Content uploaded by Peter M. Kiffney
Author content
All content in this area was uploaded by Peter M. Kiffney on Apr 17, 2014
Content may be subject to copyright.
Tributary streams create spatial discontinuities in
habitat, biological productivity, and diversity in
mainstem rivers
P.M. Kiffney, C.M. Greene, J.E. Hall, and J.R. Davies
Abstract: Lotic ecosystems are made up of numerous tributary streams forming a complex branching network. The
point where smaller tributaries flow into larger rivers, or tributary junctions, may be sites in the network where spatial
discontinuities or “hot spots” are created and maintained, because small streams funnel important materials captured
from the surrounding landscape and carry them by gravity downstream. We hypothesized that habitat complexity, envi-
ronmental productivity, and abundance of primary consumers and predators peak in mainstem rivers at or downstream
of tributary junctions. We conducted surveys in three river basins and 13 reaches to examine interdependence between
tributary streams and the larger rivers they enter. Wood abundance and volume, variability in median substrate size
(i.e., substrate heterogeneity), concentrations of nitrogen and phosphorus in water, algal biomass, and abundance of
consumers and predators peaked with a higher frequency at or downstream of tributary junctions. For several variables,
the size of the tributary relative to the main stem contributed to the strength of tributary affect. These findings suggest
that some tributary streams have fundamental effects on the larger rivers they enter. We argue that maintaining the
integrity of connections among and between ecosystems is essential for promoting habitat complexity and community
structure within river networks.
Résumé : Les systèmes lotiques sont formés de nombreux tributaires réunis en un réseau dendritique complexe. Les
points de jonction des tributaires, où les tributaires de taille inférieure se jettent dans les rivières de taille plus importante,
peuvent générer et maintenir des discontinuités ou des « points chauds » dans le réseau; en effet, les petits cours d’eau
accumulent des matériaux importants provenant du paysage environnant et les charrient vers l’aval par gravité. Nous
formons l’hypothèse selon laquelle la complexité de l’habitat, la productivité du milieu et l’abondance des consommateurs
primaires et des prédateurs atteignent toutes des maximums dans les cours principaux aux points de jonction des tributaires
ou juste en aval. Nous avons mené des inventaires dans trois réseaux hydrographiques et 13 sections de cours d’eau
pour évaluer l’interdépendance entre les tributaires et la rivière plus importante qu’ils rejoignent. Nous observons que les
maximums d’abondance et de volume de bois, de variabilité de la taille moyenne du substrat (c’est-à-dire de
l’hétérogénéité du substrat), des concentrations d’azote et de phosphore dans l’eau, des biomasses des algues et des
abondances des consommateurs et des prédateurs se situent plus fréquemment aux points de jonction des tributaires ou
juste en aval. Pour plusieurs des variables, la taille du tributaire par rapport au cours axial de la rivière contribue à
augmenter l’effet du tributaire. Ces données laissent croire que certains tributaires ont des effets radicaux sur les rivières
plus grandes qu’ils alimentent. Nous croyons que le maintien de l’intégrité des interconnections à l’intérieur des éco-
systèmes et entre les écosystèmes est essentiel pour favoriser la complexité des habitats et la structure des communautés
dans les réseaux de rivières.
[Traduit par la Rédaction] Kiffney et al. 2530
Introduction
Discontinuities, or transition zones, are ubiquitous in nature
(Naiman and Décamps 1997) and are potentially important
in maintaining biodiversity (Naiman et al. 1988; Polis et al.
1997). For example, river networks contain numerous dis-
continuities, with streams entering and leaving lakes (Rich-
ardson and Mackay 1991), rivers mixing with floodplains
(Amoros and Bornette 2002; Sabo and Power 2002), two
rivers intersecting (i.e., tributary junction or confluence) (Rice
Can. J. Fish. Aquat. Sci. 63: 2518–2530 (2006) doi:10.1139/F06-138 © 2006 NRC Canada
2518
Received 11 January 2006. Accepted 20 June 2006. Published on the NRC Research Press Web site at http://cjfas.nrc.ca on
25 October 2006.
J19102
P.M. Kiffney.1NOAA Fisheries, Northwest Fisheries Science Center, Mukilteo Biological Field Station, 10 Park Avenue, Building
B, Mukilteo, WA 98275, USA.
C.M. Greene, J.E. Hall, and J.R. Davies. NOAA Fisheries, Northwest Fisheries Science Center, 2725 Montlake Boulevard East,
Seattle, WA 98112, USA.
1Corresponding author (e-mail: peter.kiffney@noaa.gov).
et al. 2001), and rivers entering oceans (Darnaude et al.
2004). The movement of water, energy, nutrients, sediment,
wood, and organisms from one distinct habitat to another
may create sites of high habitat complexity, biological produc-
tivity, and diversity at these discontinuities or “hotspots”
(Polis et al. 1997; Power and Dietrich 2002). Unfortunately,
we know relatively little about the formation and location of
discontinuities in river networks and whether they are asso-
ciated with habitat complexity, productivity, or community
structure (Fisher 1997; Power and Dietrich 2002; Benda et
al. 2004).
Until recently, conceptual models of aquatic systems have
often downplayed or ignored the importance of discontinu-
ities. For the last 25 years, the field of aquatic ecology has
generally relied on the River Continuum Concept, which
posits that controls on aquatic systems result in linear
patterns of abundance and diversity as one proceeds down-
stream (Vannote et al. 1980). Although patterns in some fac-
tors, in particular energy inputs, support predictions of the
River Continuum Concept (Minshall et al. 1983), results
from other studies have not supported biodiversity patterns
predicted by this model (Winterbourne et al. 1981). More-
over, considerable evidence is mounting that aquatic habitat
complexity and productivity are not solely longitudinal func-
tions of river length, but are instead discontinuously distrib-
uted in river networks (Rice et al. 2001; Poole 2002; Benda
et al. 2004). In this paper, we examine whether one potential
source of discontinuities in aquatic systems — tributary con-
fluences — create gradients in productivity and diversity.
A conceptual model recently proposed by Benda et al.
(2004), which couples disturbance regime with network
structure, suggests that habitat complexity is generally greater
around tributary confluences; therefore, biological diversity
and productivity should peak at these points. Geomorph-
ologists have long appreciated that unique changes in physical
habitat occur at tributary confluences (Benda et al. 2004, and
references cited therein). Only a few studies, however, have
explicitly examined the effects of tributaries on the distribution
and abundance of organisms of the larger rivers they enter
and the mechanisms driving these patterns (Bruns et al. 1984;
Stevens et al. 1997; Rice et al. 2001). Rice et al. (2001)
found that invertebrate taxa richness peaked at tributary con-
fluences and speculated that this pattern was in response to
greater habitat diversity.
We build upon these earlier studies examining the role of
small streams on larger rivers by assessing patterns in a
range of environmental attributes that are known to affect
species’ abundance and distribution. Our study is unique in
that we examined multiple physical, chemical, and biological
endpoints, including algal biomass, density of large inverte-
brate consumers, and density and diversity of benthic and
water column fishes. Testing this hypothesis is a challenge,
primarily because potential nonlinear effects of tributaries on
their main stems might result in several possible patterns
across mainstem gradients (Fig. 1). We therefore examined
tributaries that ranged in size relative to the main stems they
enter in four different basins and used surveys that covered
an intermediate spatial scale (101–105km). These design
elements have been proposed to be important for under-
standing factors controlling fish assemblages (Fausch et al.
2002). We sampled an extensive range of ecological vari-
ables including multiple taxa that allowed us to identify
patterns in habitat complexity, water chemistry, and stream
organisms. Specifically, we tested the hypothesis that
tributary streams increase habitat complexity and resource
abundance of the larger rivers they enter and consequently
result in peaks of biological production, abundance of multi-
ple taxa, and fish diversity at and below tributaries.
Materials and methods
Study sites
Our surveys were conducted at the Cedar River, Taylor
Creek, Bacon Creek, and Finney Creek, which are located in
the foothills of the Cascade Range mountains in western
Washington, USA. This area is part of the Pacific Coastal
ecoregion (Naiman and Bilby 1998), which is generally mild
(minimum–maximum annual temperature = 4–15 °C; National
Oceanic and Atmospheric Administration (NOAA) 2005),
with wet (minimum–maximum annual precipitation = 145–
203 cm) winters and a pronounced summer dry season.
Upland and riparian vegetation are dominated by the conifers
western hemlock (Tsuga heterophylla (Raf.) Sarg.), western
redcedar (Thuja plicata Donn ex D. Don), Douglas-fir
(Pseudotsuga menziesii (Mirb.) Franco), and Sitka spruce
(Picea sitchenis (Bong.) Carrière) (Franklin and Dyrness 1969).
Dominant broadleaf riparian species are primarily black cotton-
wood (Populus trichocarpa Torr. & A. Gray), red alder
(Alnus rubra Bong.), vine maple (Acer circinatum Pursh),
and salmonberry (Rubus spectabilis Pursh).
These basins were primarily forested, with Bacon Creek
and Finney Creek tributary to the Skagit River (Fig. 2; Table 1).
© 2006 NRC Canada
Kiffney et al. 2519
Fig. 1. Hypothetical relationships of nonlinear patterns in ecological
variables (i.e., chemical, physical, or biological) resulting from
tributary inputs upon a mainstem reach. Distances from the tributary
junction or “0 m” (solid vertical line) increase traveling upstream
or downstream in metres. The broken line indicates that tributaries
may cause a localized effect resulting from various inputs (e.g.,
nutrients). The solid line shows that tributaries might also have
effects both upstream and downstream on the main stem if inputs
(e.g., hyporheic flow or gravel) are spread across an alluvial fan.
The thick line indicates that tributaries might also have effects
far downstream if inputs are large or result in greater habitat
complexity.
© 2006 NRC Canada
2520 Can. J. Fish. Aquat. Sci. Vol. 63, 2006
Fig. 2. Map of study streams: (a) Skagit River sites, (b) Cedar River and Taylor Creek sites. Circles represent study reaches, with the
small dot within each circle representing location of tributary–mainstem confluence.
Finney Creek is located on private timberland, while Bacon
Creek is located on US Forest Service and National Park
Service property. The Cedar River and Taylor Creek are
located within the city of Seattle’s watershed, which is
currently managed as a conservation area. Chester Morse
Lake, a water storage reservoir, is ~4 km upstream of the
Steele Creek – Cedar River confluence. Because there were
no pre-dam assessments, we are unable to determine
whether there have been impacts on stream habitat below
the reservoir. Potential impacts of dams include changes in
water velocity, stream depth, substrate composition, and
cover for fish. We argue, however, that reservoir operations
have not overwhelmed the impact of tributary streams on the
Cedar River main stem. The flow regime below the reservoir
is similar to flow patterns observed in nearby unregulated
rivers (United States Geographical Survey (USGS) 2006);
therefore, impacts on flow are likely minimal. Because Ches-
ter Morse Lake is on a site of a natural lake and the reach
downstream of the dam is a bedrock canyon, neither
streambed degradation nor armoring occurs downstream
(Seattle Public Utilities 2000). Moreover, any longitudinal
impacts on sediment characteristics due to the dam in the sec-
tions we surveyed were minimal because the first tributary is
relatively far from the reservoir (~4 km). Finally, none of
the tributaries are regulated. Without data, however, there
remains the possibility that the reservoir may impact down-
stream habitat, thereby confounding effects of tributaries on
the main stem.
To examine gradients in environmental characteristics of
mainstem rivers, we conducted detailed surveys on the physical,
chemical, and biological attributes at 13 tributary junctions.
We surveyed four tributary–mainstem reaches on the Cedar
River (July–August 2002) and three on Taylor (August 2002),
Finney (July–August 2003), and Bacon creeks (August–
September 2003). These are clear, cool, rocky bottom streams,
with mainstem sites relatively open because of their width,
whereas tributaries are primarily shaded. The area of our
mainstem sites ranged from 1578 to 30 121 ha, while the
ratio of tributary to mainstem area ranged over an order of
magnitude from 0.046 to 0.49 (Table 1).
Survey protocols
Reach delineation
Our survey design and methods were based on a standardized
approach developed by the United States Environmental
Protection Agency’s Environmental Monitoring and Assess-
ment Program (Kaufmann 2002). Reach lengths were 40
times the average wetted width of the main stem during our
surveys and ranged from 240 to 920 m (Table 1). Each reach
was divided into 11 equally spaced transects that were per-
pendicular to water flow, with six primary transects upstream
and five transects downstream of each tributary (Fig. 3).
Distances between transects differed among the 13 tributary–
mainstem reaches, because the distance between transects
was shorter for narrower channels and longer for wider
channels. To standardize this distance across the 13 tributary
junctions for analysis and graphical presentation, we divided
the distance between each transect and the tributary by the
intratransect distance. We called this distance between
transects as the standardized distance unit (SDU), ranging in
value from –5 to 5.
To determine the location of each transect, we used the
upstream extent of alluvial fans at the mouth of each tributary
to place the initial primary transect. The position of remaining
primary transects were then dependent on this location. We
also placed secondary transects within each reach, which
were 12.5%, 25%, and 50% of the distance between primary
transects. Secondary transects were located between 1 and
–1 SDU and ranged from 0.5 to –0.5 SDU. The purpose of
these transects was to determine if there were small-scale
patterns in water temperature, algal biomass, nutrient chemistry,
and large invertebrate and sculpin (Cottus sp.) abundance,
resulting from tributary confluences. Because we hypothe-
sized these small-scale spatial patterns would dissipate rela-
tively quickly as one moved away from the tributary junction
and maintaining this sampling intensity between all transects
was logistically a challenge, we confined these secondary
transects to between 1 and –1 SDU.
Quartz and Hatchery creeks entered Finney Creek about
50 m apart, creating a large floodplain complex. Because
© 2006 NRC Canada
Kiffney et al. 2521
Main
stem Tributary
Tributary
area (ha)
Mainstem
area (ha)
Tributary/
main stem
Reach
length (m)
Average channel
depth (cm)
Average channel
width (m) Slope
Finney Ruxell 481 10 382 0.046 680 35 16 0.8
Hatchery* 454 11 100 0.041
Quartz* 1 053 11 566 0.091 840 52 16 1.5
Bacon Jumbo 206 8 734 0.024 800 76 20 1.6
Falls 1 466 9 530 0.15 920 66 21 1.5
Oakes 324 11 715 0.028 840 75 21 1.8
Cedar Steele 281 22 814 0.012 800 77 22 1.0
Williams 600 23 800 0.025 800 95 26 0.6
Taylor 4 431 24 408 0.18 800 95 29 0.6
Rock 1 484 30 121 0.05 800 83 33 0.6
Taylor South Fork 548 1 578 0.35 240 27 6 1.7
North Fork 1 103 3 799 0.49 360 36 10 2.0
Seventeen 530 3 799 0.14 500 51 12 1.5
*Confluences were so close that we combined reaches.
Table 1. Tributary and mainstem areas, ratio of tributary to mainstem area, reach lengths, average channel depth, channel width, and
slope for each tributary–mainstem confluence.
these two tributaries were so close together, it was difficult
to determine whether observed patterns were related to indi-
vidual tributaries. Therefore, we treated this tributary complex
as one reach with one tributary input.
Physical and chemical measurements
Throughout each study reach, we estimated wood volume,
substrate size, canopy cover, and nutrient levels. Pieces of
wood >0.1 m in diameter and >1.0 m in length that were in
contact with the wetted channel were counted continually
from transect 5 through –5. Wood abundance was converted
to volume within each size class using the following for-
mula: wood volume (m3) = 3.14 r2l, where requals the radius
and lthe length of each log in each size class. These values
were then summed across size classes to determine mini-
mum wood volume per transect. To determine substrate size
and heterogeneity (coefficient of variation), pebble counts
were conducted at each primary transect. We assigned 100
to 300 rocks selected across each transect to different size
categories using the intermediate axis and a gravelometer
(Wildco Supply, Buffalo, New York). Individual stones were
selected by picking up the rock located directly in front of
the big toe of the sampler’s boot. One step was taken from
this point where another rock was selected until 100 to 300
rocks were measured. Each pebble count began and finished
at the stream edge to minimize bias introduced by location
within transects. Canopy cover was measured at primary
transects using a spherical densiometer. At each transect,
measurements were taken in four directions (facing the bank,
upstream, downstream, and facing towards the opposite bank)
in the middle of the channel and once at each bank facing
the stream.
Grab water samples were collected from primary and
secondary transects on the main stem and tributary. These
samples were analyzed for total nitrogen and phosphorus,
dissolved nitrate and nitrite, ammonia, and soluble-reactive
phosphorus (College of Oceanography, University of Wash-
ington, Seattle, Washington).
Biological measurements
Ambient periphyton biomass was determined by processing
fist-sized rocks (~8 cm wide) collected from three (left,
middle, and right side of transect) locations across each
transect. To determine periphyton and algal biomass, rocks
were scrubbed thoroughly with a brush and rinsed with
distilled water. This slurry was split into equal subsamples
and analyzed for periphyton biomass as ash-free dry mass
(AFDM) and chlorophyll a(see Kiffney et al. 2003 for
further details). To estimate the area of each stone, we used
calipers to measure the x,y, and zaxes of each stone; this
area was then divided by two because we assumed most of
the algal biomass accrued on the upper rock surface.
To characterize benthic fauna, surveys were conducted at
primary and secondary transects. A square plastic (polyvinyl
chloride pipe) frame (0.25 m2area) weighed down by sand
within the pipes was carefully placed on the stream bottom
at three locations (left, middle, and right side) across each
transect. Within each quadrat, we counted and identified
sculpin, a small benthic fish, and large, cased caddisflies
(primarily Dicosmoecus gilvipes). Placing this frame on the
stream bottom likely led to the emigration of small, mobile
insects, such as mayflies (Order Ephemeroptera); however,
we were most interested in counting large, immobile, and
easily identifiable taxa.
Continuous snorkel surveys from transect 5 through –5
were conducted at each reach to determine relative abun-
dance, composition, and size class (total length <50 mm =
class 1, 51–100 mm = class 2, 101–150 mm = class 3,
151–200 mm = class 4, 201–250 mm = class 5, >251 mm =
class 6) of stream salmonids (Oncorhynchus kisutch,Onco-
rhynchus mykiss,Oncorhynchus clarkii clarkii,Oncorhynchus
tshawytscha,Oncorhynchus nerka,Oncorhynchus gorbuscha,
Prosopium williamsoni) and bull trout (Salvelinus confluentas)
(Wydoski and Whitney 2003). Surveys were conducted
within a 2-week period to minimize affects of seasonal
migration on abundance. Depending on channel width, one
to four snorkelers moved upstream in unison from one
transect to the next (Thurow 1994). To minimize double
counts of fish, each snorkeler was assigned a lane; any fish
that moved between lanes (mostly large individuals
>150 mm total length) was assigned to one person through
communication among snorkelers. Snorkelers relayed number,
identity, and size class to recorders on the bank.
We recognize there are sampling issues with using snorkel
surveys to estimate abundance (Thompson 2003), but this
© 2006 NRC Canada
2522 Can. J. Fish. Aquat. Sci. Vol. 63, 2006
Fig. 3. Survey design used for this study. Primary transects were
5 though –5, with the 0 transect located immediately upstream of
the tributary–mainstem confluence. Secondary transects (0.125 to
–0.125) were used to sample nutrients, water temperature, algae,
and Dicosmoecus gilvipes and sculpin (Cottus sp.) density.
approach was used for the following reasons: (i) reach lengths
were long (mean = 698 m) and two rivers were large, deep,
and fast (Cedar River and Bacon Creek), which made electro-
shocking logistically difficult; (ii) water clarity of these
rivers was high (~5 m) and there were relatively few spe-
cies, with most having distinct characteristics making vi-
sual detection and identification of fish relatively easy;
(iii) we wanted to use the same method across habitat types
and stream sizes; and (iv) we were interested in estimating
relative abundances across these strata. Because we used the
same methods across transects, reaches, and streams, we
suggest that this approach was suitable for addressing our
hypotheses. Moreover, observer bias was minimized, be-
cause two individuals (Kiffney and Greene) conducted 90%
of fish counts.
We used two measures to estimate the diversity of stream
fishes. The number of fish species in these systems is low;
however, the number of size classes within a species is
relatively high as a result of complex life histories within
Salmonidae. For example, a large pool may contain two
species of salmon or trout, but support three to four size
classes of each species. We assumed that this species – size
class diversity was analogous to species diversity, especially
within a stream, because a diverse suite of life stages within
a pool may be functionally different from a pool with only
one to two size classes. Therefore, as one measure of fish
diversity, we summed the number of size classes within each
species to determine species × size class diversity. For
instance, one reach may have two species (cutthroat and
rainbow trout), with each species having four size classes, so
the species × size class diversity for this reach equals eight.
Likewise, if one species had two size classes and another
had four size classes, the diversity index would equal six.
We also used a standard diversity index, Simpson’s reciprocal
index of diversity (D–1), to measure species diversity (D=
Σ(nN–1)2), where nis the total number of organisms of a
particular species, and Nis the total number of organisms of
all species (Magurran 1988).
Statistical analysis
Sample units were transects, so multiple samples collected
within individual transects were pooled (e.g., three rock
samples were collected per transect for algal biomass; these
values were pooled to determine the mean value for each
transect).
The nonlinearity in ecological patterns along each reach
made it a challenge to quantify tributary impacts. Therefore,
we focused on the location of peak values for each response
variable as determined by measurements at each transect to
determine whether tributary junctions created discontinuities
in ecological measures. We plotted mean ± 95% confidence
intervals for the location or SDU for the peak value for each
response measure. Because tributary effects on mainstem
habitat can occur upstream and downstream of the tributary
confluence (Benda et al. 2003), we predicted that, on average,
peak values for these response measures would occur at or
downstream of each tributary junction (Fig. 1).
Because there was considerable variation in physical,
chemical, and biological attributes among basins, and almost
every variable we measured revealed major interbasin differ-
ences, we also examined tributary affects by calculating z
scores for sample data of points relative to other points within
a given basin. The zscore reveals how many standard devia-
tions away from the mean a score resides, with a positive
value indicating a zscore above the mean and negative value
below. In addition to standardizing data across basins, zscores
provide a convenient way to test for significant peaks at and
below tributary junctions. Using ztests, we examined whether
these peaks surpassed the critical zscore (assuming a two-
tailed test) for significance when compared with other values
measured within that basin. This test essentially asked
whether a peak value associated with each study reach was
within the range of variation expected for the basin. Hence,
© 2006 NRC Canada
Kiffney et al. 2523
Fig. 4. Scatterplots of response measure values vs. standardized
distance units (SDU) for (a) minimum channel wood volume at
the Quartz–Hatchery–Finney creeks confluence, (b) total phosphorus
concentrations at the Rock Creek – Cedar River confluence,
(c) water temperature at the Seventeen Creek – Taylor Creek
tributary, (d)Dicosmoecus gilvipes density at the Ruxell Creek –
Finney Creek confluence, (e) sculpin (Cottus sp.) density at the
Steele Creek – Cedar River confluence, and (f) Simpson’s recip-
rocal index of fish diversity at the Quartz Creek – Finney Creek
confluence. The shaded, vertical line represents the location of
the tributary–mainstem confluence.
for any single variable, each of the 12 tributary junctions
constituted a ztest. We then used histograms that summed
the frequency of peak values within bins defined by SDUs to
assess the overall pattern of peak values as a function of
distance from tributaries. Before analyzing peak values using
zscores, we examined whether the data were normally dis-
tributed using the Kolmogorov–Smirnov test. When this
assumption was violated, we used the logit transformation
for percentage data and log transformations on other types
of data to normalize their distributions.
Results
There was significant spatial variation in the distribution
of sample values, especially for physical responses and salmonid
abundance; however, surveys revealed distributions that were
generally consistent with patterns in Fig. 1. Specifically, we
observed peaks in a number of response measures at tribu-
tary confluences, with values declining with distance from
this point. For instance, the Quartz–Hatchery–Finney creeks
confluence was highly complex, with in-channel wood peaking
around the tributary–mainstem confluence (Fig. 4a). In gen-
eral, concentrations of limiting nutrients were higher in trib-
utaries, which in some cases translated into higher nutrient
concentrations downstream of confluences, especially in the
Cedar River (Fig. 4b). We also observed that tributary streams
influenced water temperature of mainstem rivers. Water temp-
erature downstream of the Seventeen Creek – Taylor Creek
confluence increased sharply by ~1.4 °C (Fig. 4c). In some
cases, we also observed higher densities of insect consumers
and predators below tributary streams. Density of the large
cased caddisfly, Dicosmoecus gilvipes, was about two–three
times higher near the Ruxell Creek – Finney Creek conflu-
ence compared with transects further away (Fig. 4d). Sculpin
abundance in the Cedar River was also tightly linked to trib-
utary junctions, as density was about two times greater near
the Steele Creek – Cedar River confluence than further away
(Fig. 4e). Although salmon and diversity responses were
generally more variable than productivity or chemical measures,
Simpson’s index of fish diversity peaked at the Quartz–
Hatchery–Finney creeks confluence (Fig. 4f). Patterns in fish
diversity at this tributary–mainstem confluence largely tracked
volume of in-channel wood (Fig. 4a).
When analyzed collectively, we also observed patterns
consistent with our hypothesis that tributary streams create
environmental gradients. On average, the mean location of
almost all measures occurred at or downstream of tributary
junctions (Fig. 5). For example, a higher frequency of peaks
in wood volume and median substrate size occurred at 1 SDU,
with substrate heterogeneity and stream depth occurring
about 0 SDU. Peaks in total nitrogen, phosphorus, ambient
periphyton AFDM, chlorophyll a, and sculpin abundance
occurred from 0.1 to 0.5 SDU and were less variable than
physical or other fish measures. Salmon abundance and
diversity peaked at 1 SDU, which was similar to wood
volume and substrate size. In almost all measurements of
salmon density and diversity, the 95% confidence intervals
for locations of peak values did not overlap the 0 SDU line.
Hence, peak values of these measures were significantly
downstream of tributary junctions.
The z-score analysis provided further evidence for creation
of ecological discontinuities by tributary streams (Fig. 6).
Significant peaks in wood volume (Fig. 6a) and substrate
diversity (Fig. 6b) occurred more often 0–4 SDU down-
stream of tributary junctions than elsewhere. Measures of
productivity were tightly linked with tributary junctions,
although this result may be a function of higher sampling
frequency between 1 and –1 SDU. Total phosphorus generally
© 2006 NRC Canada
2524 Can. J. Fish. Aquat. Sci. Vol. 63, 2006
Fig. 5. Mean ± 95% confidence intervals for the location (standardized distance units, SDU) of peak values for physical, chemical, and
biological response measures. AFDM, ash-free dry mass.
peaked 0–1 SDU (Fig. 6c), while total nitrogen peaked 0–
0.5 SDU (Fig. 6d). Ambient AFDM peaked at 0–1 SDU
(Fig. 6e), while ambient chlorophyll apeaked more often
between –0.5 and +0.5 SDU (Fig. 6f). These variables had
significant peaks associated with tributary inputs in 2–4 of
the 12 reaches.
Patterns in salmon abundance in relation to distance from
tributary junction differed by size class and taxa, but tended
to peak below tributaries. Juvenile salmon (<50 mm total
length) exhibited the strongest tributary-related effect, with
peaks at or downstream of tributaries in 10 of the 12 reaches,
four of which surpassed the level of significance (Fig. 7a).
The pattern for larger salmonids (>50 mm in total length)
was less clear, but significant peaks occurred more often
within 2 SDUs downstream of tributaries (Fig. 7b). The number
of species × size classes, one measure of diversity, peaked
below tributaries in 7 of the 12 reaches we examined and
showed particularly large peaks in two of these (Fig. 7c).
Simpson’s diversity showed a less robust tributary effect
than other variables. Nevertheless, peaks in Simpson’s diversity
also occurred more often close to tributary confluences, with
two observations occurring 0–1 SDU and five occurring 1–2
SDU (Fig. 7d) downstream of tributaries.
Our results confirmed significant ecological effects of
tributaries in some but not all reaches for the variables we
examined. We hypothesized that the magnitude of the tributary
© 2006 NRC Canada
Kiffney et al. 2525
Fig. 6. Spatial distributions of peak values for (a) minimum
volume of channel wood, (b) substrate diversity (defined as the
coefficient of variation (%) of median substrate size), (c) total
phosphorus (µg·L–1), (d) total nitrogen (µg·L–1), (e) ambient
periphyton ash-free dry mass (µg·cm–2), and (f) ambient chloro-
phyll abiomass (µg·cm–2) measured within 12 study reaches in
four river basins. Within each panel, each datum represents the
spatial location of the peak within the study reach, with the
tributuary junction located at zero. Hence, each bar represents
the frequency of peaks measured within five reach sections:
(a) at or between transects –4 and –5 = –4, (b)atorbetween
transects –2 and –3 = –2, (c) at or between –1 and1=0,
including all secondary transects, (d) at or between 2 and3=2,
and (e) at or between 4 and5=4(refer to Fig. 3 for position of
transects). Hatched portions of bars represent data surpassing the
0.05 level of significance when those points were compared with
others within its basin.
Fig. 7. Spatial distributions of peak values for (a) estimates of
salmon density ≤50 mm in total length (fish·m–3), (b) estimates
of salmon density ≥50 mm in total length (fish·m–3), (c) species
× size class, which represents size class diversity, and
(d) Simpson’s reciprocal index of diversity measured within 12
study reaches in four river basins. Within each panel, each datum
represents the spatial location of the peak within the study reach,
with the tributuary junction located at zero. Hence, each bar
represents the frequency of peaks measured within five reach
sections: (a) at or between transects –4 and –5 = –4, (b)ator
between transects –2 and –3 = –2, (c)atorbetween–1and1=0,
including all secondary transects, (d)atorbetween2and3=2,
and (e) at or between 4 and5=4(refer to Fig. 2 for position of
transects). Hatched portions of bars represent data surpassing the
0.05 level of significance when those points were compared with
others within its basin.
effect should be related to the percentage of the watershed
above a reach that was within the tributary. We tested this
by correlating peak zvalues in each reach with the propor-
tion of watershed area in the tributary. We found significant
correlations for three variables (Table 2) and trends in sev-
eral others. These correlations tended to be saturating func-
tions of relative tributary area, leveling off after the tributary
contribution was greater than 10%–15% of the upstream wa-
tershed (Fig. 8).
Discussion
This study provides evidence that tributary junctions
represented discontinuities in river networks. In general,
physical, chemical, and biological attributes of our study
reaches peaked at or downstream of these junctions. In most
cases, the spatial distribution of values for our response
measures was similar to hypothetical distributions, as well as
distributions predicted from biogeochemical mechanisms
operating in river systems. Furthermore, the mean location
for peak values was at or below the tributary confluence. For
example, total nitrogen and phosphorus and algal biomass
generally peaked at or below tributary junctions. In addition,
sculpin, large caddisfly, and juvenile salmonid abundance
and species × size class diversity generally peaked at or
slightly downstream of tributary junctions. We speculate that
these patterns in biotic assemblages were in response to
gradients in physical, chemical, and biological attributes
partially created by tributary streams.
Tributary streams appeared to influence a variety of physical
attributes of the mainstem rivers we examined, increasing
morphological heterogeneity of receiving streams. Wood
volume or abundance and median substrate size were generally
higher at or downstream of tributary junctions. Other physical
variables such as substrate heterogeneity and stream depth
also peaked at or slightly below tributary confluences. The
range in peak values for these physical measures was high,
overlapping the 0 SDU line, which suggests other processes
within our study reaches contributed to physical habitat
heterogeneity. Disturbance regime, such as the timing, intensity,
and frequency of floods that move wood and sediment, can
affect the spatial distribution of these materials, potentially
obscuring or redistributing tributary inputs. In some loca-
tions along Bacon and Finney creeks, steep valley walls border
the river; local bank failure at these locations was a likely
process contributing to the spatial distribution of sediment
and wood. Bank failure was evident at the –6 SDU in the
Quartz–Hatchery–Finney creeks reach, which contributed
high volumes of wood to this transect. Other factors that can
affect the strength of the tributary signal include the ratio of
tributary to mainstem area, basin size and shape, drainage
and confluence density, and network geometry (Benda et al.
2004). In addition, tributary effects on mainstem habitat can,
in some cases, extend upstream of the confluence (Benda et
al. 2003).
The resulting effects of tributaries on physical habitat and
nutrients appear to create “biogeochemical hotspots” (McClain
et al. 2003). These hot spots, defined as areas or patches that
show disproportionately high reaction rates relative to the
surrounding area, occur most frequently at boundaries or
ecotones where two distinct landscape features meet, such as
where two streams of distinct water chemistries intersect
(McClain et al. 2003). Our results support this notion, as
tributary streams created distinct gradients in water chemistry
in mainstem rivers, which were generally nutrient-poor or
oligotrophic. Long-term surveys in the Cedar River have
consistently shown that concentrations of total and dissolved
nutrients were higher in tributaries entering the main stem,
with concentrations of these elements often remaining high
downstream of tributaries (Kiffney et al. 2002). The impor-
tance of these inputs as agents of chemical change in the
main stem, of course, is dependent upon the volume of water
in the tributary relative to the main stem.
We offer two potential mechanisms to explain why some
tributaries can create biogeochemical hot spots in mainstem
rivers. Most of the tributaries we surveyed were strongly
shaded by riparian forests, thereby limiting primary produc-
tion and algal accrual (Hill et al. 1995; Kiffney et al. 2003).
Low levels of primary production in tributary rivers may
limit uptake of nutrients, so that tributary nutrients are trans-
ported into sunlit, mainstem rivers where they fuel primary
production (Power and Dietrich 2002). Hill et al. (2001)
found that dissolved nitrogen and phosphorus concentrations
in stream water increased threefold, whereas aquatic primary
production declined threefold coincident with leaf emergence.
In addition, logging and natural disturbance events have led
to riparian forests in many headwater streams of the Pacific
Northwest to be composed mostly of red alder (Volk et al.
2003). Recent studies have demonstrated that streams drain-
ing red alder forests have higher levels of nitrogen and, in
some cases, phosphorus than nearby streams dominated by
conifers (Compton et al. 2003; Volk et al. 2003). Riparian
red alder dominated many of the tributaries we surveyed,
especially in the Cedar River. Therefore, these alder-
© 2006 NRC Canada
2526 Can. J. Fish. Aquat. Sci. Vol. 63, 2006
Variable
Correlation
coefficient
Large wood volume 0.72**
Substrate CV 0.60*
Velocity 0.22
Thalweg temperature –0.45
Thalweg depth 0.29
Canopy cover –0.54
Total phosphorus 0.85**
Total nitrogen 0.69**
Ambient ADFM –0.10
Ambient chlorophyll a0.42
Salmonid fry (≤50 mm) density 0.10
Larger salmonid (≥50 mm) density 0.03
Species × size classes 0.39
Simpson’s index –0.20
Note: Correlations were computed with zscores of vari-
ables and logit transformations of relative tributary area data.
Substrate CV, substrate coefficient of variation; ambient
AFDM, ash-free dry mass on rocks. *, p< 0.1; **, p< 0.05.
Table 2. Pearson’s correlation coefficents of vari-
ables with the relative tributary area, measured as
the proportion of total basin area flowing into a
study reach within the tributary.
dominated tributaries may have high nutrient loads as a
result of inputs of nutrient-rich leaf litter. If this is so, then
the condition (e.g., community composition) of riparian forests
in tributary streams may be tightly connected to the nutrient
dynamics of downstream portions of the watershed.
In agreement with our predictions and potentially in response
to peaks in nutrient resources, we observed that peaks in
both periphyton AFDM and chlorophyll aoccurred more
often at or slightly downstream of tributary junctions. Algal
biomass is extremely variable in both space and time in river
ecosystems because of the variety of factors that affect algal
communities (Biggs et al. 1998). For example, stream flow
(Peterson and Stevenson 1992), water temperature (Bothwell
1988), nutrients (Peterson et al. 1993), biotic interactions
(Hill et al. 1995), and habitat heterogeneity (Cardinale et al.
2002) contribute to variation in algal biomass, with these
factors likely interdependent. Our study supports this view
and suggests that tributary confluences may be locations
within the river network where these factors interact to affect
algal dynamics. For example, we found that total phosphorus
was positively correlated with chlorophyll density when
including all transects (P.M. Kiffney and C. Greene, unpub-
lished data). We speculate that greater standing stocks of
algal biomass may contribute to higher levels of consumer
(i.e., D. gilvipes abundance) and predator biomass and abun-
dance at tributary junctions by increasing resource supply.
Salmon abundance and diversity and sculpin abundance
peaked downstream of tributary junctions, potentially in
response to patterns of channel complexity and productivity.
These data support the hypothesis that tributary junctions
were nodal habitats for stream fishes, especially juvenile
trout and salmon and sculpin, during summer. A number of
interdependent mechanisms may contribute to these patterns.
Tributary streams funnel detritus and nutrients into mainstem
habitats, providing additional resources for invertebrates in
the main stem that sculpin and drift-feeding fish consume.
Invertebrates also drift out of headwater streams, providing
food for fish residing in the main stem (Wipfli and Gregovich
2002). Wood abundance or volume is also an important
predictor of fish abundance and diversity in streams (e.g.,
Johnson et al. 2005), as it serves a number of important
functions for fish, such as providing refugia and a surface
area for insect prey (Collier and Halliday 2000). We hypoth-
esize that the combination of higher productivity and habitat
complexity in mainstem rivers downstream of tributary junc-
tions was partially responsible for distribution patterns of
© 2006 NRC Canada
Kiffney et al. 2527
Fig. 8. Relationships of four variables as a function of the tributary area relative to the total amount of area upstream of each study
reach. Different basins are represented by different symbols: solid circles, Finney Creek; open circles, Bacon Creek; solid triangles,
Cedar River. One basin (Taylor Creek) was removed because of large differences of tributary stream order compared with the other
three basins.
stream fish. A recent study found that the diversity of electric
fishes (Gymnotiformes) was higher in the mainstem Amazon
River downstream of tributary rivers (Fernandes et al. 2004).
These authors also hypothesized that tributary rivers carry
nutrients and organic materials that provide food and habitat
resources for electric fishes in the main stem.
Caveats and implications
Although we observed that gradients in physical, chemical,
and biological attributes of river systems appeared to be cen-
tered on tributary junctions, other factors may account for
this variation. This study was observational, which limits
inferences because we could not disentangle mechanisms by
which tributary junctions influence communities. Our study
was confined to summer low-flow periods; therefore, the
biological patterns we observed might be different during
different seasons or years. For example, with the onset of the
rainy season, fish in the main stem may migrate into tribu-
tary streams to escape high flow events (Brown and Hartman
1988).
Despite these issues, this study has a number of important
implications for river ecology and conservation. One of the
salient findings of this study is it illustrates the fundamental
effects of tributary inputs on aquatic ecological processes.
Moreover, these effects appear to increase as a function of
the ratio of tributary to watershed area until this ratio reaches
15%. Given the strong patterns we found, researchers should
consider tributary inputs as a large source of ecological vari-
ation. If areas downstream of tributaries have elevated levels
of substrate, nutrients, and biological production, studies
overlooking these spatial patterns might come to erroneous
conclusions concerning common biological processes. We
provide three examples that illustrate how this might occur
(Fig. 9). In the first case (Fig. 9a), values measured down-
stream of tributaries obscured a common correlation between
water temperature and canopy cover that was evident in values
measured upstream (water temperature negatively correlated
with canopy closure; Kiffney et al. 2003), which is signifi-
cant if points at or downstream of tributaries are ignored (r=
–0.71, p< 0.05). In the second case (Fig. 9b), there was no
relationship between chlorophyll density and phosphorus
level (r= 0.02, p> 0.1) upstream of tributaries, but this rela-
tionship was significant downstream of tributaries (r= 0.60,
p< 0.05). The correlation across all sites is still significant
(r= 0.40, p< 0.05), driven primarily by high values down-
stream of tributaries. A final example (Fig. 9c) shows a positive
relationship between wood abundance and large salmonid
abundance and illustrates how extreme values associated
with tributaries might drive the strength of a correlation (r=
0.60, p< 0.05), which would not be significant in the
absence of those data points (r= 0.02, p> 0.1). These pat-
terns, therefore, illustrate that tributary inputs can obscure or
drive study results if researchers ignored their importance in
survey designs.
This study also supports the hypothesis that both produc-
tivity and habitat diversity were higher on mainstem reaches
at or downstream of tributary junctions. The abundance of
stream salmonids also peaked below tributary junctions,
potentially in response to increased habitat heterogeneity
and productivity. Based on the observed patterns of produc-
tivity and habitat diversity and the corresponding patterns of
© 2006 NRC Canada
2528 Can. J. Fish. Aquat. Sci. Vol. 63, 2006
Fig. 9. Examples of how tributary junctions can affect common
relationships found in aquatic systems. In each panel, solid circles
represent reach units downstream of tributary junctions, while
open circles represent units at or upstream from confluences.
(a) Input of warmer water from tributaries of Cedar River obscures
the correlation of thalweg temperature and canopy cover. (b)In
Bacon Creek, ambient chlorophyll density is correlated with total
phosphorus downstream but not upstream of tributary junctions.
(c) High densities of large (≥50 mm) salmonids were associated
with logjams downstream of tributary confluences on Finney
Creek, driving a significant salmonid – wood volume correlation.
fish distributions, habitat heterogeneity appears to be the
primary driver of fish communities. Habitat structure has
also been found to be an important factor shaping reef fish
communities (Syms and Jones 2000). However, fish likely
take advantage of both productivity and habitat heterogeneity
increases below tributaries by selecting downstream areas
that conveys the benefits of both gradients.
We argue that this study has important implications for
habitat conservation, as tributary steams appear to be impor-
tant sources of habitat heterogeneity in river networks. The
homogenization of habitat is thought to be one of the most
important factors driving loss of biodiversity (Cardinale et
al. 2002). A number of recent studies have shown that habitat
diversity in the form of the abundance of wood, size and
depth of pools, channel complexity, and substrate composi-
tion are higher where tributary streams enter mainstem rivers
(Rice et al. 2001; Benda et al. 2004; Fernandes et al. 2004).
In the Pacific Northwest and in other temperate regions,
75%–90% of all river kilometres are composed of tributary
streams, resulting in watersheds with a highly branched
appearance. If tributary streams are important drivers of habitat
and biological diversity in mainstem rivers, the degradation
of tributary habitat could have profound and lasting impacts
on habitat and biodiversity of the rivers they enter.
Our study provides evidence to suggest that the connec-
tion between tributary and main stem was important for the
flow of ecologically important materials (nutrients, sediment,
wood) that contributed to habitat heterogeneity and bio-
diversity in the main stem. A variety of organisms, such as
Pacific salmon, rely on a diversity of habitats within a water-
shed from headwaters to the estuary to maximize fitness. We
speculate that maintaining the integrity of connections
within river networks is critical to protecting and restoring
salmon populations. Our findings, therefore, suggest that
conservation plans for mainstem rivers include not only
riparian buffers but also entire drainages of large ecologi-
cally important tributaries entering main stems.
Acknowledgements
We recognize the involvement and support of the many
Earthwatch volunteers during 2003 and Ralph Riley,
Earthwatch Coordinator for the Skagit River Conservation
Initiative. Without their help, this research would not have
been possible. We also thank the Earthwatch Institute and
the Northwest Fisheries Science Center’s (NWFSC) internal
grant program for funding. K. Guilbault and A. Ritchey pro-
vided heroic assistance, as did a number of volunteers from
the NWFSC. Seattle Public Utilities provided access to sites
on the Cedar River, while Cascade Timberlands provided
access to Finney Creek.
References
Amoros, C., and Bornette, G. 2002. Connectivity and complexity
in waterbodies of riverine floodplains. Freshw. Biol. 47: 761–
776.
Benda, L., Veldhuisen, C., and Black, J. 2003. Debris flows as
agents of morphological heterogeneity at low-order confluences,
Olympic Mountains, Washington. GSA Bull. 115: 1110–1121.
Benda, L., Poff, N.L., Miller, D., Dunne, T., Reeves, G., Pess, G.,
and Pollock, M. 2004. The network dynamic hypothesis: how
channel networks structure riverine habitats. Bioscience, 54:
413–427.
Biggs, B.J.F., Stevenson, R.J., and Lowe, R.L. 1998. A habitat matrix
conceptual model for stream periphyton. Arch. Hydrobiol. 143:
21–56.
Bothwell, M.L. 1988. Growth rate responses of lotic periphytic
diatom communities to experimental enrichment of phosphorus:
the influence of temperature and light. Can. J. Fish. Aquat. Sci.
45: 261–270.
Brown, T.G., and Hartman, G.F. 1988. Contribution of seasonally
flooded lands and minor tributaries to the production of coho
salmon in Carnation Creek, British Columbia. Trans. Am. Fish.
Soc. 117: 546–551.
Bruns, D.A., Minshall, G.W., Cushing, C.E., Cummins, K.W., Brock,
J.T., and Vannotte, R.L. 1984. Tributaries as modifiers of the
river continuum concept: analysis by polar ordination and regression
models. Arch. Hydrobiol. 99: 208–220.
Cardinale, B.J., Palmer, M.A., Swan, C.M., Brooks, S., and Poff,
N.L. 2002. The influence of substrate heterogeneity on biofilm
metabolism in a stream ecosystem. Ecology, 83: 412–422.
Collier, K.J., and Halliday, J.N. 2000. Macroinvertebrate–wood
associations during decay of plantation pine in New Zealand
pumice-bed streams: stable habitat or trophic subsidy. J. North
Am. Benthol. Soc. 19: 94–111.
Compton, J.E., Church, M.R., Larned, S.T., and Hogsett, W.E.
2003. Nitrogen saturation in forested watersheds of the Oregon
Coast Range: the landscape role of N2-fixing red alder. Ecosys-
tems, 6: 773–785.
Darnaude, A.M., Salen-Picard, C., Polunin, N.V.C., and Harmelin-
Vivien, M.L. 2004. Trophodynamic linkage between river runoff
and coastal fishery yield elucidated by stable isotope data in the
Gulf of Lions (NW Mediterranean). Oecologia, 138: 325–332.
Fausch, K.D., Torgersen, C.E., Baxter, C.V., and Li, H.W. 2002.
Landscapes to riverscapes: bridging the gap between research
and conservation of stream fishes. Bioscience, 52: 483–498.
Fernandes, C., Podos, J., and Lundberg, J. G. 2004. Amazonian
ecology: tributaries enhance the diversity of electric fishes.
Science (Washington, D.C.), 305: 1960–1962.
Fisher, S.G. 1997. Creativity, idea generation, and the functional
morphology of streams. J. North Am. Benthol. Soc. 16: 305–
318.
Franklin, J.F., and Dyrness, C.T. 1969. Vegetation of Oregon and
Washington. United States Department of Agriculture, Forest
Service, Portland, Ore. PNW-80.
Hill, W.R., Ryon, M.G., and Schilling, E.M. 1995. Light limitation
in a stream ecosystem: responses by primary producers and con-
sumers. Ecology, 76: 1297–1309.
Hill, W.R., Mulholland, P.J., and Marzolf, E.R. 2001. Stream eco-
system responses to forest leaf emergence in spring. Ecology,
82: 2306–2319.
Johnson, S.L., Rodgers, J.D., Solazzi, M.F., and Nickelson, T.E.
2005. Effects of an increase in large wood on abundance and
survival of juvenile salmonids (Oncorhynchus spp.) in an Oregon
coastal stream. Can. J. Fish. Aquat. Sci. 62: 412–424.
Kaufmann, P.R. 2002. Physical habitat characterization. In Envi-
ronmental monitoring and assessment program — surface waters:
western pilot study field operations manual for wadeable streams.
Edited by D.V. Peck, J.M. Lazorchak, and D.J. Klemm. US
Environmental Protection Agency, Washington, D.C. pp. 103–
160.
Kiffney, P.M., Volk, C.J., and Hall, J. 2002. Community and eco-
system attributes of the Cedar River and tributaries before arrival
of anadromous salmonids. NMFS-Watershed Program, Technical
report submitted to Seattle Public Utilities, Seattle, Wash.
© 2006 NRC Canada
Kiffney et al. 2529
Kiffney, P.M., Richardson, J.S., and Bull, J.P. 2003. Responses of
periphyton and insect consumers to experimental manipulation
of riparian buffer width along headwater streams. J. Appl. Ecol.
40: 1060–1076.
Magurran, A. 1988. Ecological diversity and its measurement.
Princeton University Press, Princeton, N.J.
McClain, M.E., Boyer, E.W., Dent, C.L., Gergel, S.E., Grimm,
N.B., Groffman, P.M., Hart, S.C., Harvey, J.W., Johnston, C.A.,
Mayorga, E., McDowell, W.H., and Pinay, G. 2003. Bio-
geochemical hotspots and hot moments at the interface of terres-
trial and aquatic ecosystems. Ecosystems, 6: 310–312.
Minshall, G.W., Petersen, R.C., Cummins, K.W., Bott, T.L., Cushing,
C.E., and Vannote, R.L. 1983. Interbiome comparisons of stream
ecosystem dynamics. Ecol. Monogr. 53: 1–25.
Naiman, R.J., and Bilby, R.E. 1998. River ecology and manage-
ment in the Pacific Coastal Ecoregion. In River ecology and
management. Edited by R.J. Naiman and R.E. Bilby. Springer-
Verlag, New York. pp. 1–12.
Naiman, R.J., and Décamps, H. 1997. The ecology of interfaces —
riparian zones. Annu. Rev. Ecol. Syst. 28: 621–658.
Naiman, R.J., Décamps, H., Pastor, J., and Johnston, C.A. 1988.
The potential importance of boundaries to fluvial ecosystems. J.
N. Am. Benthol. Soc. 7: 289–306.
NOAA. 2005. Washington climate summaries [online]. Western
Regional Climate Center, Reno, Nev. Available from http://www.
wrcc.dri.edu/summary/climsmwa.html [accessed May 2006].
Peterson, B.J., Deegan, L., Helfrich, J., Hobbie, J.E., Hullar, M.,
Moller, B., Ford, T.E., Hershey, A., Hiltner, A., Kipphut, G.,
Lock, M.A., Fiebig, D.M., McKinley, V., Miller, M.C., Vestel,
J.R., Venutllo, R., and Volk, G. 1993. Biological responses of a
tundra river to fertilization. Ecology, 74: 653–672.
Peterson, C.G., and Stevenson, R.J. 1992. Resistance and resilience
of lotic algal communities: importance of disturbance timing
and current. Ecology, 73: 1445–1461.
Polis, G.A., Anderson, W.B., and Holt, R.D. 1997. Toward an inte-
gration of landscape and food web ecology: the dynamics of
spatially subsidized food webs. Annu. Rev. Ecol. Syst. 28: 289–
316.
Poole, G.C. 2002. Fluvial landscape ecology: addressing unique-
ness within the river discontinuum. Freshw. Biol. 47: 641–660.
Power, M.E., and Dietrich, W.E. 2002. Food webs in river net-
works. Ecol. Res. 17: 451–471.
Rice, S.P., Greenwood, M.T., and Joyce, C.B. 2001. Tributaries,
sediment sources, and the longitudinal organization of macro-
invertebrate fauna along river systems. Can. J. Fish. Aquat. Sci.
58: 824–840.
Richardson, J.S., and Mackay, R.J. 1991. Lake outlets and the
distribution of filter feeders: an assessment of hypotheses. Oikos,
62: 370–380.
Sabo, J.L., and Power, M.E. 2002. River–watershed exchange:
effects of riverine subsidies on riparian lizards and their terres-
trial prey. Ecology, 83: 1860–1869.
Seattle Public Utilities. 2000. Final Cedar River watershed habitat
conservation plan. City of Seattle, Wash.
Stevens, L.E., Shannon, J.P., and Blinn, D.W. 1997. Colorado River
benthic ecology in Grand Canyon, Arizona, USA: dam, tributary
and geomorphological influences. Regul. Rivers Res. Manag.
13: 129–149.
Syms, C., and Jones, G.P. 2000. Disturbance, habitat structure, and
the dynamics of a coral-reef fish community. Ecology, 81: 2714–
2729.
Thompson, W.L. 2003. Hankin and Reeves’ approach to estimating
fish abundance in small streams: limitations and alternatives.
Trans. Am. Fish. Soc. 132: 69–75.
Thurow, R.F. 1994. Underwater methods for study of salmonids in
the intermountain west. United States Department of Agricul-
ture, Forest Service, Boise, Idaho. INT-GTR-307.
USGS. 2006. Washington stream flows [online]. National Water
Information System. Available from http://waterdata.usgs.gov/
wa/nwis/current/?type=lakewash&group_key=NONE [accessed
June 2006; updated daily].
Vannote, R.L., Minshall, G.W., Cummins, K.W., Sedell, J.R., and
Cushing, C.E. 1980. The river continuum concept. Can. J. Fish.
Aquat. Sci. 37: 130–137.
Volk, C.J., Kiffney, P.M., and Edmonds, R.L. 2003. Role of riparian
red alder (Alnus rubra) in the nutrient dynamics of coastal
streams of the Olympic Peninsula, Wash., U.S.A. Am. Fish. Soc.
Spec. Publ. 34: 213–228.
Winterbourne, M.J., Rounick, J.S., and Cowie, B. 1981. Are New
Zealand ecosystems really different? N.Z. J. Mar. Freshw. Res.
15: 321–328.
Wipfli, M.S., and Gregovich, D.P. 2002. Export of invertebrates
and detritus from fishless headwater streams in southeastern
Alaska: implications for downstream salmonid production.
Freshw. Biol. 47: 957–969.
Wydoski, R.S., and Whitney, R.R. 2003. Inland fishes of Washington.
2nd ed. University of Washington Press, Seattle, Wash.
© 2006 NRC Canada
2530 Can. J. Fish. Aquat. Sci. Vol. 63, 2006