatial variability in river fish–habitat relation-ships remain unanswered: How finely tuned arelongitudinal patterns in fish assemblages to keyhabitat factors such as thermal heterogeneity,channel morphology, and velocity? Can the ef-fects of temperature on fish assemblages be iso-lated from the effects of other factors? How areassemblage patterns at one scale mediated bycontext at larger spatial scales? Does perceptionof habitat relationships change with the spatialextent of a study? We propose that these ques-tions can be addressed only by adapting andchanging the manner in which fish assemblageand habitat data are collected and analyzed.

Here we illustrate a new approach to collect-ing and analyzing fish assemblage and habitatdata that provides a more spatially continuousview of fishes and the riverine landscapes, or“riverscapes,” they inhabit (Fausch et al. 2002).Our objectives were to (1) collect spatially con-tinuous data on fish assemblage structure andhabitat along the length of three rivers with con-trasting physical environments, (2) characterizeand compare longitudinal patterns and habitatrelationships (water depth, velocity, channel gra-dient, and water temperature) among and withinthese riverscapes, and (3) evaluate the effect ofspatial extent and geographical context of sur-vey data on observed fish–habitat relationships.

METHODS

Study Area

We studied fish assemblages in three small riv-ers in the Blue Mountains of northeastern Ore-gon: the Middle Fork John Day (MFJD; upper49 km), the North Fork John Day (NFJD; upper70 km), and the Wenaha River (WEN; lower 35km; Figure 1). Study section elevations rangedfrom 500 m in the lower WEN to 1,700 m in theupper NFJD and shared a similar geology ofColumbia River basalt at lower elevations andfolded metamorphosed rocks partially overlainby volcanic tuff in headwater reaches (Orr et al.

22torgersen.p65 7/28/2006, 10:00 AM474

Landscape Influences on Longitudinal Patterns of River Fishes 475

Figure 1. Study area and river sections surveyed for fish assemblages in northeastern Oregon. Study riversincluded (A) the Middle Fork John Day (MFJD), (B) the North Fork John Day (NFJD), and (C) the Wenaha River(WEN). Black dots indicate the spatial extent and continuity of underwater visual surveys.

22torgersen.p65 7/28/2006, 10:00 AM475


1992). Although the NFJD study section had thelargest drainage area and the highest elevations,the WEN received more annual precipitation andhad higher summer base flow (Table 1). Longi-tudinal gradients in elevation and annual pre-cipitation were steepest in the WEN, followedby the NFJD and the MFJD. Maximum summerwater temperature patterns reflected differencesin streamflow among basins and represented arange of cool and cold thermal environments(Table 1).

Seasonal weather patterns throughout thestudy area are typical of high desert climates withhot, dry summers and cold, relatively wet win-ters (–15–38°C; Loy et al. 2001). The Blue Moun-tains ecoregion is characterized by contrasts intemperature, precipitation, and vegetation cor-responding with steep elevation gradients(Clarke and Bryce 1997). Canyons and alluvialvalleys in the Wenaha and John Day River ba-sins are vegetated with mixed conifer forest (pon-derosa pine Pinus ponderosa, grand fir Abiesgrandis, Douglas-fir Pseudotsuga menziesii, west-ern larch Larix occidentalis, and lodgepole pinePinus contorta) on the upslopes and broadleafassemblages of black cottonwood Populustrichocarpa, willow Salix spp., and red alder Alnusrubra in the valley bottoms. The upper NFJD andthe WEN are designated wild and scenic riverssituated within public wilderness areas, whereasthe MFJD flows mainly through private cattle

ranches. Land-use impacts are minimal in therelatively pristine WEN compared to the NFJDand the MFJD, which have experienced exten-sive mining, grazing, and logging during the lastcentury.

Fish Assemblages

Native fish species common in the study riversincluded four salmonids, three catostomids, fourcyprinids, and two cottids. Two nonnative fishes(brook trout Salvelinus fontinalis and small-mouth bass Micropterus dolomieu) were ex-tremely rare and therefore not included in ouranalysis. We selected a subset of species for as-semblage analysis based on their relative abun-dance and ease of identification underwater(Figure 2). We noted sculpins Cottus spp.,longnose dace Rhinichthys cataractae, and moun-tain sucker Catostomus platyrhynchus duringsurveys but did not include them in analysis be-cause they were difficult to detect and identifyunderwater, as determined by comparisons ofsnorkeling and electrofishing in selected sectionsof the MFJD (H. W. Li, unpublished data).

Fish assemblage zones overlapped in each ofthe study rivers and provided an excellent op-portunity to evaluate patterns in assemblagestructure in relation to water temperature andchannel morphometry. Cold- and coolwatertemperature classifications were based on species

Table 1. Physical characteristics of study sections in the Middle Fork John Day (MFJD), the North Fork JohnDay (NFJD), and the Wenaha (WEN) rivers.

River Drainage Summer Waterkilometer area Elevation Stream Precipitation base flow temperature

River (rkm)a (km2)b (m) orderc (cm/year) (m3/s)d (°C)e

MFJD 62�117 1,000 1,000�1,300 4th�5th 35�60 1.4 21.1�25.2NFJD 95�165 1,600 800�1,700 4th�5th 50�90 2.8 19.1�25.0WEN 0�35 750 500�1,100 4th�5th 50�150 5.7 15.1�21.3a Distance upstream from mouth.b Drainage area at lower boundary of study section.c Range in stream order between upper and lower boundaries of study section (determined from 1:100,000-scale U.S. Geological Surveytopographic maps).d Streamflow estimates are approximations of summer low-flow conditions based on field measurements in late August and September 1997�1999.e Range in mean maximum water temperature on 1�7 August 1998 at the upstream and downstream boundaries of the study sections.

22torgersen.p65 7/28/2006, 10:00 AM476


ranges, spawning seasons, spawning tempera-tures, and physiological optima (Zaroban et al.1999). Coldwater species included only thesalmonids, whereas coolwater species comprisedboth catostomids and cyprinids.

Longitudinal Surveys of Fish Distributionand Aquatic Habitat

We conducted extensive snorkel surveys to quan-tify longitudinal patterns in river fish assemblagesduring summer low-flow conditions in July–

August 1996 (MFJD), 1997 (NFJD), and 1998(WEN). Underwater snorkel surveys provideaccurate assessments of fish abundance in flow-ing waters and offer an alternative to electro-fishing when it is restricted by managementagencies or when rivers are too large to sampleeffectively with a backpack electrofisher and toosmall to sample by boat (Cunjak et al. 1988;Zubik and Fraley 1988; Thurow and Schill 1996;Mullner et al. 1998; Joyce and Hubert 2003). Weevaluated the distribution and abundance ofriver fishes using a modified version of point

Cyprinidae Catostomidae

Salmonidae

1

9

8

76

5

43

2

1. Oncorhynchus mykiss rainbow trout (R T)2. Salvelinus confluentus bull trout (BT)3. Oncorhynchus tshawytscha juvenile Chinook salmon (JCS)4. Prosopium williamsoni mountain whitefish (MW)5. Ptychocheilus oregonensis northern pikeminnow (NP)6. Richardsonius balteatus redside shiner (RS)7. Rhinichthys osculus speckled dace (SD)8. Catostomus macrocheilus largescale sucker (LS)9. Catostomus columbianus bridgelip sucker (BS)

SpeciesID Common name

Figure 2. River fish assemblage surveyed in northeastern Oregon. Benthic fish species, including longnosedace Rhinichthys cataractae, mountain sucker Catostomus platyrhynchus, torrent sculpin Cottus rhotheus, andPaiute sculpin C. beldingii, were noted during surveys but not included in analyses. The species code used insubsequent figures is listed after the common name.

22torgersen.p65 7/28/2006, 10:00 AM477


abundance sampling (Persat and Copp 1990).The objective of modified point abundance sam-pling was to collect large numbers of closelyspaced samples (<100 m separation), providinga relatively continuous assessment of fish distri-bution (Figure 1). Although the size of the studyrivers prevented us from estimating the efficiencyof our sampling procedure (e.g., via compari-son with estimates from multiple-passelectrofishing), other work conducted in riversof similar size has shown that visual estimatesprovide an accurate (though perhaps imprecise)assessment of fish distribution (H. W. Li and P.B. Bayley, Oregon State University, unpublisheddata). Three short gaps (2–3 km) in the exten-sive surveys of the MFJD and the NFJD occurredwhere access was denied to private lands or wheresteep canyons and rapids made sampling toodangerous. We divided survey sections intoreaches of equal length and sampled fishes andhabitat with two-person crews consisting of adiver and a data recorder walking along the shore.Divers counted fish in two or more passes nearshore and mid-channel in an upstream or down-stream direction depending upon water depthand velocity. Using this approach, a diver–recorder crew was capable of surveying an aver-age of 2–4 km per day.

Divers recorded fish abundances in categoriesindicating whether a species was dominant(>50%), common (10–50%), or rare (<10%) inrelation to the total number of fish observed ina sample unit. Relative abundance provided in-formation on the composition but not the abso-lute abundance of fish species in a given channelunit. Relative abundances were representative ofthe proportion of all fish of all species estimatedto be in a sampled channel unit. This measureof abundance is particularly useful for determin-ing the ecological relationships among fish spe-cies (Rahel 1990; Rahel and Hubert 1991;Reynolds et al. 2003). In all cases, the divers werehighly experienced in fish identification andevaluated their estimates of fish abundance regu-larly through repeat dives of the same channelunit by different divers. In addition to collecting

data on fish assemblages, field crews collectedinformation on channel morphology (e.g., sidechannel/main channel, depth, width, velocity)and water temperature and recorded geographiccoordinates (±100 m) of individual sample unitswith a handheld global positioning system(GPS). Field crews placed slow- and fast-waterhabitats in four categories corresponding to wa-ter velocity: (1) pools, (2) slow-moving glides,(3) fast-moving glides, and (4) riffles (Bisson etal. 1982). Categorical estimates of current veloc-ity explained 68% of the variation in currentvelocity measured with a flowmeter (n = 33, P <0.001, y = 0.39 + 0.26x + 0.08x2).

Geographical Analysisand Remote Sensing

A geographical information system (GIS) wasessential for mapping, displaying, and analyzingthe large number of sample points required toassess spatial patterns in aquatic habitat and fishdistribution (Figure 1). We mapped sampledchannel units as individual points linked to adatabase containing information on fish abun-dance and habitat characteristics. Longitudinalanalysis was accomplished using route and dy-namic segmentation procedures in ARC/INFOGIS (ESRI 1996; Radko 1997). We derived digi-tal hydrography layers from 1:5,000-scale aerialphotographs (MFJD) and 1:100,000-scale topo-graphic maps (NFJD and WEN). Route-measurecoordinates, defined as the distance upstreamfrom the mouth (i.e., river kilometer, rkm), servedas a common axis with which to compare longi-tudinal profiles of fish distribution and aquatichabitat. We generated a channel gradient profilefrom a 10-m digital elevation model (DEM) bysampling elevation every 100 m along the riverchannel and then calculating gradient using a500-m moving window. Overlays of evenly spacedsample points on the spatially continuous chan-nel gradient profile provided a coarse estimateof gradient that was consistent among channelunits. Those units varied in length and generallydecreased in size in an upstream direction.

22torgersen.p65 7/28/2006, 10:00 AM478


We assessed spatially and temporally continu-ous patterns in water temperature with airbornethermal infrared (TIR) remote sensing and au-tomated instream thermographs (Torgersen etal. 2001). Aerial surveys occurred on cloudlessdays 4–9 August 1998 at 1300–1400 hours. Ther-mographs served as ground-truth points for TIRremote sensing and provided temporal data nec-essary for comparing relative difference in meanand maximum water temperatures within andamong basins.

Data Analysis

To evaluate spatial patterns and associations infish distribution, channel morphology, watervelocity, and temperature, we compared peaksand troughs in fish abundance to longitudinalprofiles of habitat. We scaled the relative abun-dance estimates for fish (dominant, common,and rare) and categorical estimates of water ve-locity to 1.0, and then plotted fish and habitatvariables versus distance upstream from the rivermouth. To identify spatial trends in longitudinalprofiles, we used locally weighted scatterplotsmoothing (LOWESS), a robust, nonparamet-ric regression technique used to identify trendsin heterogeneous ecological data (Trexler andTravis 1993). Locally weighted regression calcu-lations used a second-degree polynomialsmoothing function in SigmaPlot statistical soft-ware (SPSS 2001). The objective of graphicalanalysis with LOWESS was to explore spatiallycontinuous patterns and evaluate the extent towhich longitudinal changes in species distribu-tion and habitat were gradual or abrupt in con-trasting riverine environments.

Multivariate analysis was necessary to distin-guish patterns in fish assemblage structure bothwithin and among rivers. Standard parametricmultivariate methods (e.g., principal compo-nents analysis, detrended correspondence analy-sis, and canonical correspondence analysis) arecommonly applied in studies of river fishes be-cause they reduce complex species matrices intotwo or more dimensions, or axes, representing

gradients in assemblage structure (Hughes andGammon 1987; Rahel and Hubert 1991; Paller1994; Taylor et al. 1996). However, these methodsare not appropriate for analyzing nonnormallydistributed data sets, such as the spatially con-tinuous fish assemblage and habitat data col-lected in this study (McCune 1997). Therefore,we computed multivariate ordinations withnonmetric multidimensional scaling (NMS) inPC-ORD, a software package specifically de-signed for multivariate analysis of ecological data(McCune and Mefford 1999). Nonmetric mul-tidimensional scaling is a nonparametric proce-dure that calculates axis scores based on rankeddistances and therefore alleviates the problemsof zero truncation caused by heterogeneous eco-logical data sets (Clarke 1993; Tabachnick andFidell 2001).

We calculated two-dimensional solutions inNMS using the Sørensen distance measure and15 runs of real data with up to 200 iterations toevaluate stability. Because of the extremely largesample size, only 30 Monte Carlo runs were suf-ficient to evaluate the probability (� = 0.05) thatordination axes explained more variation thanwould be expected by chance. To identify envi-ronmental gradients associated with ordinationaxes, we constructed joint plots and biplots(Jongman et al. 1995) of samples and species inordination space and examined Pearson corre-lations between variables in a habitat matrix(mean depth, maximum depth, water velocity,channel gradient, and water temperature) andordination axis scores. The statistical significanceof Pearson correlations provided a relative meansto compare correlation strength among habitatvariables and ordination axes rather than to testspecific hypotheses (McCune and Grace 2002).To facilitate interpretation of the ordinations, werotated the point cluster around the centroid toalign habitat variable vectors in the joint plotswith the primary and secondary ordination axes.We produced ordinations with fish species andsamples plotted in ordination space by calculat-ing species scores with weighted averaging. Wethen labeled the primary and secondary gradients

22torgersen.p65 7/28/2006, 10:00 AM479


in fish assemblage structure (ordination axes 1 and2, respectively) according to the two habitat vari-ables with which they were most highly correlated.

To evaluate the effect of spatial extent andgeographical context on observed assemblagepatterns and fish–habitat relationships, we per-formed multiple ordinations on subsets of ourdata from varying lengths of each river and com-pared gradients in assemblage structure withinand among rivers. We divided each river sectioninto 10 reaches of varying lengths (e.g., rkm 0–70, rkm 0–65, rkm 0–60, etc.) and performed aseparate ordination for each reach. This processis essentially a scaling analysis that quantifies theeffects of spatial extent and geographic contexton assemblage composition. Specifically, by com-paring Pearson correlations of environmentalvariables with ordination axis scores along thelongitudinal profile, we were able to examine thecombined effects of spatial extent (i.e., reachlength) and geographic context on the observedrelative influences of habitat on fish assemblagestructure.

RESULTS

Longitudinal Patterns ofIndividual Fish Species

Longitudinal patterns in fish distribution weregradual for some species and abrupt for others,and differed markedly among rivers. In theMFJD, patterns were driven by differences in thedistribution of juvenile Chinook salmonOncorhynchus tshawytscha and rainbow trout O.mykiss versus mountain whitefish Prosopiumwilliamsoni, catostomids, and cyprinids (Figure3A). Juvenile Chinook salmon and rainbow troutwere both relatively abundant in the middle sec-tion of the river (rkm 85–100). Rainbow troutincreased in relative abundance upstream of rkm105, whereas juvenile Chinook salmon were mostcommon in a single reach downstream of rkm100. Mountain whitefish were relatively abun-dant downstream of the reaches with high rela-tive abundances of Chinook salmon and rainbow

trout (rkm 83–88). Catostomids (bridgelipsucker Catostomus columbianus and largescalesucker C. macrocheilus) had different patterns ofrelative abundance depending on the species.Peaks in the relative abundance of bridgelipsucker occurred downstream of rkm 65, at rkm85, and upstream of rkm 110, and largescalesucker were relatively abundant at rkm 72 andrkm 97. Speckled dace Rhinichthys osculus andredside shiner Richardsonius balteatus were com-mon throughout the MFJD but exhibited localpeaks in relative abundance at rkm 83 (both spe-cies) and peaks and troughs, respectively, at rkm101. Northern pikeminnow Ptychocheilusoregonensis were relatively rare in the MFJD ex-cept in reaches downstream of rkm 65 and atrkm 107–112.

Fish distribution in the NFJD exhibited dis-tinct peaks and troughs in the relative abundanceof juvenile Chinook salmon and mountainwhitefish but was more gradual for rainbowtrout, bull trout Salvelinus confluentus, largescalesucker, bridgelip sucker, speckled dace, redsideshiner, and northern pikeminnow (Figure 3B).Rainbow trout increased in abundance gradu-ally in an upstream direction but were rare inthe uppermost reaches of the NFJD. Bull troutincreased in abundance gradually in an upstreamdirection from the lowermost occurrence at rkm150. Bull trout and coolwater species (catosto-mids and cyprinids) did not overlap spatially.

Salmonids dominated the fish assemblage inthe WEN (Figure 3C). Juvenile Chinook salmonwere relatively abundant throughout the studysection but were most abundant in the middlereaches of the WEN (rkm 13–25). Rainbow troutwere common throughout the study section butincreased in relative abundance in downstreamreaches (rkm 0–8) and in the uppermost reach ofthe study section (rkm 33–35). Relative abun-dances of bull trout and juvenile Chinook salmonincreased gradually in an upstream direction andreached a peak at rkm 20–23. Mountain white-fish were relatively abundant throughout the lower23 km of the WEN but decreased dramatically inrelative abundance upstream of rkm 23. Largescale

22torgersen.p65 7/28/2006, 10:00 AM480


sucker and northern pikeminnow occurredthroughout the lower 23 km of the WEN but rep-resented a relatively small part of the fish assem-blage, except in the lower reaches (rkm 0–3) wherelargescale sucker were nearly as common asmountain whitefish and rainbow trout.

Associations between Fish Species andLongitudinal Patterns of Aquatic Habitat

Longitudinal patterns of fish distribution corre-sponded with patterns in aquatic habitat, butthese associations were nonlinear and complex

Figure 3. Longitudinal patterns of fish distribution in (A) the Middle Fork John Day (MFJD), (B) the North ForkJohn Day (NFJD), and (C) the Wenaha River (WEN). Trend lines are smoothed values from locally weightedscatterplot smoothing (LOWESS) of near-continuous fish survey data. Dashed horizontal bars below each trendline depict the spatial continuity of fish surveys and provide a relative indicator of the number of data pointsused to calculate LOWESS regressions. Relative abundance represents a continuum of rare to dominant on ascale of 0 to 1; panels are separated and scaled differently to clarify individual species-abundance patterns.See Figure 2 for definitions of species codes.

AAAAA BBBBB

CCCCC

22torgersen.p65 7/28/2006, 10:00 AM481


(Figures 3 and 4). In the MFJD, peaks in the rela-tive abundance of juvenile Chinook salmon wereassociated with peaks in channel gradient andtroughs in water temperature (Figures 3A and

4A). Peaks in the relative abundance of moun-tain whitefish, bridgelip sucker, speckled dace,and redside shiner corresponded with the high-est peak in maximum depth at rkm 83 (Figures

AAAAA BBBBB

CCCCC

Figure 4. Longitudinal patterns of aquatic habitat in (A) the Middle Fork John Day (MFJD), (B) the North ForkJohn Day (NFJD), and (C) the Wenaha River (WEN). Trend lines are smoothed values from locally weightedscatterplot smoothing (LOWESS) of spatially continuous (channel gradient and temperature) and near-continu-ous survey data (depth and water velocity). Dashed horizontal bars below each trend line depict the spatialcontinuity of habitat surveys and provide a relative indicator of the number of data points used to calculateLOWESS regressions. Water velocity is scaled to 1.0 and represents a continuum of slow- to fast-water aquatichabitats. Mean daily water temperatures were recorded on the day of synoptic surveys with thermal infraredremote sensing.

22torgersen.p65 7/28/2006, 10:00 AM482


3A and 4A). In the NFJD, spatial associationsbetween fish distribution and channel morphol-ogy and water temperature were not as pro-nounced as they were in the MFJD (Figures 3A,3B, 4A, and 4B). The two highest peaks in therelative abundance of juvenile Chinook salmoncorresponded with the highest peak (rkm 120)and the lowest trough in water velocity (rkm150). Peaks in the relative abundance of moun-tain whitefish corresponded with peaks in maxi-mum depth. In the WEN, peaks in the relativeabundance of rainbow trout were associated withhigh-velocity downstream reaches (rkm 0–8)and high-gradient reaches upstream (rkm 33–35) (Figures 3C and 4C). Peaks in the relativeabundance of bull trout and juvenile Chinooksalmon coincided with a peak in maximum wa-ter depth and a trough in water velocity.

Multivariate Gradients in Fish AssemblageStructure and Aquatic Habitat

Fishes exhibited distinct differences in assem-blage structure with respect to habitat variablesin the three rivers. Variation in fish assemblagecomposition in the MFJD corresponded withhabitat gradients in depth, water velocity, chan-nel gradient, and, to a lesser degree, water tem-perature (Figure 5A and Table 2). The primaryordination axis (depth and water velocity) ex-plained 73% of the variation in fish assemblagestructure, and the secondary axis (channel gra-dient and water temperature) explained 19% ofthe variation (P < 0.05). Fishes were strongly seg-regated among shallow riffles (rainbow trout,juvenile Chinook salmon, mountain whitefish,and speckled dace) and deep pools (redsideshiner, bridgelip sucker, northern pikeminnow,and largescale sucker). Fish species most stronglycorrelated with the primary axis (depth and wa-ter velocity) included bridgelip sucker, northernpikeminnow, largescale sucker, redside shiner,and rainbow trout. Species strongly associatedwith the second axis (channel gradient and tem-perature) included juvenile Chinook salmon,rainbow trout, and northern pikeminnow (Table

Axis 1 (73%)

Highgradient

Low gradient

Shallow Fast

Deep Slow

MFJD

Axi

s 2

(19%

)

Cool

Warm

Axis 1 (57%)

Axi

s 2

(26%

)

Warm Low gradient

ColdHigh gradient

NFJD

Shallow

Deep Slow

Fast

Axis 1 (46%)

Axi

s 2

(43%

)

Cool Low gradient

ColdHigh gradient

WEN

Shallow Fast

Deep Slow

Figure 5. Ordination of nonmetric multidimensionalscaling (NMS) analysis of fish assemblage structure in(A) the Middle Fork John Day (MFJD), (B) the NorthFork John Day (NFJD), and (C) the Wenaha River(WEN). Fish species are plotted in ordination space,in which each fish outline indicates the position of thespecies� centroid with respect to the ordination axes.Solid triangles are sample units in species space. Theamount of variation explained by each ordination axisis shown in parentheses. See Figure 3 for a key to thefishes.

AAAAA

BBBBB

CCCCC

22torgersen.p65 7/28/2006, 10:00 AM483


2). Mountain whitefish and speckled dace occu-pied intermediate positions.

Fish assemblages in the NFJD were structuredalong gradients of water temperature, channelgradient, and depth (Figure 5B). Temperature andchannel gradient were strongly correlated with theprimary axis, which explained 57% of the varia-tion in fish assemblage structure (P < 0.05) (Fig-ure 5B and Table 2). The distribution of fishspecies with respect to the primary ordinationaxis (temperature and channel gradient) indi-cated a separation between coolwater fishes(redside shiner, largescale sucker, bridgelipsucker, northern pikeminnow, and speckleddace) and rainbow trout and bull trout. JuvenileChinook salmon and mountain whitefish werepositioned at an intermediate location with re-spect to the primary ordination axis (tempera-ture and channel gradient). Fish species moststrongly correlated with the primary axis in-

cluded rainbow trout, speckled dace, and redsideshiner (Table 2). The secondary axis explained26% of the variation in fish assemblage structure(P < 0.05) and was associated primarily with meanand maximum water depth (Table 2). With theexception of bull trout and rainbow trout, fishesin the NFJD were generally grouped into deep-water (mountain whitefish, largescale sucker,juvenile Chinook salmon, and northern pike-minnow) and shallow-water (bridgelip sucker,speckled dace, and redside shiner) assemblages.

Fishes in the WEN responded to gradients intemperature, channel gradient, depth, and wa-ter velocity (Figure 5C, Table 2). Coldwater fishes(juvenile Chinook salmon, bull trout, and rain-bow trout) were most abundant in colder, up-stream reaches, while coolwater fishes (northernpikeminnow and largescale sucker) were mostcommon downstream. Mountain whitefish,largescale sucker, and juvenile Chinook salmon

Table 2. Pearson correlation coefficients of species and habitat variables versus axis scores from ordinations offish assemblage structure in entire survey reaches. The surveyed lengths in the Middle Fork John Day, the NorthFork John Day, and the Wenaha rivers are 49, 70, and 35 km, respectively. Ordinations were calculated fromrelative abundance data using nonmetric multidimensional scaling (NMS). The statistical significance of corre-lations between axis scores and habitat variables, indicated with one or two asterisk symbols (P < 0.05 or P <0.001, respectively), provides a relative means to compare correlation strength among variables and ordina-tion axes.

Middle Fork North ForkJohn Day River John Day River Wenaha River

(n = 261) (n = 244) (n = 179)

Variable Axis 1 Axis 2 Axis 1 Axis 2 Axis 1 Axis 2

SpeciesBull trout � � 0.30 �0.11 0.22 0.31Juvenile Chinook salmon �0.12 0.87 �0.24 0.29 0.50 0.76Rainbow trout �0.69 0.79 0.79 �0.18 0.24 �0.46Mountain whitefish �0.15 0.11 �0.29 0.71 �0.86 0.40Northern pikeminnow 0.92 �0.50 �0.42 0.06 �0.29 0.31Largescale sucker 0.76 0.05 �0.44 0.09 �0.50 0.19Bridgelip sucker 0.95 �0.39 �0.40 �0.30 � �Redside shiner 0.74 �0.37 �0.54 �0.23 � �Speckled dace �0.51 �0.40 �0.78 �0.42 � �

HabitatTemperature 0.28** �0.13* �0.76** 0.00 �0.65** �0.04Channel gradient �0.16* 0.32** 0.57** �0.02 0.35** �0.07Maximum depth 0.40** �0.12* �0.22** 0.30** �0.18* 0.35**Mean depth 0.32** -0.05 �0.18** 0.28** �0.23** 0.29**Water velocitya �0.38** 0.04 0.14 �0.22** 0.11 �0.28**a Water velocity is a categorical variable that represents a continuum of slow- to fast-water aquatic habitats.

22torgersen.p65 7/28/2006, 10:00 AM484


were strongly correlated with ordination scoreson the primary axis, which explained 46% of thevariation in the ordination (P < 0.05) (Figure5C and Table 2). On the secondary ordinationaxis, depth and water velocity explained 43% ofthe variation in fish assemblage structure (P <0.05). Fish species most strongly associated withthe secondary axis included juvenile Chinooksalmon, rainbow trout, and mountain whitefish(Table 2). Of the three coldwater fishes in theWEN, juvenile Chinook salmon exhibited thestrongest positive association with the second-ary ordination axis (water depth and slow-waterhabitats) (Table 2).

Effects of Spatial Extent andGeographical Context on Observed

Fish�Habitat Relationships

The observed relationships between fish assem-blage structure and aquatic habitat changed de-pending on the spatial extent and geographicalcontext of analysis (Figure 6). In the MFJD andthe NFJD, the relative influences of temperatureand channel morphology increased and de-creased, respectively, when the spatial extent ofthe data set was increased. Crossover points inthe trends of correlations indicated where tem-perature and channel morphology explainedapproximately equal amounts of variation in fishassemblage structure (Figure 6). In the MFJD,the warmest of the rivers, the crossover pointoccurred in the upper portion of the study sec-tion (rkm 48), whereas in the NFJD, this transi-tion occurred in the lower 20 km of the studysection (rkm 115). In the WEN, the coldest andalso the shortest river, the relative influence ofchannel morphology on fish assemblage struc-ture increased as the spatial extent of the dataset was increased. In the WEN, there was no con-sistent spatial trend in the relative influence oftemperature on fish assemblage structure, andthere was no crossover point at which watertemperature and channel morphology explainedequal amounts of variation in fish assemblagestructure.

DISCUSSION

Longitudinal patterns of river fish in the MFJD,NFJD, and WEN ranged from gradual to abruptand differed substantially among species andamong rivers. At the scale of entire river sections

Figure 6. Scale-dependent effects of temperature andchannel morphology on river fish assemblage struc-ture in the Middle Fork John Day (MFJD), the NorthFork John Day (NFJD), and the Wenaha River (WEN).Pearson correlation coefficients (r) indicate the rela-tive influence of temperature and channel morphol-ogy on river fish assemblage structure over a range ofspatial extents. Crossover points indicate the spatialextent and location where temperature and channelmorphology explained approximately equal amountsof variation in river fish assemblage structure.

22torgersen.p65 7/28/2006, 10:00 AM485


(30–70 km), longitudinal patterns of water tem-perature and channel gradient corresponded tozonation from a coldwater assemblage (Salmon-idae) to a coolwater minnow–sucker assemblage(Cyprinidae–Catostomidae) as predicted by theriver continuum concept and current models ofriver fish distribution (Vannote et al. 1980; Li etal. 1987; Rahel and Hubert 1991). However, em-bedded within the broad-scale template of cool-and coldwater fish assemblage zones, the distri-bution of fishes was highly variable and reflectedreach-scale variation in channel morphology(i.e., depth and water velocity) and water tem-perature. Fish assemblage structure was particu-larly variable along the length of the MFJD wherethe longitudinal thermal gradient was not so pro-nounced as in the NFJD and the WEN. Althoughother studies have suggested a high degree ofspatial heterogeneity in river fish distribution atspatial extents of 30–70 km (Stewart et al. 1992;Roper and Scarnecchia 1994), few studies havedescribed such patterns with spatially continu-ous data (Baxter 2002; Torgersen 2002). Thisstudy provides an example of how spatially con-tinuous data can be collected and analyzed toevaluate landscape influences on longitudinalpatterns of cool- and coldwater fish assemblagesin three Pacific Northwest rivers.

Effects of Landscape Context onFish�Habitat Relationships

Longitudinal patterns of species and assem-blages.—Detailed studies of the spatial distribu-tion of fishes within entire river sections (10–100km) are useful for evaluating how fish–habitatrelationships change across scales and in differ-ent spatial contexts (Fausch et al. 1994). Riverfish responses to channel gradient provide a casein point. Within a given river basin, rainbowtrout/steelhead (anadromous rainbow trout) aregenerally associated with relatively steeper,swifter habitats than juvenile Chinook salmon(McMichael and Pearsons 1998; Montgomery etal. 1999). Similarly, salmonids are usually asso-ciated with higher channel gradient than

warmwater fishes in North America (Rahel andHubert 1991), but this relationship is difficult tointerpret because elevation, channel gradient,and coldwater temperatures are often closelycorrelated (Isaak and Hubert 2000; Isaak andHubert 2001). However, we were able to isolatethe influence of channel gradient on species dis-tribution from that of stream temperature. Ourobservations of fish distribution in the MFJD andNFJD provided a unique opportunity to evalu-ate the response of a coldwater fish (juvenileChinook salmon) to relatively high-gradientreaches over a range of water temperatures. Forexample, we found that the relative abundancepatterns of juvenile Chinook salmon in thewarmer MFJD and lower NFJD were positivelyassociated with relatively high-gradient reaches.In contrast, in the upper, colder section of theNFJD, a peak in the relative abundance of juve-nile Chinook salmon corresponded with a localtrough in channel gradient. Thus, landscape con-text (i.e., geographic and thermal conditions)reversed the observed relationship between thisspecies and channel gradient. This effect oflandscape context was also observed in patternsof fish assemblage structure. Comparisons be-tween assemblage structure in the MFJD andthe WEN indicated that juvenile Chinooksalmon were associated with the shallow, fast-water fish assemblage at warm temperatures(MFJD) but were more associated with thedeep, slow-water fish assemblage at cold tem-peratures (WEN).

Potential ecological mechanisms.—The poten-tial ecological mechanisms underlying thischanging habitat relationship require furtherinvestigation. The apparent reversal in habitatselection by juvenile Chinook salmon may berelated to species interactions and bioenerget-ics. For example, coolwater fish species, such asredside shiner and northern pikeminnow, havea physiological advantage over coldwater fishesin the relatively warmer sections of the MFJDand lower NFJD. Coolwater fishes were the mostabundant species in deep, slow-water habitats inthe MFJD and the lower NFJD and may exclude

22torgersen.p65 7/28/2006, 10:00 AM486


juvenile Chinook salmon through competitionwith redside shiner (Reeves et al. 1987) orthrough predation by northern pikeminnow(Isaak and Bjornn 1996). Increased riffle use bysalmonids has also been shown to occur in re-sponse to higher metabolic demands at warmerwater temperatures (Smith and Li 1983) becausefaster current velocities provide higher inverte-brate drift rates and may actually balance out theincreased metabolic costs of maintaining a po-sition in faster current.

The heterogeneous distribution of river fisheswe observed in these watersheds may also be in-fluenced by historical constraints on distribution(geomorphology and biogeographic history),land-use history, and the spawning distributionof adults (Harding et al. 1998; Williams et al. 2003).All three factors likely play roles in the assemblagepatterns of the study rivers, particularly in theMFJD and NFJD, which have complex historiesof land use and have undergone considerablechannel restructuring (Torgersen et al. 1999).

Spatial Scale of Observation andGradients in Fish Assemblage Structure

The spatial scale of analysis influenced the ob-served gradients in fish assemblage structure. Anumber of studies have evaluated changing habi-tat relationships of individual fish species acrossmultiple spatial scales (Fausch et al. 1994; Poizatand Pont 1996; Torgersen et al. 1999; Baxter andHauer 2000; Thompson et al. 2001; Torgersen andClose 2004). However, investigations of the effectsof scale on observed patterns of species diversityand assemblage structure are less common (Wil-son et al. 1999). This is largely due to the expenseof collecting data that are of sufficient resolutionand extent to conduct sequential analyses whilevarying the spatial dimensions of the data set. Therelative roles of temperature and channel mor-phology in structuring river fish assemblages areknown to change as the spatial extent and loca-tion in the drainage are altered (Matthews 1998).However, the specific quantitative relationshipbetween spatial scale and the observed effects of

these two variables has not been previously de-scribed. The quantitative approach that we em-ployed in this study may be useful in river fishecology and management both for understand-ing fish–habitat relationships and for identifyingtransitions in fish assemblage structure.

Transition zones between cool- and coldwaterfish assemblages were difficult to identify in thisstudy because the assemblages overlapped con-siderably in all three study rivers. However, cross-over points in the relative influences oftemperature and channel morphology on fishassemblage structure were useful for identifyingpotential habitat-specific transitions betweencool- and coldwater fish assemblages. Differencesin the structure of deepwater assemblages in theMFJD, NFJD, and WEN indicated that deep poolenvironments may be occupied by either cold-or coolwater fishes, depending on water tempera-ture. Others have observed similar transitions atdeep pools when continuously electrofishinglarge Pacific Northwest rivers (R. M. Hughes,Oregon State University, Corvallis, unpublisheddata). Without data of such high spatial resolu-tion and extent, we would not have detected suchcrossover points in the habitat relationships offish assemblages.

The crossover point in the relative influence ofchannel morphology (i.e., depth and velocity)versus water temperature on fish assemblagestructure may provide a useful index for assess-ing and monitoring biological potential for cool-and coldwater fishes in rivers. In the MFJD andthe NFJD, crossover points in assemblage struc-ture occurred at 20–22°C (mean daily tempera-ture during the hottest week of the year). Thistemperature range corresponds with the highestmean weekly temperatures recommended forcoldwater fish species cited by Armour (1991) andthe thermal transition zone recorded by Taniguchiet al. (1998) for trout and nontrout assemblagesin the Rocky Mountains. Because this is the firstdescription of such crossover points in fish assem-blage structure, more examples are needed froma range of rivers over a broader geographic areato test the application and further develop the

22torgersen.p65 7/28/2006, 10:00 AM487


utility of this approach for examining transitionsin fish assemblage structure.

Spatially Continuous Analysisof Fish�Habitat Relationships

The response of riverine fishes to habitat het-erogeneity at intermediate scales is poorly un-derstood (Fausch et al. 1994, 2002). In part, thisis due to the relative ease of either assessing broad-scale patterns in fish distribution with respect togeographic variation in elevation and air tem-perature (Rahel and Nibbelink 1999) or observ-ing fine-scale patterns in fish behavior inindividual pool–riffle sequences and in the labo-ratory (Reeves et al. 1987; Taniguchi et al. 1998).In both site-based and laboratory approaches, thenumber of samples used to evaluate statisticalrelationships is relatively small. To collect thelarge number of samples necessary for evaluat-ing spatially continuous patterns in fish distri-bution, we used snorkeling and relativeabundance estimates. In many instances, relativeabundance categories (abundant, common, rare)and presence–absence data are sufficient foridentifying important trends in river fish assem-blages (Rahel 1990); however, estimates of rela-tive abundance and presence–absence may beunreliable if they are uncorrected for samplingefficiency (Bayley and Dowling 1993). Neverthe-less, many mountain rivers are too large to samplewith a backpack electrofishing unit and too smallto sample with boat electrofishing gear (Hugheset al. 2002; Mebane et al. 2003). These logisticalchallenges make it difficult to validate visual es-timates of fish abundance. To compensate for thelack of precision in snorkeling surveys, we useda modified version of point abundance sampling(Persat and Copp 1990) and found that largenumbers of visual estimates of fish abundancewere quite effective for quantifying spatial pat-terns in river fish assemblages. Other studies havesuccessfully employed snorkeling and less rigor-ous electrofishing methods (single-pass) toevaluate patterns of fish distribution in smallstreams (Hankin and Reeves 1988; Thurow and

Schill 1996; Mullner et al. 1998; Bateman et al.2005). New sampling methods and statistical ap-proaches, such as hydroacoustics, point abun-dance sampling, and replicate sampling (Barkerand Sauer 1995; Duncan and Kubecka 1996; Caoet al. 2001), are all applicable to surveys of riverfishes and represent an area needing more re-search in order to better describe and understandthe spatial distribution of river fishes.

ACKNOWLEDGMENTS

We are indebted to the generous field supportprovided by numerous technicians and volun-teers, including L. Weaver-Baxter, C. Lacey, C.Krueger, R. Scarlett, S. Robertson, P. Howell, K.Dwire, K. Wright, J. Li, B. Hasebe, P. Jacobs, B.Landau, K. Hychka, E. Phillips, and S. Hastie.Additional technical support was provided by theUmatilla and Malheur National Forests. We alsothank the land owners in the Middle Fork JohnDay River basin for allowing us to conduct fish-eries research on their properties. Thermal re-mote sensing flights were coordinated by R. Fauxwith Watershed Sciences, Inc. and Snowy ButteHelicopters. GIS facilities and consulting wereprovided by K. Christiansen and the Aquatic–Land Interactions research group at the PacificNorthwest Research Laboratory, Forestry Sci-ences Laboratory in Corvallis, Oregon. Researchfunding was provided by the U.S. Environmen-tal Protection Agency/National Science Founda-tion Joint Watershed Research Program(R82–4774–010 for ecological research) and theBonneville Power Administration (project No.88–108 for salmon research). Insightful com-ments and constructive criticism from R. Hughesand three anonymous reviewers improved thefocus and quality of the manuscript.

REFERENCES

Armour, C. L. 1991. Guidance for evaluating and rec-

ommending temperature regimes to protect fish.

U.S. Fish and Wildlife Service Biological Report

90(22), Washington, D.C.

22torgersen.p65 7/28/2006, 10:00 AM488


Barker, R. J., and J. R. Sauer. 1995. Statistical aspects

of point count sampling. Pages 125–130 in C. J.

Ralph, J. R. Sauer, and S. Droege, editors. Moni-

toring bird populations by point counts. U.S. Forest

Service, General Technical Report PSW-GTR-149,

Albany, California.

Bateman, D. S., R. E. Gresswell, and C. E. Torgersen.

2005. Evaluating single-pass catch as a tool for iden-

tifying spatial pattern in fish distribution. Journal

of Freshwater Ecology 20(2):335–345.

Baxter, C. V. 2002. Fish movement and assemblage

dynamics in a Pacific Northwest riverscape. Doc-

toral dissertation. Oregon State University, Corvallis.

Baxter, C. V., and F. R. Hauer. 2000. Geomorphology,

hyporheic exchange, and selection of spawning

habitat by bull trout (Salvelinus confluentus). Ca-

nadian Journal of Fisheries and Aquatic Sciences

57:1470–1481.

Bayley, P. B., and D. C. Dowling. 1993. The effect of

habitat in biasing fish abundance and species rich-

ness estimates when using various sampling meth-

ods in streams. Polskie Archiwum Hydrobiologii

40:5–14.

Belliard, J., P. Boet, and E. Tales. 1997. Regional and

longitudinal patterns of fish community structure

in the Seine River basin, France. Environmental

Biology of Fishes 50:133–147.

Bisson, P. A., J. L. Nielsen, R. A. Palmason, and L. E.

Grove. 1982. A system of naming habitat types in

small streams, with examples of habitat utilization

by salmonids during low streamflow. Pages 62–73

in N. B. Armantrout, editor. Acquisition and utili-

zation of aquatic habitat inventory information.

American Fisheries Society, Bethesda, Maryland.

Bult, T. P., R. L. Haedrich, and D. C. Schneider. 1998.

New technique describing spatial scaling and habi-

tat selection in riverine habitats. Regulated Rivers:

Research & Management 14:107–118.

Cao, Y., D. P. Larsen, and R. M. Hughes. 2001. Evalu-

ating sampling sufficiency in fish assemblage sur-

veys: a similarity-based approach. Canadian

Journal of Fisheries and Aquatic Sciences 58:

1782–1793.

Clarke, K. R. 1993. Non-parametric analyses of

changes in community structure. Australian Jour-

nal of Ecology 18:117–143.

Clarke, S. E., and S. A. Bryce. 1997. Hierarchical sub-

divisions of the Columbia Plateau and Blue Moun-

tains ecoregions, Oregon and Washington. U.S.

Forest Service, Portland, Oregon.

Collares-Pereira, M. J., M. F. Magalhaes, A. M.

Geraldes, and M. M. Coelho. 1995. Riparian eco-

tones and spatial variation of fish assemblages in

Portuguese lowland streams. Hydrobiologia

303:93–101.

Cunjak, R. A., R. G. Randall, and E. M. P. Chadwick.

1988. Snorkeling versus electrofishing: a compari-

son of census techniques in Atlantic salmon rivers.

Naturaliste Canadien 115:89–93.

Duncan, A., and J. Kubecka. 1996. Patchiness of lon-

gitudinal fish distributions in a river as revealed

by a continuous hydroacoustic survey. ICES Jour-

nal of Marine Science 53:161–165.

ESRI (Environmental Systems Research Institute).

1996. ARC/INFO GIS, version 7.0.4. ESRI,

Redlands, California.

Fausch, K. D., S. Nakano, and K. Ishigaki. 1994. Dis-

tribution of two congeneric charrs in streams of

Hokkaido Island, Japan: Considering multiple fac-

tors across scales. Oecologia 100:1–12.

Fausch, K. D., C. E. Torgersen, C. V. Baxter, and H. W.

Li. 2002. Landscapes to riverscapes: bridging the

gap between research and conservation of stream

fishes. BioScience 52:483–498.

Hankin, D. G., and G. H. Reeves. 1988. Estimating to-

tal fish abundance and total habitat area in small

streams based on visual estimation methods. Ca-

nadian Journal of Fisheries and Aquatic Sciences

45:834–844.

Harding, J. S., E. F. Benfield, P. V. Bolstad, G. S. Helfman,

and E. D. B. Jones. 1998. Stream biodiversity: the

ghost of land use past. Proceedings of the National

Academy of Sciences 95:14843–14847.

Horwitz, R. J. 1978. Temporal variability patterns and

the distributional patterns of stream fishes. Eco-

logical Monographs 48:307–321.

Huet, M. 1959. Profiles and biology of western

European streams as related to fish management.

Transactions of the American Fisheries Society

88:155–163.

Hughes, R. M., and J. R. Gammon. 1987. Longitudi-

nal changes in fish assemblages and water quality

22torgersen.p65 7/28/2006, 10:00 AM489


in the Willamette River, Oregon. Transactions of

the American Fisheries Society 116:196–209.

Hughes, R. M., P. R. Kaufmann, A. T. Herlihy, S. S.

Intelmann, S. C. Corbett, M. C. Arbogast, and R.

C. Hjort. 2002. Electrofishing distance needed to

estimate fish species richness in raftable Oregon

rivers. North American Journal of Fisheries Man-

agement 22:1229–1240.

Isaak, D. J., and T. C. Bjornn. 1996. Movement of

northern squawfish in the tailrace of a lower Snake

River dam relative to the migration of juvenile

anadromous salmonids. Transactions of the Ameri-

can Fisheries Society 125:780–793.

Isaak, D. J., and W. A. Hubert. 2000. Are trout popula-

tions affected by reach-scale stream slope? Cana-

dian Journal of Fisheries and Aquatic Sciences

57:468–477.

Isaak, D. J., and W. A. Hubert. 2001. A hypothesis about

factors that affect maximum summer stream tem-

peratures across montane landscapes. Journal of

the American Water Resources Association 37:

351–366.

Jongman, R. H. G., C. J. F. ter Braak, and O. F. R. van

Tongeren. 1995. Data analysis in community and

landscape ecology. Cambridge University Press,

Cambridge, UK.

Joyce, M. P., and W. A. Hubert. 2003. Snorkeling as an

alternative to depletion electrofishing for assess-

ing cutthroat trout and brown trout in stream

pools. Journal of Freshwater Ecology 18:215–222.

Li, H. W., C. B. Schreck, C. E. Bond, and E. R. Rextad.

1987. Factors influencing changes in fish assem-

blages of Pacific Northwest streams. Pages 192–202

in W. J. Matthews, and D. C. Heins, editors. Com-

munity and evolutionary ecology of North Ameri-

can stream fishes. University of Oklahoma Press,

Norman.

Loy, W. G., S. Allan, A. R. Buckley, and J. E. Meacham.

2001. Atlas of Oregon. University of Oregon Press,

Eugene.

Matthews, W. J. 1998. Patterns in freshwater fish ecol-

ogy. Chapman and Hall, New York.

McCune, B. 1997. Influence of noisy environmental

data on canonical correspondence analysis. Ecol-

ogy 78:2617–2623.

McCune, B., and J. B. Grace. 2002. Analysis of eco-

logical communities. MjM Software Design,

Gleneden Beach, Oregon.

McCune, B., and M. J. Mefford. 1999. Multivariate

analysis of ecological data, version 4.27. MjM Soft-

ware, Gleneden Beach, Oregon.

McMichael, G. A., and T. N. Pearsons. 1998. Effects of

wild juvenile spring chinook salmon on growth

and abundance of wild rainbow trout. Transactions

of the American Fisheries Society 127:261–274.

Mebane, C. A., T. R. Maret, and R. M. Hughes. 2003.

An index of biological integrity (IBI) for Pacific

Northwest rivers. Transactions of the American

Fisheries Society 132:239–262.

Montgomery, D. R., E. M. Beamer, G. R. Pess, and T. P.

Quinn. 1999. Channel type and salmonid spawn-

ing distribution and abundance. Canadian Jour-

nal of Fisheries and Aquatic Sciences 56:377–387.

Mullner, S. A., W. A. Hubert, and T. A. Wesche. 1998.

Snorkeling as an alternative to depletion electro-

fishing for estimating abundance and length-class

frequencies of trout in small streams. North Ameri-

can Journal of Fisheries Management 18:947–953.

Naiman, R. J., H. Decamps, J. Pastor, and C. A.

Johnston. 1988. The potential importance of

boundaries to fluvial ecosystems. Journal of the

North American Benthological Society 7:289–306.

Orr, E. L., W. N. Orr, and E. M. Baldwin. 1992. Geol-

ogy of Oregon. Kendall/Hunt Publishing,

Dubuque, Iowa.

Paller, M. H. 1994. Relationships between fish assem-

blage structure and stream order in South Caro-

lina Coastal Plain streams. Transactions of the

American Fisheries Society 123:150–161.

Persat, H., and G. H. Copp. 1990. Electric fishing and

point abundance sampling for the ichthyology of

large rivers. Pages 197–209 in I. G. Cowx, editor.

Developments in electric fishing. Cambridge Uni-

versity Press, Cambridge, UK.

Poizat, G., and D. Pont. 1996. Multi-scale approach to

species-habitat relationships: juvenile fish in a large

river section. Freshwater Biology 36:611–622.

Poole, G. C. 2002. Fluvial landscape ecology: address-

ing uniqueness within the river discontinuum.

Freshwater Biology 47:641–660.

22torgersen.p65 7/28/2006, 10:00 AM490


Radko, M. A. 1997. Spatially linking basin-wide stream

inventories in a geographic information system.

U.S. Forest Service, General Technical Report INT-

GTR-345, Ogden, Utah.

Rahel, F. J. 1990. The hierarchical nature of commu-

nity persistence: a problem of scale. The American

Naturalist 136:328–344.

Rahel, F. J., and W. A. Hubert. 1991. Fish assemblages

and habitat gradients in a Rocky Mountain-Great

Plains stream: biotic zonation and additive patterns

of community change. Transactions of the Ameri-


Rahel, F. J., and N. P. Nibbelink. 1999. Spatial patterns

in relations among brown trout (Salmo trutta) dis-

tribution, summer air temperature, and stream size

in Rocky Mountain streams. Canadian Journal of

Fisheries and Aquatic Sciences 56:43–51.

Reeves, G. H., F. H. Everest, and J. D. Hall. 1987. Inter-

actions between the redside shiner (Richardsonius

balteatus) and the steelhead trout (Salmo gairdneri)

in western Oregon: the influence of water tempera-

ture. Canadian Journal of Fisheries and Aquatic

Sciences 44:1603–1613.

Reynolds, L., A. T. Herlihy, S. V. Gregory, and R. M.

Hughes. 2003. Electrofishing effort requirements

for assessing species richness and biotic integrity

in western Oregon streams. North American Jour-

nal of Fisheries Management 23:450–461.

Roper, B. B., and D. L. Scarnecchia. 1994. Summer dis-

tribution of and habitat use by chinook salmon

and steelhead within a major basin of the South

Umpqua River, Oregon. Transactions of the Ameri-


Sheldon, A. L. 1968. Species diversity and longitudinal

succession in stream fishes. Ecology 49:193–198.

Smith, J. J., and H. W. Li. 1983. Energetic factors in-

fluencing foraging tactics of juvenile steelhead

trout, Salmo gairdneri. Pages 173–180 in D. L. G.

Noakes, editor. Predators and prey in fishes. Dr.

W. Junk Publishers, The Hague, The Netherlands.

SPSS. 2001. SigmaPlot, version 8.0. SPSS Science,

Chicago.

Stewart, B. G., J. G. Knight, and R. C. Cashner. 1992.

Longitudinal distribution and assemblages of fishes

of Byrd’s Mill Creek, a southern Oklahoma

Arbuckle Mountain stream. The American Natu-

ralist 37:138–147.

Tabachnick, B. G., and L. S. Fidell. 2001. Using multi-

variate statistics. Allyn and Bacon, Boston.

Taniguchi, Y., F. J. Rahel, D. C. Novinger, and K. G.

Gerow. 1998. Temperature mediation of competi-

tive interactions among three fish species that re-

place each other along longitudinal stream

gradients. Canadian Journal of Fisheries and

Aquatic Sciences 55:1894–1901.

Taylor, C. M., M. R. Winston, and W. J. Matthews. 1996.

Temporal variation in tributary and mainstem fish

assemblages in a Great Plains stream system.

Copeia 2:280–289.

Thompson, A. R., J. T. Petty, and G. D. Grossman. 2001.

Multi-scale effects of resource patchiness on for-

aging behaviour and habitat use by longnose dace,

Rhinichthys cataractae. Freshwater Biology 46:

145–160.

Thurow, R. F., and D. J. Schill. 1996. Comparison of

day snorkeling, night snorkeling, and electro-

fishing to estimate bull trout abundance and size

structure in a second-order Idaho stream. North

American Journal of Fisheries Management

16:314–323.

Torgersen, C. E. 2002. A geographical framework for

assessing longitudinal patterns in stream habitat

and fish distribution. Doctoral dissertation. Or-

egon State University, Corvallis.

Torgersen, C. E., and D. A. Close. 2004. Influence of

habitat heterogeneity on the distribution of larval

Pacific lamprey (Lampetra tridentata) at two spa-

tial scales. Freshwater Biology 49:614–630.

Torgersen, C. E., R. N. Faux, B. A. McIntosh, N. J. Poage,

and D. J. Norton. 2001. Airborne thermal remote

sensing for water temperature assessment in rivers

and streams. Remote Sensing of Environment

76:386–398.

Torgersen, C. E., D. M. Price, H. W. Li, and B. A. McIn-

tosh. 1999. Multiscale thermal refugia and stream

habitat associations of chinook salmon in north-

eastern Oregon. Ecological Applications 9:301–319.

Trexler, J. C., and J. Travis. 1993. Nontraditional re-

gression analysis. Ecology 74:1629–1637.

Vannote, R. L., G. W. Minshall, K. W. Cummins, J. R.

22torgersen.p65 7/28/2006, 10:00 AM491


Sedell, and C. E. Cushing. 1980. The river con-

tinuum concept. Canadian Journal of Fisheries and

Aquatic Sciences 37:130–137.

Wiens, J. A. 1989. Spatial scaling in ecology. Functional

Ecology 3:385–397.

Williams, L. R., C. M. Taylor, M. L. Warren, and J. A.

Clingenpeel. 2003. Environmental variability, his-

torical contingency, and the structure of regional

fish and macroinvertebrate faunas in Ouachita

Mountain stream systems. Environmental Biology

of Fishes 67:203–216.

Wilson, J. B., J. Steel, W. King, and H. Gitay. 1999. The

effect of spatial scale on evenness. Journal of Veg-

etation Science 10:463–468.

Zaroban, D., M. Mulvey, T. Maret, R. Hughes, and G.

Merritt. 1999. Classification of species attributes

for Pacific Northwest freshwater fishes. Northwest

Science 73:81–93.

Zubik, R. J., and J. J. Fraley. 1988. Comparison of snor-

kel and mark-recapture estimates for trout popu-

lations in large streams. North American Journal

of Fisheries Management 8:58–62.

22torgersen.p65 7/28/2006, 10:00 AM492

Documents

Landscape Influences on Longitudinal Patterns of River ...faculty.washington.edu/cet6/pub/Torgersen_etal_2006.pdf · Analysis of FishŒHabitat Relationships Christian E. Torgersen*,1