HYDRAULIC HABITAT PREFERENCES OF THE TORRENTIAL MAYFLY
Epeorus longimanus (Ephemeroptera: Heptageniidae): THE ECOLOGICAL
IMPORTANCE OF NEAR-BED FLOWS

by
Trent Hoover
B.Sc. (Sp.), The University of Alberta, 1994

THESIS SUBMITTED IN PARTIAL FULFILMENT OF
THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE
in
BIOLOGY
© Trent Hoover, 2001
THE UNIVERSITY OF NORTHERN BRITISH COLUMBIA

September 2001
All rights reserved . This work may not be reproduced in whole or in part,
by photocopy or other means, without the permission of the author.

UNJVERSiTV OF NORTHERN
BRITISH COLUMBiA

LIBRARY

Prince George, BC

Abstract
The larval stage of the mayfly Epeorus longimanus (Ephemeroptera: Heptageniidae) is an
inhabitant of torrential stream habitats.

It possesses several adaptations to high velocity

environments, including a flattened body shape, a tilted head shield, and a "sucker-like" arrangement
of abdominal gills. In order to relate the behavioural and morphological adaptations of E. longimanus
to its habitat requirements, the distribution of this mayfly and a suite of environmental parameters
were measured at a range of spatial scales (i.e. watershed, within-stream, and within-stone scales).
Benthic macroinvertebrates were collected at 39 stream sampling sites throughout the lower
portion of the Torpy River watershed in eastern British Columbia, in order to (1) identify the particular
habitat preferences of larval Epeorus, and (2) examine the community structure of benthic
macroinvertebrates in a northern watershed . Several environmental variables (e.g. discharge, stream
surface slope, substrate size, pH, conductivity, riparian vegetation, etc.) were measured and related
to patterns of faunal abundance using canonical correspondence analysis (CCA); this ordination
technique divided the collected invertebrate taxa into four functional assemblages, each with welldefined habitat requirements and trophic relationships. The results of the CCA showed that Epeorus
larvae prefer high discharge streams with coarse substrata.

To further examine the habitat

requirements of this genus, benthic samples and several descriptors of the flow environment of the
stream environment (e.g. mean velocity, near-bed velocity, depth, substrate size, Froude number,
etc.) were obtained at 50 regularly located sampling sites in two adjacent, high-discharge streams
with coarse bed material.

Observed patterns of abundance of Epeorus were significantly and

negatively related to velocity, depth, channel Reynolds number, and relative roughness.
The near-bed hydraulic environment of shallow torrential streams was characterised by
measuring velocity profiles, near-bed (U 0002 m} and mean (U 0 . 50 ) velocities, and shear stresses (-rw)
over the surface of five experimentally deployed and three naturally occurring stones in a highdischarge stream in the Torpy River watershed. The velocity profiles measured above the stones
regularly deviated from the "classic" log-normal shape . The profiles were often "wedge-shaped";
velocities were greatest a few millimetres above the bed, and decreased logarithmically below and
above this height. Wall shear stresses and near-bed velocities generally increased from the front to
the rear of each stone.

ii

The daytime and night-time distributions of E. /ongimanus were recorded and related to shear
stress, periphyton biomass, and substrate characteristics (e.g. stone roughness, topography). During
the daytime, larvae preferred areas of the stone surface with high shear stress ; during the night-time ,
larvae preferred areas of the stone surface with higher elevation and attached boundary layer flows.
Periphyton density was significantly related to stone surface roughness and stone surface
topography. A stone reversal experiment suggested that hydrodynamic factors , rather than food
(periphyton) availability, proximally influence the microdistribution of E. /ongimanus larvae; however,
the precise nature of the forces to which they respond remains unknown. E. longimanus larvae were
also found to exhibit a strong diurnal migration, generally migrating to the upper surface of streambed
stones at night, and retreating to the underside of the stones during the day.
This study represents one of the first detailed examination of the relationship between the
distribution of microscale hydrodynamic parameters, (e.g. shear stress, near-bed velocity) and
benthic organisms at organism-defined spatial scales. The results demonstrate that flu id dynamics
are the proximate factor that determines the microdistribution of benthic organisms in torrential
stream environments.

Additional research is required to investigate the ecological importance of

these small-scale hydrodynamic parameters. In order to understand the behaviour and ecology of
benthic stream organisms, models of flow in natural stream channels must be expanded to include
patterns of flow at small, organism-defined scales immediately adjacent to the bed .

iii

Table of Contents

Abstract ii
Table of Contents iv
List of Tables vii
List of Figures ix
Acknowledgements xiv
Chapter 1

The influence of fluid dynamics on freshwater invertebrates: morphological,

behavioural, and ecological adaptations
Summary 1
Introduction 2
Stream hydrodynamics 4
Mean flows in channels 4
Reynolds number 5
Froude number 6
Near-bed flows 7
The boundary layer 7
Shear velocity and shear stress 8
The influence of bed geometry on stream flows 13
Meso-scale patterns: the link between macro- and micro-hydraulics 17
Benthic-lotic coupling in the stream environment 18
Invertebrate drift: migration using bulk flow 23
Ontogenetic shifts in hydraulic habitat and microdistribution 25
Hydrodynamics, prey, and predation: eating in the fast lane 27
Fine-scale flows: behavioural and morphological hydrodynamics 31
Fine-scale manipulations of flow 31
The fluid mechanics of suspension feeding and 'erosional hunting' 33
Directions for future research 37
References 44

iv

Chapter 1 -Figures 50
Chapter 2 Environmental influences on macroinvertebrate distribution at reach and watershed
scales 54
Summary 54
Introduction 54
Methods 58
Results 64
Discussion 70
References 85
Chapter 2 - Figures 90
Chapter 3 Near-bed hydrodynamics above boulders in shallow streams 96
Summary 96
Introduction 97
Methods 98
Results 106
Discussion 118
References 131
Chapter 3 - Figures 134
Chapter 4

Microdistribution of Epeorus /ongimanus and periphyton on torrential stream

boulders 160
Summary 160
Introduction 161
Methods 165
Results 172
Discussion 178
References 191
Chapter 4 - Figures 195
Chapter 5 Torrential stream microhabitats: flow, form, and function 213
Summary 213
Physical- biological coupling in the stream environment 214

v

The relationship between mean and near-bed flows 214
Near-bed flows in torrential streams 215
Ecohydraulics of Epeorus /ongimanus 217
Stream hydraulics I invertebrate interactions 218
Directions for future research 221
References 223
Chapter 5 - Figures 225
Appendix 1 Formulae and notation 228
Appendix 2 Near-bed velocity profiles and wall shear stress (tw) estimation 230
Appendix 3 Ontological changes in Epeorus Jongimanus during the study period (Sept. - Oct.
1998) 233
Appendix 4 Preston-static tube 234
Appendix 5 Periphyton sampler 236
Appendix 6 Stone survey form 239

vi

List of Tables

Chapter 1

Table 1-1 . Selected examples of behavioural and morphological adaptations of stream invertebrates
to hydrodynamics 39-42
Chapter 2

Table 2-1.

Comparison of environmental factors between streams sampled in microhabitat

preference analysis 62
Table 2-2. Major environmental characteristics of the tributaries of the lower Torpy River watershed

65
Table 2-3. lntraset correlation coefficients of environmental variables with the first two axes of the
Canonical Correspondence Analysis (CCA) 68
Table 2-4. Summary of Canonical Correlation Analysis (CCA) of lower Torpy River watershed data
68
Table 2-5. Summary of ecological data for common macroinvertebrate taxa collected in tributaries of
the lower Torpy River 69
Table 2-6. Cumulative fit of Epeorus (Heptageniidae) abundance data against environmental axes
(canonical correspondence analysis) 70
Table 2-7. Simple linear and standard multiple regressions of Epeorus microdistribution and various
hydraulic and substrate characteristics 70
Chapter 3

Table 3-1.

Stone surface geometry and velocity profile shape categories for all sampled stone

surface locations 109
Table 3-2. Summary of flow conditions for all sampled stone surface locations 110
Table 3-3. Velocity profile shapes and regression parameters used in the estimation of U* and tw
(velocity gradient method} 111
Table 3-4. Water surface slope (S 5 ) parameters for the 5 deployed and 2 natural stones 113
Chapter 4

Table 4-1. Comparison of measured parameters at all sampling dates 170

vii

Table 4-2 . Post-hoc comparison of means (Tukey's HSD test) following the significant ANCOVA on
night-time distribution of Epeorus /ongimanus and streamwise (x) position 173
Table 4-3.

Multiple regression analyses of the relationship between the microdistributions of

periphyton and Epeorus /ongimanus (daytime and night-time distributions) and wall shear
stress (-rw) 174
Table 4-4. Multiple regression analyses of the relationship between the microdistribution of Epeorus
/ongimanus (daytime and night-time) and streambed characteristics 175

Table 4-5.

Multiple regression analyses of the relationship between the microdistribution of

periphyton to streambed characteristics and grazing mayfly densities 177

viii

List of Figures

Chapter 1

Figure 1-1. Reynolds number conditions for the occurrence of turbulent flows in open channels. 50
Figure 1-2. Representation of the boundary layer in shallow water. 50- 51
Figure 1-3. Diagram of the forces acting on a particle within a velocity gradient. 51
Figure 1-4. (a) Dorsal and (b) side views of a larval Sclerocyphon, showing the streamlined body
form and lateral laminae (reproduced from Smith and Dartnall 1980). 52
Figure 1-5. Diagram of the patterns of flow used by Pseudiron centra/is to excavate prey from sand
beds (reproduced from Vogel 1994; redrawn from Soluk and Craig 1990). 52
Figure 1-6.

Diagram of feeding posture adopted by larvae of the river mayfly Ametropus neavi

(reproduced from Soluk and Craig 1988). 53
Figure 1-7. Diagram of the flow patterns around the body of the black fly larva (Simulium vittatum)
(reproduced from Merritt et al. 1996). 53
Chapter 2

Figure 2-1. Taxonomic richness of macroinvertebrate taxa in relation to stream discharge. 90
Figure 2-2.

Conductivity versus stream discharge in tributaries of the lower Torpy River, British

Columbia. 91
Figure 2-3. Canonical correspondence analysis (CCA) biplot for the 24 benthic macroinvertebrate
taxa and the 11 environmental variables of the lower Torpy River, British Columbia. 91 - 92
Figure 2-4. Relationship between abundance of Epeorus larvae and channel depth (D) in streams 48
and 56 in the Torpy River watershed, British Columbia. 93
Figure 2-5. Relationship between abundance of Epeorus larvae and mean velocity in streams 48 and

56 in the Torpy River watershed, British Columbia. 93
Figure 2-6. Relationship between abundance of Epeorus larvae and channel Reynolds number (Re)
in streams 48 and 56 in the Torpy River watershed, British Columbia. 94
Figure 2-7. Relationship between abundance of Epeorus larvae and relative roughness (D/kave) in
streams 48 and 56 in the Torpy River watershed, British Columbia . 94

ix

Figure 2-8. Relationship between mean velocity (U) and depth (D) measured in streams 48 and 56,
located in the Torpy River watershed , British Columbia. 95
Figure 2-9. Relationship between near-bed velocity (Uo.o2m- velocity measured 2 em above the bed)
and mean velocity (U- measured at 0.6D). 95
Chapter 3

Figure 3-1 . Map of British Columbia, showing the location of the Torpy River watershed. 134
Figure 3-2. Diagrammatic representation of the grid system and deployment of the experimentally
deployed stones. 134
Figure 3-3. Stream bed and water surface topography of experimentally deployed stone 7 (centreline
transect) measured on Sept. 5, 1998. 135
Figure 3-4.

Stream bed and water surface topography of experimentally deployed stone 31

(centreline transect) measured on Sept. 2, 1998. 135
Figure 3-5.

Stream bed and water surface topography of experimentally deployed stone 25

(centreline transect) measured on Sept. 8, 1998. 136
Figure 3-6 .

Stream bed and water surface topography of experimentally deployed stone 23

(centreline transect) measured on Sept. 30, 1998. 136
Figure 3-7.

Stream bed and water surface topography of experimentally deployed stone 10

(centreline transect) measured on Sept. 30, 1998. 137
Figure 3-8. Stream bed and water surface topography of naturally positioned stone N01 (centreline
transect) measured on Sept. 26 , 1998. 137
Figure 3-9 . Stream bed and water surface topography of naturally positioned stone N02 (centreline
transect) measured on Sept. 27, 1998. 138
Figure 3-10. Stream bed and water surface topography of naturally positioned stone N03 (centreline
transect) measured on Aug . 7, 1998. 138
Figure 3-11 .

Stream bed and water surface topography of the gravel riffle site (GR) measured on

Aug. 17, 1998. 139
Figure 3-12. Velocity profiles obtained along the centreline of stone 7. 140
Figure 3-13. Velocity profiles obtained along the centreline of stone 31 . 141
Figure 3-14. Velocity profiles obtained along the centreline of stone 25. 142
Figure 3-15. Velocity profiles obtained above the centres of stone 23 and stone 10. 143
X

Figure 3-16. Velocity profile obtained above the centres of naturally positioned stones N01, N02, and
N03, and the gravel riffle site. 144
Figure 3-17. The distribution of wall shear stress (•w) over the upper surface of stone 7. 145
Figure 3-18. The distribution of wall shear stress (•w) over the upper surface of stone 31. 146
Figure 3-19. The distribution of wall shear stress (•w) over the upper surface of stone 25. 147
Figure 3-20. The distribution of wall shear stress (•w) over the upper surface of stone 23. 148
Figure 3-21 . The distribution of wall shear stress (•w) over the upper surface of stone 10. 149
Figure 3-22. The distribution of wall shear stress (•w) over the upper surface of naturally positioned
stone N01. 150
Figure 3-23. The distribution of wall shear stress (•w) over the upper surface of naturally positioned
stone N02. 151
Figure 3-24. Relationship between wall shear stress measured by the PST (•w-PsT) and wall shear
stress estimated from shear velocity based on the slope of dU I dlnz (•w-dU/dlnz). 152
Figure 3-25. Relationship between wall shear stress measured by the PST (•w-PsT) and wall shear
stress estimated from shear velocity based on the slope of dU I dlnz (•w-dU/dlnz) (data
separated by velocity profile type). 153
Figure 3-26 .

Comparison of the velocity profiles obtained over naturally positioned boulders,

experimentally deployed boulders, and a gravel riffle (velocities are normalised by shear
velocity (U*) as measured by the PST). 154
Figure 3-27.

Comparison of the velocity profiles obtained over naturally positioned boulders,

experimentally deployed boulders, and a gravel riffle (velocities are normalised by velocities
at mid-depth (Uo.so) ). 155
Figure 3-28 . Relationship between Froude number (Fr) and local Reynolds number (Rex) at velocity
sampling locations over the upper surface of the stones (including all experimentally
deployed and naturally positioned stones). 156
Figure 3-29. Relationship between Froude number (Fr) and shear velocity (U*). 157
Figure 3-30. Relationship between shear velocity (U*; measured using the PST) and water velocity,
at mid-depth (U 0 .50 ) and near the bed (Uo.oo 2m). 158

xi

Figure 3-32.

Diagrammatic representation of the flow field above isolated boulders in shallow

streams (D/dk < 2). 159
Chapter 4

Figure 4-1. Diagrammatic representation of the upper surface of a stone, showing the distribution of
each of the types of factors sampled. 195
Figure 4-2A. E. longimanus larvae density among stones, and between day- and night-time periods
196
Figure 4-28. E. longimanus larvae density in areas of the substrate with different boundary layer flow
conditions. 197
Figure 4-3. Relationship between daytime and night-time microdistributions of E. longimanus larvae.
198
Figure 4-4. The distribution of E. longimanus larvae (day and night) along the centre line of the upper
surface of the experimental stones. 199
Figure 4-5. Relationship between E. longimanus density (day and night) and wall shear stress (-rw).
200
Figure 4-6. Distribution of larval E. longimanus density, periphyton density, and shear stress (•w)
along the centreline transect of stone 7. 201
Figure 4-7. Distribution of larval E. longimanus density, periphyton density, and shear stress (•w)
along the centreline transect of stone 31. 202
Figure 4-8 . Distribution of larval E. longimanus density, periphyton density, and shear stress (-rw)
along the centreline transect of stone 23 . 203
Figure 4-9. Distribution of larval E. /ongimanus density, periphyton density, and shear stress (-rw)
along the centreline transect of stone 10. 204
Figure 4-10. Distribution of larval E. longimanus density, periphyton density, and shear stress (-rw)
along the centreline transect of stone 25. 205
Figure 4-11 . Periphyton (chlorophyll a) densities along the centre line of the upper surface of the
experimentally deployed stones. 206
Figure 4-12.

Periphyton density in areas of the bed (i.e. experimentally deployed stones) with

different near-bed flow conditions. 207

xii

~----------------------

Figure 4-13.

---

Densities of E. /ongimanus larvae on the upper surface of the five experimentally

deployed stones during September 26-27, 1998. 208
Figure 4-14 . Densities of E. longimanus larvae on the upper surface of the five experimentally
deployed stones during October 3 - 4, 1998. 209
Figure 4-15. Change in E. /ongimanus density recorded on the upper surface of the experimentally
deployed stones during October 3 - 4, 1998. 21 0
Figure 4-16 . The effect of stone reversal on the positioning of E. longimanus larvae. 211, 212
Chapter 5

Figure 5-1 . The relationship between mean and near-bed flows. 223
Figure 5-2 . Diagrammatic representation of the flow field above isolated roughness elements in
shallow, torrential streams (where D <3d). 224
Figure 5-3.

Diagram of the direct and indirect relationships between flow, Epeorus longimanus,

primary producers, and predator abundance. 225

xiii

Acknowledgements
I would like to thank and acknowledge my wife and intellectual inspiration, Shelley. Without
her help, I would have long since gone mad.
In addition, I would like to extend thanks to the following people who helped me along the
path, in ways both large and small : Joe Ackerman, Max Blouw, Steve Bruce, Grace Cho, Doug Craig ,
Michelle Dube, Terri Fortune, Dan Gable, Laura Genn, Allan Gottesfeld, Glen and Hope Hoover (my
parents), Dev Khurana, Jacquie Lee, Staffan Lindgren, Jen McConnachie, Mike Nishizaki, Ellen
Petticrew, Mark Shrimpton, Steve Springer, Jill Sutton , Steven Vogel, and Noboru Yonemitsu . Your
interest and assistance were appreciated .

xiv

Chapter 1

The influence of fluid dynamics on freshwater invertebrates:
morphological, behavioural, and ecological adaptations

Summary
1.

Hydrodynamic forces are potentially the most important factors influencing the distribution,
biology, and trophic interactions of benthic invertebrates in lotic environments.

2.

Mean flows can be characterised by a series of directly measured hydrodynamic variables (i.e.
mean velocity, depth, discharge) and parameters (Reynolds number, Froude number). Reynolds
number and Froude number are of value in comparing flows between sites, as they are nondimensionalised indices describing flow regime and surface flow conditions, respectively.

3. Measurements of flows immediately adjacent to the bed are relatively difficult to obtain, and as
such, are rarely encountered in the literature. However, because they more closely represent the
flows that benthic invertebrates experience, characterisation of fine-scale flows immediately above
a substrate provides greater insight into the microhabitat preferences of stream invertebrates than
do descriptors of mean flows. Fine-scale flow descriptors include shear velocity, shear stress,
roughness Reynolds number, boundary layer thickness, and the thickness of the viscous
sublayer. The importance of quantifying stream flow at several spatial scales is discussed.
4.

The microhabitat preferences of benthic stream invertebrates are closely related to the
hydrodynamic characteristics of the environments they inhabit. Flow microenvironments influence
the predator-prey interactions, movement, and feeding efficiency of benthic organisms. Generally,
primary consumers possess specialised morphological and behavioural adaptations to specific
hydraulic habitats, while predators tend to be 'hydraulic generalists'.

5.

Many invertebrates utilise the current to migrate downstream ('drift'). The numbers of individuals
drifting generally peaks at night, suggesting that this behaviour may be an adaptation to search for
more favourable microhabitats while avoiding visually oriented fish predation . However, sizerelated increases in drift suggest that entry into the drift is not always active, and may, in some
cases, be the result of the erosion of invertebrates from the bed during times of increased activity.

6.

Stream invertebrates, especially those inhabiting high-velocity habitats, often demonstrate a shift
in microhabitat preferences as they grow. This type of ontogenetic shift may be largely due to the
changes in hydrodynamic forces (i.e. lift, drag, and shear stress) that larvae experience as they
increase in size.

7. In order to maintain contact with the substrate, benthic invertebrates that inhabit areas of the
streambed exposed to high velocities often exhibit marked morphological adaptations that
manipulate local flow fields to reduce the erosional effects of lift and drag. Adaptations include
flattened bodies that concentrate lift over the legs and body shapes that delay boundary layer
separation.
8.

Suspension-feeding invertebrates (generally Trichoptera and simuliid Diptera) depend on suitable
near-bed flow velocities to ensure an adequate supply of seston . Due to the intimate relationship
between flow velocity and suspension-feeder ecology, suspension-feeding taxa often exhibit
marked morphological adaptations for specific flow regimes.

The behavioural adaptations of

mayfly species that manipulate flow fields to excavate prey or suspension-feed are also
discussed.

Introduction
The physical forces created by moving water are recognised as the most important factors
shaping the benthic community in rivers and streams. The ever-present, generally unidirectional flow
of streams has shaped the morphology (Weissenberger et al. 1991; Collier 1994; Pommen and Craig
1995), behaviour (Osborne and Herricks 1987; Soluk and Craig 1990; Wetmore et al. 1990;
Holomuzki and Messier 1993), diversity (Growns and Davis 1994; Quinn and Hickey 1994) and
community structure (Lancaster and Hildrew 1993) of the invertebrate fauna which inhabit this
complex environment.
The distributions of benthic stream fauna are spatially and temporally variable.

Many

macroinvertebrate species migrate to the underside of streambed stones during the daylight hours,
and return to the exposed, upper surfaces of the stones during the night (Allan et al. 1986; Peckarsky
1996).

Examination of the daytime and night-time distributions of benthic fauna reveals species-

specific patterns of abundance, which are in turn related to a variety of physical and biological factors .
While many taxa demonstrate marked preferences for sheltered, depositional habitats (Bouckaert and

2

Davis 1999), others prefer exposed conditions (e.g. the upper surfaces of stones), where they are
subjected to high velocities, high shear stress, and relatively thin boundary layers (Nowell and Jumars
1984).

While it is more expensive energetically to move about and forage in such conditions,

invertebrates that inhabit this type of extreme microhabitat may benefit from more abundant algal food
resources (Quinn et al. 1996), greater oxygen availability (Wiley and Kohler 1980; Golubkhov et al.
1992), and decreased predation risk (Peckarsky et al. 1990; Hansen et al. 1991; Hart and Merz 1998).
As such, the benthic macroinvertebrates that inhabit lotic systems are subject to an ecological
dilemma. Benthic organisms benefit from hydraulic conditions that maintain interstitial microhabitats,
elevate oxygen levels, reduce predation, and provide an energetically inexpensive method of
downstream movement. These same hydrodynamic forces, however, are those acting to dislodge
grazers from feeding sites, erode the streambed, and limit feeding opportunities of both predators and
herbivores.

The morphological and behavioural adaptations that have evolved in benthic fauna

ensure that the ecological benefits of inhabiting these extreme habitats do not outweigh the costs .

Thesis overview
In this study, manipulative field experiments and surveys of naturally occurring benthic
communities in a montane watershed in east-central British Columbia were used to examine the
relationship between the distribution of stream invertebrates and physical factors across a range of
spatial scales. Specifically, the ecological preferences of a single stream invertebrate, the torrential
mayfly Epeorus longimanus (Ephemeroptera: Heptageniidae), were examined in detail. Mayflies of
the genus Epeorus are grazers, and feed largely on growths of periphyton that encrust the upper
surface of stones in high-velocity, mountainous streams (Resh and Rosenberg 1984). Three separate
studies were used to examine the habitat requirements of this mayfly.

(1) The large-scale (i.e.

watershed- to stream-scale) habitat preferences and community associations of Epeorus spp. were
examined in the tributaries of the lower portion of the Torpy River watershed using canonical
correspondence analysis, a multivariate ordination technique. (2) The reach-scale habitat preferences
of Epeorus spp. were measured by comparing local densities of this mayfly with a range of variables
that describe channel and flow conditions (e.g. depth, mean velocity, and bed roughness). (3) The
within-stone microhabitat preferences of E. /ongimanus were further characterised by comparing the
microdistributions of mayfly larvae, near-bed water velocity, bed shear stress, periphyton, and several

3

stone surface variables (e.g. stone roughness) over the surface of five experimentally deployed and
three naturally occurring streambed stones. These measurements allowed for an assessment of the
factors that determine the distribution of torrential mayflies in high-gradient, mountainous streams.

Stream hydrodynamics
When describing flow in open channels, physical scientists commonly employ a large number
of hydrodynamic parameters and ratios. As such, the task of the stream ecologist interested in the
interactions between moving water and the community structure, behaviour, or functional morphology
of benthic invertebrates can be challenging, as much of the existing literature provides little insight into
which measurements are biologically or ecologically relevant.

In addition, the technology for

measuring flows at small spatial scales in the field is often expensive and/or difficult to employ. Not
surprisingly, these difficulties have resulted in some misinterpretation of hydrodynamic phenomena
(see Gore 1978; Buffagni et al. 1995). Fortunately, there has been an increased interest in these
issues (e.g. Collier 1994; Hart et al. 1996; Bouckaert and Davis 1999), and several excellent reviews
on the flow microenvironment of streams and rivers have been published (Nowell and Jumars 1984;
Davis 1986; Davis and Barmuta 1989; Carling 1992; Hart and Finelli 1999).

Mean flows in channels
The movement of water is the most dominant physical feature in rivers and streams (Statzner
and Higler 1986; Davis and Barmuta 1989), yet only relatively recently have ecologists begun to
examine the biology and community ecology of benthic fauna in the context of relevant hydrodynamic
theory (but see Ambuhl 1959). However, before one can characterise the flows experienced by
macroinvertebrates at organism-defined spatial scales, an understanding of the basic principles of the
mean motion of water in channels is essential.

Even in straight, flat-bottomed channels, flow is

complex and highly three-dimensional (White 1999). As streams are rarely, if ever, straight, and
generally have rough, irregular beds, flows are spatially and temporally variable, and difficult to model
(Young 1992; 1993; Hart et al. 1996). As a result, characterisation of flows in natural channels is
usually achieved through extensive measurement.

The most basic description of the stream

environment necessarily includes mean velocity (U), channel depth (D), discharge (Q), and the wetted
perimeter (P).

An indication of turbulence intensity can be obtained by recording high-frequency

4

variations in velocity (u'); however, the response times of most instruments currently used to measure
velocity in streams (e.g. propeller velocimeters) are too low to make such measurements meaningful
(see Bouckaert and Davis 1998).

Reynolds number
The mean motion of flow in streams can be characterised by the Reynolds number (see Davis
and Barmuta 1989). Analytic solutions (e.g. Navier-Stokes) that describe boundary layer flows do not
exist for the conditions that are found within naturally occurring lotic systems, especially rivers and
streams with irregular bed geometry and water surface topography. Solutions of the Navier-Stokes
equation exist only for steady, uniform flows, where depths of flow do not vary over time, and shape,
size, roughness, and slope of the channel are constant in the y direction. Such conditions are rarely,
if ever, found in natural systems. Yet, the Navier-Stokes equation allows for a description of the flow
in any given circumstance when inertial effects predominate over viscous effects, or vice versa. The
resultant ratio of inertial to viscous forces is known as the Reynolds number (Re), and is an indicator
of the turbulent properties of flow. Reynolds number is determined as

Re = Ul/u

(1-1)

where U is velocity, I is the length scale, and u is the kinematic viscosity of water.

When

characterising flow in open channels, U is the mean channel velocity (generally measured at 60% of
the total depth; see White 1999), and the length scale I is the total channel depth or the hydraulic
diameter. Where calculated Reynolds numbers are less than unity (see figure 1-1 ), viscous forces
dictate flow behaviour, and the influence of inertia can be neglected; such flow is said to be laminar.
Streamlines are smooth and linear, and the water apparently 'slides' in layers (Carling 1992).
However, when Reynolds numbers exceed unity, inertial effects predominate, and the immediate
effects of viscosity can be neglected (Vogel 1994 ). Flow in this case is said to be turbulent, and is
characterised by irregular, secondary motions. However, the critical value at which flow becomes
turbulent varies depending on the frame of reference and the conditions of the boundary. In open
channel flows, Reynolds numbers less than 500 suggest laminar flow, and the transition to fully
turbulent flow occurs between Re = 10,000 and Re = 100,000. With the exception of extremely
5

shallow streams of water {i.e. a sheet of water sluicing over bedrock) or unusually slow streams {i.e.
shallow, nearly stagnant channels), mean flow in natural systems is nearly always turbulent. Flows
where the Reynolds number falls between 500 and 10000 are said to be transitional, and may
oscillate between turbulent and laminar flow. However, the range of Reynolds numbers at which flows
become turbulent is dependent on the conditions of the boundary.

Streams whose beds are

composed of coarse substrata make the transition to turbulence at lower Reynolds numbers than
streams and rivers whose bottoms are made up of sand or fine gravels (Vogel 1994 ). In flow past the
different regions around a submersed object, both laminar and turbulent flow may be present
(Silvester and Sleigh 1985; Vogel 1994).

Froude number
The mean motion of flow in streams can also be described using the Froude number. The
Froude number (Fr) is a dimensionless ratio of inertial to gravitational forces, and is calculated as

Fr = U/(gD)0 ·5

(1-2)

where U is mean velocity, g is acceleration due to gravity, and D is channel depth. Froude number is
used to distinguish between the two predominant flow regimes in moving water. Froude numbers less
than unity (Fr < 1) are seen in flows which are characterised by deep, smooth flowing water. This is
termed subcritical, or tranquil, flow. Flow disturbances such as surface waves are propagated both up
and downstream in subcritical conditions. Froude numbers greater than unity (Fr > 1) exist in flows
that are characterised by broken, white water. This is alternately termed supercritical, or shooting,
flow. In these circumstances flow disturbances are propagated only downstream. The transition from
supercritical flow to subcritical flow is often accompanied by an abrupt elevation of the water surface
known as a hydraulic jump (a form of shock that occurs when downstream depths are too great to
maintain supercritical flow; see White 1999). Where flows accelerate from subcritical to supercritical
flow, an abrupt decrease in surface elevation known as a hydraulic drop can also be seen.
Reynolds number and Froude number are widely used to describe flow in open channels
(Vogel 1994 ). Although their use has been criticised by some workers as inappropriate (in that they
describe mean rather than near-bed flows; e.g. Hart et al. 1996), Reynolds number and Froude

6

number often correlate highly with distributions of benthic macroinvertebrates (Gore 1978; Wetmore et
al. 1990; Buffagni et al. 1995). Reynolds number and Froude number are also of crucial importance in
laboratory-based research, where flumes are employed to simulate the stream environment.

An

accurate reproduction of field conditions relies heavily on the measurements and design of
experimental flumes (Lacoursiere and Craig 1990); well-designed flumes provide simplification of flow
conditions so that flows can be "summarized from their constituent parts" (Craig 1993).

Near-bed flows
Flows moving over, through, and around the substratum of streams are very complex.
Patterns of bulk flow, such as the plunging jet upstream of a hydraulic jump (where Fr > 1) or the fully
turbulent flows of a bouldery, torrential stream (where Re > 100,000) are readily observed, measured,
and understood. More difficult to observe are the complex, fine-scale patterns of flow that occur at the
interface between streambed and moving water. Stream organisms exist almost exclusively in a zone
that extends only a few millimetres above the surface of the substrate, and it is in this zone that the
majority of the biotic processes in streams take place.

The boundary layer
As the water moves over the surface of the streambed, friction is created between the solid
substrate and the moving water immediately adjacent to the surface. As a result, water immediately
adjacent to the boundary has a velocity of zero. This 'no-slip' condition (referring to the fact that when
a fluid moves over a solid, the fluid tends to shear above the boundary rather than 'slip' over the
surface) results in a characteristically logarithmic velocity gradient above the bed.

Although the

velocity gradient is generally steepest immediately adjacent to the surface of the substrate, the
velocity-retarding influence of the boundary may extend to the surface of the water in shallow flows,
while in deeper channels the flow structure far from the bed may be entirely unaffected.

This

boundary-influenced layer of water is termed the boundary layer (see Figure 1-2). The thickness of
the boundary layer (o) has been variously defined as the distance from the boundary at which water
velocity reaches 90% (Silvester and Sleigh 1984; Vogel 1994) or 99% (Prandtl 1952; Jumars and
Nowell 1984; White 1999) of the mean velocity. There is no sharply defined delineation to the outer
limit of the boundary layer, as the transition from viscid to inviscid flow is continuous (Silvester and

7

Sleigh 1984). The thickness and flow conditions of the boundary layer (see Figure 1-2) are influenced
by a number of physical factors, including water depth and velocity, channel slope, substrate
roughness , and the dynamic viscosity of the fluid.
Flow conditions within the boundary layer can be described by the local Reynolds number,

(1-3)

Rex = U.xlu

where u. represents velocity at the distance downstream (x) over which the boundary layer has
developed (usually the distance from the leading edge of a flat plate), and u is kinematic viscosity.
Relatively large local Reynolds numbers indicate correspondingly thin boundary layer thicknesses.
However, while the rate and nature of boundary layer growth over flat plates with well-defined leading
edges are well understood (White 1999), the growth of boundary layers over rough bottoms in natural
systems is not (see Bathurst 1994; Buffin-Belanger and Roy 1999). Over the topographically rough
bottoms of natural streams, boundary layers repeatedly grow, detach, reattach, and grow again as
water flows downstream. As a result of the regular interruption of boundary layer growth, boundary
layers in gravel/cobble streams rarely, if ever, reach an equilibrium thickness (Nowell and Jumars
1984; White 1999). Unfortunately, measurement of boundary layer conditions in naturally occurring
systems remains relatively rare. With the exception of the fine-scale measurements obtained by Hart
et al. (1996), the majority of studies on near-bed flow environments have failed to measure water
velocities close enough to the bed to describe the hydraulic habitats experienced by stream benthic
fauna .

Shear velocity and shear stress
Due to the exaggerated deformation of the velocity profile near the bottom, the boundary layer
can also be regarded as a zone of shearing flow (see Figure 1-3). Shear velocity (U*), identified as an
ecologically important factor by several workers (e .g. Carling 1992), is related to the shear stress
acting on the bed . Shear stress (or, wall shear stress t w), is defined as

(1-4)

8

where the square of shear velocity (U*) is multiplied by the density of the fluid (p ). The relationsh ip
between velocity (u) and shear velocity (U*) within the boundary layer is described by the von Karman
- Prandtllaw of velocity distribution, or the law of the wall (see Bergeron and Abrahams (1992) for a
complete discussion),

Uz

where

Uz

=U*/K ln[(z-d)/Zo]

(1-5)

is velocity at height above the substrate z, K is von Karman's constant (empirically

determined to be 0.4), d is the "zero-plane displacement", and Zo is the bed roughness length (see
Appendix 2).

However, while the ecological significance of shear velocity and shear stress are

recognised, they are rarely determined in field studies . The relatively rare appearance of U* in the
literature is due, in part, to its method of determination.

Estimates of shear velocity are typically

derived from the slope obtained by regressing velocity (u) on the logarithm of height above the bed (In
z),

(1-6)

where K is von Karman's constant, u1 and u2 are velocities at distances from the boundary z 1 and z2 .
This relationship can alternatively, be expressed as

U* = K m

(1-7)

where m is the slope of the regression of velocity (u) on the log of the distance from the boundary (z) .
Several measurements of velocity in the log-layer adjacent to the boundary are required to obtain an
accurate velocity profile and statistically accurate estimate of U*. Moreover, equations 1-5 and 1-6
can only be applied in the log-layer where the plot of U on In z is linear. Obtaining a sufficient number
of velocity measurements can be difficult; the few instruments capable of measuring flows at the fine
spatial resolution required are extremely difficult to deploy close to the streambed in shallow, naturallyoccurring channels (see Hart et al. 1996).
9

Davis and Barmuta (1989) reported that a reasonable estimate of the shear velocity U* could
be obtained from measurements of the relative bed roughness (a ratio of channel depth D to the
height of the roughness elements k) and mean velocity U.

(1-8)

U/U* = 5.75 log(12 D/k)

Unfortunately, this method of estimating shear velocity is limited to applications where only reachaveraged values of U* are desired, and is probably of little value in determining U* at the small spatial
scales necessary to characterise the hydraulic habitats of benthic invertebrates.

The principal

variables in Equation 1-8 (i.e. U, D, and k) are descriptors of average, rather than near-bed, channel
characteristics. Hart et al. ( 1996) found that the velocity 10 em above the surface of stream cobbles
was a poor predictor of velocities near the surface, making any estimation of near-bed flow
characteristics from mean channel velocities suspect.

In addition, Carling (1992) states that for

natural channels, estimates of roughness factors are related both to the size and packing
characteristics of gravel or cobble streambeds . Consequently, determinations of shear velocity (U*)
are best made from velocity profile data (using Equation 1-6), rather than estimation methods (as in
Equation 1-8). Higher values of U* will indirectly influence the growth of both algal and invertebrate
taxa by increasing the fluxes of dissolved materials in near-bed flows. Higher shear velocities result in
more rapid diffusion of nutrient supplies to periphytic diatoms (see Vogel 1994 ), and higher oxygen
supply to mayflies such as Epeorus (Palmer 1995) and Ecdyonurus (Buffagni et al. 1995); many
benthic organisms thrive in high velocity, exposed microhabitats (Vogel1994).
The near-bed flows to which benthic invertebrates are exposed can also be described using
the Reynolds roughness number (Re*), an index of turbulence at the fluid-boundary interface. The
Reynolds roughness number is given as

Re*= U*k5 /v

(1-9)

where U* is shear velocity, k5 is Nikuradse's roughness factor, and v is kinematic velocity. The
viscous sublayer remains intact when Re* < 3.5; such flows are said to be smooth-turbulent (Carling
1992; see also Davis and Barmuta 1989). When the values of Re* fall between 3.5 and 68 (3.5 and
10

100 in Vogel 1994 }, near-bed flows are in a transitional state, and the viscous sublayer develops
intermittent turbulent disruptions. When the bed roughness exceeds the theoretical height of the
viscous sublayer (see equation 1-10 below) or shear velocities (U*) become too great, the viscous
sublayer disappears, and hydraulically rough-turbulent flow occurs (Re* > 68) (Carling 1992). The
thickness of the viscous (or laminar) sublayer (8'} can be estimated using the following formula,

8' = 11.6 vI U*

(1-10}

where v is kinematic viscosity, and U* is shear velocity (Carling 1993). Ambuhl, in his seminal work
on the microhabitats of stream insects (1959), suggested that species living in high velocity
environments were, in fact, living in the viscous sublayer, experiencing only minimal turbulence and
relatively low velocities. While this represented a major advancement in behavioural hydrodynamic
theory, his findings were not as widely applicable as he believed . In most naturally occurring flows,
the viscous sublayer is absent (Carling 1992), and near-bed flows are fully turbulent.

In addition,

Statzner and Holm (1982) found that boundary layers were much thinner, relative to the body height of
the mayfly Ecdyonurus, than Ambuhl had reported (see below). Statzner and Holm suggested that
the conflicting results were due, in part, to the relatively large size of the tracer particles used by
Ambuhl (larger particles may be unsuitable for detecting the fine-scale details of velocity gradients in
the boundary layer). Staztner and Holm (1982; 1989) demonstrated that the patterns of flow around
the bodies of benthic insects are more complex than previously thought; rather than being protected
from the forces of the current by a viscous sublayer or thick boundary layer, velocity gradients are
actually steeper over the bodies of the insects where they protrude into the flow .
Organisms that inhabit streambed surfaces that are exposed to moving water must be able to
maintain contact with the substrate. However, the velocity gradients in the microhabitats preferred by
many benthic stream organisms are often quite steep (see figure 1-3). As a result, the forces of lift
and drag acting to dislodge individual invertebrates can be quite substantial. The force due to drag
(F ct) is defined as

(1-11)

11

where Cd is the drag coefficient (a function of shape, orientation, and Reynolds number), p is the
density of the fluid, S is surface area (alternatively defined as the frontal area of an organism;
Weissenberger et al. 1991 ), and U is velocity (Vogel 1994). The biological validity of this relationship
was demonstrated by Weissenberger et al. (1991 ), who found that drag acting on three benthic
macroinvertebrate species was proportional to the square of near-bed velocity. However, as noted by
Vogel (1994), application of Equation 1-11 to estimate the drag experienced by benthic organisms
must be approached with care; the drag coefficient (Cd) is not a constant but rather a dimensionless
form of drag (drag per unit area divided by the dynamic pressure) and varies with velocity. At the
relatively high Reynolds numbers (Re ~ 1000, where the characteristic length is body length)
experienced by macroinvertebrates inhabiting high velocity habitats, flow separation generally occurs
somewhere along the body of the invertebrate. As a result, dynamic pressure differences between
front and rear are substantial (the energy of decelerating fluid in the turbulent wake at the rear is lost
through viscous damping and dissipated as heat) and the total drag is almost entirely due to pressure
drag rather than skin friction (Vogel 1994).
The strong velocity gradients inhabited by many benthic invertebrates also generate lift forces
acting to dislodge organisms from the substrate. The force due to lift (F 1) can be calculated as

(1-12)

where the variables p and U represent the density of water and velocity, C1 is the coefficient of lift
(again, a function of shape, orientation, and Reynolds number), and S represents area (Vogel 1994).
Statzner and Holm (1982; 1989) found that as flow crossed the bodies of various organisms it was
compressed, creating even stronger velocity gradients over the organism. Statzner and Holm (1982)
noted that maximum compression occurred over the thorax of Ecdyonurus venosus, and interpreted
this as an adaptation to concentrate the forces of lift over the legs of the organism. Weissenberger et
al. (1991) noted that the generally 'airfoil'-like form of many aquatic invertebrate taxa (e.g. the mayfly
Epeorus) generate substantial lift forces, and also found that the mayfly Ecdyonurus was able to

actively reduce lift by altering its body orientation.
Invertebrates inhabiting high velocity, torrential habitats should possess morphological and/or
behavioural adaptations that identify them as having adopted one of two strategies to maintain contact
12

with the substrate. Organisms should either be notably streamlined (a shape that delays boundary
layer separation, reducing pressure drag but experiencing higher lift as a trade-off), or should more
closely approximate a bluff body (reducing lift by increasing pressure drag due to early flow
separation). Rhithrogena and Epeorus (Heptageniidae: Ephemeroptera) appear to belong to the first
class; lift produced by their streamlined bodies is countered by a sucker-shaped arrangement of
abdominal gills which acts as a 'sucker' to help keep them in contact with the substrate (see Vogel
1994; Collier 1994). Pupal blepharicerid (Diptera) belong to the second class; the low pressure zone
associated with vortices produced by separating flow at the rear of the body enhance respiration.

The influence of bed geometry on stream flows
The presence of large-scale flow structures, such as horseshoe (solenoidal) vortices , wake
separation zones, and eddies are often accounted for in contemporary stream research (Nowell and
Jumars 1984; Bouckaert and Davis 1998; Buffin-Belanger and Roy 1998).

For example, Young

(1992; 1993) proposed a system to classify near-bed flow regimes based on the height and spacing of
bed roughness elements, which serves primarily as a descriptor of bed/flow interactions rather than a
method of classifying flows immediately adjacent to the bed . For example, when the mean roughness
height exceeds mean depth, the flow is categorised as chaotic. This characterisation, which includes
the majority of torrential flows , suggests that the near-bed flows of shallow channels with substrates
composed primarily of boulders are unpredictable. An earlier classification by Davis and Barmuta
(1989) was based on additional hydraulic parameters, such as Reynolds number, Froude number,
shear velocity, and the thickness of the viscous sublayer. While these parameters are essential for
the description of the near-bed flow environment, they cannot be measured easily in all instances.
Davis and Barmuta (1989), therefore, proposed that the distribution of near-bed flow regimes can be
estimated from vertical velocity profiles or even single measurements of mean velocities . However,
while several studies have measured near-bed velocity gradients in order to establish the link between
mean and near-bed flows in natural stream channels (e.g. Hart et al. 1996; Buffin-Belanger and Roy
1998), the ecological validity of the proposed classification of Davis and Barmuta (1989) remains
uncertain.
Bed or wall shear stress is one of the most important factors limiting the distribution of benthic
invertebrates (Davis and Barmuta 1989; Dittrich and Schmedtje 1995; Robertson et al. 1997).
13

Statzner and Muller (1989) developed a set of standardised hemispheres of different densities (FST
hemispheres) which , when deployed on the stream channel bottom, indicate the total shear stress at
the site. FST hemisphere results are, however, influenced by local bed topography (Dittrich and
Schmedtje 1995) and provide a spatially integrated measurement of shear stress (Statzner and Muller
1989). The latter attribute has lead to their widespread use in the determination of the hydraulic
conditions in streams at the reach and patch scale (Peckarsky et al. 1990; Lancaster and Hildrew
1993; Waringer 1993; Robertson et al. 1997).

However, this technique provides little insight into

small-scale distributions of bed shear stress on the surface of the substrate or the responses of
individual insects to local shear stress production.
Bed shear stress has also been quantified using estimates of shear velocity obtained from the
slope of the semi-logarithmic plot of the velocity profile (e.g. Wiberg and Smith 1991 ; Li 1994; Rempel
et al. 1999). In cases where the channel bed is topographically simple and boundary layer profiles
can be accurately predicted from existing theory (see Schlichting 1979), this method provides wall
shear estimates at the spatial resolution required to describe small-scale flow environments near the
bed . However, although widely applied, the validity of this method of estimating wall shear stress is
limited for several reasons .

The model applies to topographically simple channels where the

boundary layer has a characteristically logarithmic vertical distribution of velocity (Bergeron and
Abrahams 1992). This velocity distribution is often assumed to be universal in open channels (e.g.
Davis and Barmuta 1989), but bed irregularities (bedforms), including sand ripples and dunes (Li
1994), pebble clusters (Buffin-Belanger and Roy 1998), cobbles (Bergeron 1994; Bathurst 1994), and
streambed boulders (Hart et al. 1996), create pressure gradients which distort the boundary layer
velocity distribution . In addition , boundary layer development can be inhibited in instances where the
flow depth is limited , such as shallow tidal flows and shallow rivers (Nowell and Church 1979). As a
result, estimates of shear stress obtained from either mean velocity measurements or relatively
coarse velocity profile measurements in natural channels are potentially in error. Substantial errors in
this method were confirmed by Hart et al. (1996), who demonstrated that boundary layer profiles in
natural channels are likely to be too heterogeneous and complex to be easily explained by existing
boundary layer theory. The relationship between bed geometry, mean flows, and wall parameters can
only be examined by detailed mapping of velocities above the substrate (Davis and Barmuta 1989;
Wetmore et al. 1990).
14

The distribution of velocity above a topographically complex streambed is a product of the
relationship between bed geometry and various flow variables.

Bathurst (1994) showed that in

streams where the ratio of depth (D) to bed material diameter {dk) was greater than 10, velocity
profiles adopted a semi-logarithmic distribution.

Nowell and Church (1979), experimentally

manipulating roughness spacing in a flume using a slightly lower relative submergences (D/dk = 8),
found that the velocity profile above the bed roughness elements (Lego® blocks, in their case)
noticeably deviated from the normally accepted semi-logarithmic shape. Nowell and Church (1979)
further speculated that in natural channels with relative submergences lower than D/dk= 8, velocity
profiles would not follow a simple logarithmic distribution, and methods that estimate mean velocity
based on a few measurements and the assumption of a log-normal profile would have significant
errors. Young ( 1992) stated that in cases where D/dk :::; 3, flow is chaotic, the structure of the flow is
very complex, and the distribution and magnitude of near-bed velocities will be dependent on the
shape of the local bed geometry.
The flow in high gradient, mountain channels with coarse bed material is known to be
substantially different from that in topographically simple channels. Jarrett (1990) demonstrated that
the velocity profiles of mountain rivers (D/dk > 3) are non-logarithmic. Velocity profiles recorded by
Jarrett are distinctively "S-shaped"; near-bed velocities are lower and near-surface velocities are
higher than a logarithmically distributed velocity profile.

He suggested that near-bed velocities are

reduced as a result of form drag induced by the cobble and boulder bed material. Bathurst (1994)
also investigated the velocity distribution of mountain rivers , and found similarly S-shaped velocity
profiles in mountain river sites. However, Bergeron ( 1994) examined the flow structure in gravel-bed
streams whose relative submergences ranged from 1.63 > D/dk > 8.58, and found that the majority of
velocity profiles were neither semi-logarithmic in shape nor s-shaped .

Velocity profiles over the

topographically simple portions of the bed conformed closely to a "typical" semi-logarithmic shape, but
then, as flow passed over obstacles on the bed {bedforms), the rising bed elevation created local flow
acceleration and an associated pressure gradient. Bergeron (1994) concluded that the resulting nonlogarithmic velocity profiles consisted of two or more semi-log linear segments joined at inflection
points ("knots").

Flow profiles some distance downstream of the bedforms continued to be non-

logarithmic due to boundary layer separation and associated regions of recirculating flow.

15

Although the degree of relative submergence (D/dk) may seem like a simple and easily
employed rule of thumb by which the degree of near-bed velocity profile distortion may be estimated, it
is important to note that almost all of the above studies were conducted at relatively coarse spatial
scales . Li (1994), investigating near-bed flows over sand ripples at high spatial resolutions, found
substantial near-bed velocity profile distortion even though relative submergences were relatively high
(D/dk > 10). Similarly, Way et al. (1995), studying the relationship between macroinvertebrate habitat
requirements and bedform I flow interactions over preformed, grooved concrete substrates with high
relative submergences (D/dk »1 0), also found velocity profiles that deviated from the expected lognormal distribution. This suggests that velocity profile distortion may develop near the bed in any case
where water flows over a geometrically irregular surface, creating heterogeneous pressure gradient
distributions.
Although there is a solid empirical foundation for the examination of near-bed flows in natural
stream channels, their general applicability, especially in torrential streams, is limited for three
reasons . First, the vast majority of studies are conducted in streams and rivers where D/dk > 1.
Torrential streams often have relative submergences less than 1 due to relatively shallow flows and
extremely coarse bed material.

As relative submergences decrease, flows should become

increasingly complex and three-dimensional.

Second, due to the types of instrumentation used,

previous measurements of velocity profiles in natural streams have often failed to determine the shape
of the velocity profile immediately adjacent to the bed {but see Li 1994; Hart et al. 1996). Third , the
bed compositions of stream channels investigated in most studies are generally composed of finer
materials than that regularly found in torrential stream channels. Flows over large bedforms are likely
to be substantially different than flows over finer, more evenly distributed bed materials .
A number of techniques exist to quantify flows with in a few millimetres of the bed . Of these
instruments, only a few, such as the constant-temperature anemometer (CTA) and the laser-Doppler
anemometer (LOA), are capable of measuring velocities at the spatial resolutions required to quantify
near-bed flow variables. Unfortunately, these types of instruments are notoriously difficult to deploy in
the field (Hart et al. 1996). Other more routinely deployed devices, such as propeller flow meters
(Wetmore et al. 1990), bucket-wheel current meters (Rempel et al. 1999), electromagnetic current
meters (EMCMs) (Bergeron 1994; Buffin-Belanger and Roy 1998), and acoustic doppler velocimeters
(ADVs) (Bouckaert and Davis 1998) are robust but are incapable of measuring flows within millimetres
16

of the bed. The measurement of pressure difference by a Preston tube resting on a solid surface has
been used to directly measure shear stress (Preston 1954 ), providing an alternative to estimating
shear stress using the semi-logarithmic regression method. The addition of a static tube to a Preston
tube deployed in this fashion results in a Preston-static tube (PST), a device capable of measuring
wall shear stresses in flows characterised by curved streamlines (Ackerman et al. 1994). In addition
to measuring shear stresses, a Preston-static tube can also be used to measure water velocities
above the substrate {i.e. deployed as a Pitot-static tube) at spatial scales approximating the diameter
of the dynamic tube (Ackerman and Hoover 2001 ).

Meso-scale patterns: the link between macro- and micro-hydraulics
Natural systems are patchy in space and time. Stream ecosystems are heterogeneous, both
in terms of physical (hydrodynamic parameters, substrate size, light, oxygen) and biological (algal and
faunal densities) characteristics (e.g. Palmer 1995). Ecologists have long sought to understand how
small-scale heterogeneity in the stream environment affects the population dynamics, community
ecology, and individual behaviour of benthic taxa (see Nowell and Jumars 1984; Davis and Barmuta
1989; Carling 1992 for reviews). As hydrodynamic parameters are the most important environmental
factors influencing patterns of invertebrate diversity and abundance (Statzner and Higler 1986), the
ecological,

behavioural, and

morphological

adaptations

of benthic organisms

to hydraulic

environments are predictably diverse.
Perhaps the single greatest topic of discussion among workers investigating the biology of
invertebrates in lotic systems is the relevance of routinely measured parameters.

For example,

variables that describe channel morphology may be correlated with the abundance or distribution of
certain species.

Buffagni et al. (1995) found that the distribution of the mayfly Rhithrogena

semicolorata is related to bed roughness.

However, given the complex relationships that exist

between bed geometry and hydrodynamic factors, R. semicolorata may, in fact, be responding to a
suite of hydraulic variables (i.e. mean and near-bed velocity, depth, turbulence, or substrate
roughness) that are functionally related to channel morphology (see Newbury 1984; Robert et al.
1996; White 1999). Stream macroinvertebrates are taxonomically and morphologically diverse (Resh
and Rosenberg 1984), but generally range in size from one to several millimetres. Thus, relating
patterns of invertebrate abundance to hydraulic or hydrologic factors quantified at relatively large

17

scales (one to several orders of magnitude larger than the body size of benthic taxa) may provide little
useful information on the proximal factors that influence the behaviour of individual organisms. These
methodologies may, however, help understand the population biology of benthic taxa. While many
benthic invertebrates are recognised as having well-defined preferences and adaptations for particular
velocity ranges and substrate sizes, the mechanisms that link large-scale physical factors to smallscale patterns of invertebrate behaviour are not well understood.

Benthic-lotic coupling in the stream environment
The interaction between flow and streambed geometry creates localised variations in nutrient
supply, food exchange, and erosive shear (Davis 1986). In turn, these factors have considerable
indirect influences on the biotic processes of the stream environment, such as predator-prey and
grazer-producer interactions (e.g. Osborne and Herricks 1987; Peckarsky and Wilcox 1989). The
patterns of invertebrate abundance and diversity that are a product of these abiotic and biotic
interactions have been recognised by a number of workers (Jowett et al. 1991; Holomuzki and
Messier 1993; Quinn and Hickey 1994; Collier et al. 1995). The association of patterns of abundance
with hydrodynamic parameters has led to the development of habitat suitability models for several
species (e.g. Deleatidium spp.; Jowett et al. 1991 ). Increasing numbers of studies of this type have
led to the development of methodology and techniques appropriate for the description of the
characteristics and distribution of 'patches' of hydraulically similar microhabitats across the
streambed.

Several comprehensive reviews (Nowell and Jumars 1984; Davis 1986; Davis and

Barmuta 1989; Carling 1992) have stressed the importance of appropriately describing the small-scale
hydrodynamic variation inherent in the stream environment, and suggest a suite of variables to do so.
Indeed, significant relationships have been found between the abundances of invertebrate taxa and
both simple (e.g. discharge, depth, and velocity) and complex (e.g. Reynolds number, Froude
number) hydrodynamic parameters.
Several studies have examined the microdistributions of sedentary and semi-sessile species
(i.e. suspension-feeding caddisfly and simuliid (black fly) larvae; see Table 1-1 ). Suspension-feeding

caddisflies (Trichoptera) are a trophically important group as they trap and process suspended
particulate organic matter (POM) in a stream segment, increasing the efficiency of the local recycling
of available resources (Wallace and Merritt 1980; Voelz and Ward 1996).
18

Similarly, suspension

feeding black fly larvae (Diptera: Simuliidae) constitute an important pathway for energy flow in stream
ecosystems (Merritt et al. 1996), transforming collected ultrafine POM (0.45 - 50 11m) into fine POM
(50 11 m - 1mm), a form more readily utilised by stream organisms (Merritt 1987). Voelz and Ward
(1996) found that the distributions of the suspension-feeding caddisflies Arctopsyche grandis and
Brachycentrus occidentalis were largely restricted to the lower sides and bottoms of stones, where

they were exposed to velocities less than 20 em s·1 . Individuals of both species aggregate underneath
the leading edge of cobbles and boulders. While the flow patterns beneath stones would be complex
(Statzner et al. 1988), flow entering the underside of stones from the front might be relatively
unidirectional compared to the water exiting the side and rear of the stones . Voelz and Ward (1996)
suggest that the distributions of these net-spinning caddisflies reflect the best local filtering velocities,
and may also reflect areas with desirable levels of turbulence. Several studies (e.g. Wetmore et al.
1990) have reported that suspension-feeding caddisflies prefer the exposed tops and sides of stones
where the flow is typically rapid, has higher Froude numbers (average Fr = 0.6), and higher stream
surface slopes.

These microhabitats represent areas of converging streamlines, and the locally

highest rates of seston delivery.

However, Voelz and Ward (1996) suggest that the observed

microhabitat preferences may, in fact, be seasonally variable. B. occidentalis and A. grandis exhibit
seasonal shifts in positioning behaviour, moving into more exposed locations only during the late
summer months when many of the previous studies had been conducted. Given that the underside of
stones and interstitial spaces represent an important niche in streams (Voelz and Ward 1996;
Robertson et al. 1997; also see Resh and Rosenberg 1989), future research on the nature of flows
beneath substrate elements is needed. For example, the presence of a solenoidal (horseshoe) vortex
around the front of the stone (Nowell and Jumars 1984; Craig 1996; Bouckaert and Davis 1998) would
presumably alter the velocity and turbulence of the water entering the underside of stones, further
influencing the microhabitat preferences of these suspension-feeding caddisflies .
The net spinning hydropsychid caddisfly Hydropsyche also relies on local current to deliver an
adequate supply of seston; as such, current velocity acts as an important selective pressure in
determining net mesh aperture and food particle size (Fuller and Mackay 1980).

Using

hydrodynamically calibrated artificial substrates deployed in a natural stream, Osborne and Herricks
(1987) found distributions of Hydropsyche to be related to flow patterns and turbulence intensity rather
than average velocity. High degrees of hydraulic niche overlap were noted among the four species of
19

Hydropsyche studied (H. betteni, H. spama, H. cheilonis, and H. bronta); larvae of all four species

were found in regions of rapidly spiralling vortices . These vortices may serve to increase the rates of
seston capture by increasing capture rates without substantially increasing drag on the body of the
insect (see Lacoursiere 1992).

The findings of Osborne and Herricks (1987) agree with earlier

hypotheses (e.g. Fuller and Mackay 1980) which suggest that variations in turbulence could be an
important factor determining interspecific differences in microhabitat preferences among hydropsychid
caddisflies. Similarly, Quinn and Hickey (1994) found that the suspension-feeding caddisfly taxa of
New Zealand rivers were primarily associated with patches of high velocity and high turbulence
intensity, and suggested that these two variables increase the particle capture rates of net-spinning
suspension-feeders . Turbulence is thought to be important to suspension-feeding caddisflies with
respect to their energy budget; while an organism will probably expend more energy maintaining a
position in areas of higher velocity, the construction of nets in an area of rapid spiralling flows might
act to increase rates of seston delivery (Osborne and Herricks 1987). This conclusion is supported in
part by the findings of Peckarsky et al. (1990), who found that densities of Hydropsyche instabilis
increase with FST hemisphere number, an indicator of surface velocities and shear stresses (see
Statzner and MUller 1989). Peckarsky et al. (1990) also note, however, that H. instabilis spins a net of
relatively coarse mesh, which may be regarded as an adaptation for high velocities . Similarly, the
predatory caddisfly Plectrocnemia conspersa, which spins a silk net to ensnare invertebrate prey,
modifies the design and aperture size of its net as flow rates increase (Townsend and Hildrew 1979).
The heterogeneous distribution of near-bed velocity in streams plays an ecologically important
role in determining the distribution of stream microcrustacea, a group comprised of small-bodied
cladocerans, copepods, and ostracods . Stream microcrustacea maintain an important trophic link in
lotic ecosystems by feeding on detritus and detritus-associated organisms. While the patterns of
microdistribution and abundance of other benthic invertebrate taxa have been associated with specific
hydraulic habitats defined by a variety of hydrodynamic factors, Robertson et al. (1997) found that the
distributions of microcrustacea are influenced by the fluctuations in flow associated with flood events
('spates').

Epibenthic microcrustacea , taxa which inhabit the surface of the substratum, are

particularly susceptible to erosion; ostracods can be displaced at velocities as low as 2 em s·1 , and the
cyclopoid copepod Eucyclops becomes numerous in the drift at 7.5 em s·1 (Richardson 1992). The
average near-bed velocities of streams in which benthic microcrustacea abound frequently exceed
20

these values.

During and after flood events, microcrustacea persist in flow refugia, areas of the

streambed that retain low hydraulic stresses, even at high flows.

Following the periods of high

discharge that accompany rainfall or seasonal runoff, these flow refugia provide a source of
microcrustacea from which denuded or disturbed areas of the bed can be repopulated.

Epibenthic

microcrustacea tend to be relatively broad-bodied, and dominate the species assemblage in streams
that have large dispersive fractions (essentially, areas of the stream bed which maintain low shear
stresses, even during disturbance events; Robertson et al. 1997). Similarly, Shiozawa (1991) found
that epibenthic microcrustacea were associated with low-velocity patches of the streambed containing
deposits of fine organic material and silts . Interstitial species, which are often slender and vermiform,
comprise the majority of microcrustacean species in high velocity streams that have low dispersive
fractions.

Unlike epibenthic species, interstitial taxa, such as harpacticoid copepods, benefit from

periods of high erosive forces.

High bed shear stresses minimise the deposition of fine particles

which, over time, tend to clog the pore space in bed gravels, reducing both flow-through rates and
levels of oxygen experienced by subsurface taxa (Robertson et al. 1997).
A test of the intermediate disturbance hypothesis by Townsend et al. (1997) provides an
interesting perspective on the ecological role of flow refugia in determining the community composition
of stream benthos.

The intermediate disturbance hypothesis, whose early uses include an

examination of patterns of diversity in coral reefs and tropical rain forests by Connell (1978), proposes
that in patchy and dynamic environments, intermediate levels of physical disturbance should produce
the highest species diversity. Intermediate levels of disturbance establish a ecological compromise
between highly unstable systems where frequent disturbances produce a community composed
primarily of a limited number of highly tolerant taxa able to rapidly colonise impacted patches, and
stable systems where rare disturbances result in communities dominated by a few, competitively
superior, species.

Daily disturbances in flow (a result of hydropower regulation of the channel

discharge) resulted in a loss of heptageniid mayflies in Swedish rapids (Malmqvist and Englund 1996),
a trend attributed to increases in the sedimentation of transported fine particles associated with lower
current velocities (generally, particles < 1 mm in diameter). In an undisturbed catchment, Townsend
et al. ( 1997) found that macro invertebrate diversity was greatest at sites that experienced intermediate
levels of flood-related disturbance. Variations in taxonomic richness were related to indices of bed
disturbance, and Townsend et al. ( 1997) suggest that the primary location of flow refugia were low-

21

velocity 'dead spaces' between large substrate elements . Invertebrate diversity decreased as the
proportion of the bed material composed of fine sediments increased.

Periods of high flow are

thought to reduce the amount of fine sediment stored within streambeds, increasing oxygen
availability (Robertson et al. 1997) and the percentage of available interstitial space.

However,

Townsend et al. (1997) acknowledge that the small-scale hydrodynamic factors that invertebrates
detect and in turn respond to remain largely unexplored.
The microdistribution of benthic grazers is also a function of the distribution of algal food
resources. However, the growth of epiphytic algae (especially diatoms, the preferred food of many
grazing invertebrate taxa) is, in turn, also a function of small-scale patterns of flow and variations in
velocity (Quinn et al. 1996; see also Resh and Rosenberg 1984). As a result, the environmental
factors that determine the microhabitats of grazers and algae can be difficult to separate. Jowett et al.
(1991) found that while the abundance of the mayfly Deleatidium was correlated with a combination of
Froude number and substrate size, it was most highly correlated with periphyton biomass. Similarly,
Palmer (1995) found that the mayflies Epeorus and Baetis grew to larger sizes in high velocity
environments with homogeneously distributed food resources, while Poff et al. (1990) demonstrated
that current has a significant influence on algal abundance and invertebrate species composition in
high velocity streams.

The thinner boundary layers and higher shear velocities associated with

microhabitats exposed to high velocities would reduce the diffusional distance to the boundary,
increasing rates of nutrient and oxygen delivery to organisms on the stream bed (Carling 1992; Davis
1986). This would benefit the growth of both periphyton and invertebrate grazers, many of which have
relatively inefficient gills (Wiley and Kohler 1980; Golubkhov et al. 1992). An accurate separation of
the flow requirements of primary producers and consumers awaits an experimental manipulation of
algal communities across a range of microhabitat velocities.

Invertebrate drift: migration using bulk flow
Flowing water is used widely by stream macroinvertebrates for movement. While upstream
movements of benthic invertebrates by crawling (Winterbottom et al. 1997) and swimming against the
current (Abelson 1997; Rader and McArthur 1995) have been recorded, the vast majority of migratory
movements in lotic environments are downstream, in the direction of least resistance.

Either

accidentally or intentionally, stream benthic invertebrates often enter the moving water column above
22

the substrate, and are carried downstream. Stream invertebrates actively enter the drift in order to
escape fish and invertebrate predators, avoid competitors, or migrate to areas of higher food
availability (Hildrew and Townsend 1982; Wooster and Sih 1995; Forrester 1994). Other factors
influencing invertebrate drift include sunlight, discharge, turbidity, oxygen, and substrate (see Wiley
and Kohler (1984) for a review). Although diurnal patterns vary between species, the numbers of
stream invertebrates drifting during the day is generally low and constant, followed by dramatic
increases at night (Sagar and Glova 1992). Rates of invertebrate drift are highest at dawn and dusk
in many species. Several authors (e.g. Forrester 1994) have hypothesised that diel periodicity in drift
addresses a significant source of selective pressure. Drifting invertebrates are an important source of
food for many stream fishes, which feed almost entirely during the day as they rely on visual acuity to
locate and attack prey (Walsh et al. 1988). As such, there should be strong selection pressure
against behaviours (including drift) that expose individuals to visually oriented predation by fish.
Invertebrate drift is generally thought to be an active process, where invertebrates detach
from the substrate in order to migrate downstream (Wooster and Sih 1995). However, the precise
mechanisms by which invertebrates detach from the streambed to initiate downstream movement
have not yet been investigated.

Whether stream insects simply detach from the surface of the

streambed, swim briefly upward, or initiate a change in body position that creates sufficient lift to impel
them upward into the water column is as yet unknown. However, several authors have noted that
there are disproportionately high densities of large size classes of many species in the drift (e.g.
Lancaster et al. 1996; Sagar and Glova 1992). Similarly, it has been noted that large invertebrates
are more susceptible to erosion than smaller members of the same species; this variation in risk of
detachment from the substrate is presumably due to the greater pressure drag and lift experienced by
larger individuals (Davis 1986; also see Weissenberger et al. 1991 , and equations 1.11 and 1.12).
Winterbottom et al. (1997) found that larger stoneflies of the predatory species Leuctra nigra were
more likely to become detached from the substrate during changes in discharge than small
individuals. Both Winterbottom et al. (1997) and Lancaster and Hildrew (1993) note that small L. nigra
were able to move more easily in high velocities. Peckarsky and Cowan (1995) found that the activity
levels of several predator (Megarcys signata and Kogotus modestus) and prey (Cinygmu/a sp.,
Epeorus deceptivus , Baetis bicaudatus, and Ephemerella infrequens) species increased at night, as

did their use of the exposed upper surfaces of stones . Given that stone surfaces exposed to the flow
23

generally have higher local velocities , higher shear stresses, and thinner boundary layers than
interstitial surfaces (Nowell and Jumars 1984; Davis 1986), it is possible that the diel periodicity of drift
is due, at least in part, to the accidental dislodgement of diurnally active individuals moving over highrisk hydraulic patches rather than intentional detachment for the purpose of migration.

This is

supported by the conclusions of Poff and Ward (1991 ), who interpreted the increases in drift
associated with periods of increased discharge to an increases in the accidental detachment ("scour'')
of invertebrates from the streambed . Periodic drift due to accidental detachment should be more
pronounced in species that lack specific adaptations to high flow environments, yet are present in a
wide range of hydraulic microhabitats (i.e. hydraulic generalists) .
Once entrained in the drift, the responses of stream invertebrates vary. The length of time
that an invertebrate spends in the drift before returning to the stream bottom (the return rate) is a
result of the interaction between the hydraulic transport properties of the stream channel and the
behavioural responses of the individual (Lancaster et al. 1996). Hydraulic 'dead zones' (areas of the
stream bed where water is held in transient storage - stream margins, turbulent eddies, vortices
attached to boulders or logs, and backflows associated with pools or bends) provide opportunities for
invertebrates to regain the substrate. Lancaster et al. (1996) found that channels with the highest
mean velocities had the lowest return rates, but also report that streams whose beds had high
proportions of dead zones had relatively high rates of return , regardless of velocity.

They also

suggest that turbulence and channel-specific depth may increase the length of time invertebrates are
entrained .
By altering their behaviour, some invertebrates can alter the length of time spent in the drift.
At very low velocities, some invertebrate taxa can actively increase time spent drifting (Poff and Ward
1991 ). Winterbottom et al. (1997) found that the length of time spent in the drift by the leuctrid stonefly
Leuctra nigra was entirely dependent on flow and the hydraulic characteristics of the channel. The

inability of L. nigra to behaviourally modify its return rate is consistent with the findings of Lancaster et
al. (1990) which show that L. nigra is a poor colonizer and disperses slowly. However, the length of
time that the nemourid stonefly Nemoura pictetii spends in the drift is independent of flow factors .
Behavioural control of return rate allows nemourids to colonise new substrates rapidly (Winterbottom
et al. 1997; Lancaster et al. 1996), an attribute which allows individuals to minimise the length of time
they are exposed to predation by fish . The capacity to efficiently utilise downstream transport in

24

moving waters is also important to the feeding ecology of stream herbivores. The mayflies Baetis
bicaudatis and Epeorus deceptivus grew to larger body sizes when high velocities allowed them to

drift successfully between high resource patches (Palmer 1995). Interestingly, the growth of these
two grazing species was maximised when they entered the drift less frequently.

Increases in time

invested to 'search' (in the drift) for higher resource patches may be beneficial; however, time spent
searching results in lost feed ing opportunities in a patch of known resource availability.

Ontogenetic shifts in hydraulic habitat and microdistribution
The influence of moving water on a stationary body is dependent on spatial scale, as
indicated by the Reynolds number. As benthic invertebrates generally increase in size and mass
during their growth and development, one might expect to find size-related shifts in behaviour and
microhabitat preference (see Table 1-1).
The spatial niches of organisms change as they increase in size.

Osborne and Herricks

(1987) found ontogeny-related differences in microhabitat velocities among four ecologically similar
species of the caddisfly Hydropsyche. H. cheilonis, H. sparna, and H. bronta occupied areas of the
substratum where maximum velocities reached 25 em s·1 , while the larvae of H. betteni were found in
velocities as high as 35 em s·1 . Although the degree of hydraulic niche overlap between the four
species is high, H. betteni is roughly 10 - 15% larger at each instar than the other three hydropsychid
species, which are virtually identical in size. Osborne and Herricks (1987) noted a similar intraspecific trend; larger, higher instar larvae of the four Hydropsyche species were capable of inhabiting
regions of higher microhabitat velocities.

The results of Osborne and Herricks suggest that size-

related differences in hydrodynamic tolerances are a function of body size, rather than functional
changes in capture net mesh dimensions.

Collier et al. (1995) found that larger larvae of the

hydrobiosid caddisflies Hydrobiosis parumbripennnis and Costachorema callistum tended to be more
common in higher velocities than smaller larvae. Collier et al. (1995) concur that this difference is a
function of the larger physical sizes of higher instars, and suggest that the large anal prolegs used by
caddisflies to move over the substrate in fast currents may be more effective in maintaining position
as they increase in size.
Intraspecific variations in distribution of the leptophlebiid mayfly Deleatidium is size related as
well (Collier 1994). While the velocity preferences of this species are relatively broad, it is generally
25

found in areas of higher flow (Jowett et al. 1991). Collier (1994) found that the abundance of smaller
nymph size classes was greatest in relatively low-velocity patches (<40 em s·1 ), while larger nymphs
reached their highest densities in higher velocity habitats (>90 em s·1 ). Collier (1994) suggests that
size-related differences in distribution might be a reflection of changes in oxygen requirements as
individuals grow. Increases in size result in lower surface-to-volume ratios, which may, in turn, restrict
respiratory gas exchange rates.

Higher flow rates thin the boundary layers associated with

invertebrate gills, resulting in increased diffusion of respiratory gases (Nowell and Jumars 1984).
However, while oxygen and food supplies may be more readily available at high velocities, the
proportion of total drag attributable to pressure drag is greatest for larger, late instar larvae (see Vogel
1994).

For stream insects, especially grazers foraging in high velocity microhabitats, this is a

necessary tradeoff. Individuals can either avoid areas where high velocities and erosional forces (lift
and drag) threaten to detach them from the substrate, or possess morphological and/or behavioural
adaptations to counter the effects of lift and drag (Weissenberger et al. 1991; Collier 1994 ). For
example, Buffagni et al. (1995) found that the velocity preferences of the mayflies Rhithrogena
semico/orata and Ecdyonurus venosus, both obligate inhabitants of high flow environments, did not

change with increasing size. Rather, larger individuals of both species tended to prefer substrates of
greater roughness. Buffagni et al. (1995) suggest that bed roughness may provide a better descriptor
of complex hydraulic characteristics near the stream bottom, citing turbulence and 'force of flow' as
possible factors . These species may be taking advantage of flow microrefugia, small-scale surface
irregularities that produce localised areas of relatively thick boundary layers (Davis and Barmuta
1986).

Larger mayflies may find maintenance of position in these areas to be less expensive

energetically. A similar use of flow microrefugia was noted in the caddisfly Hydropsyche (Osborne
and Herricks 1987); larvae are often found associated with small depressions and imperfections in the
substrate surface. This may allow these suspension-feeding caddisflies to hold their nets in the higher
velocities of the upper boundary layer, while restricting the exposure of the bulk of their bodies to the
low shear stress, lift, and drag of a locally-thickened boundary layer.
Benthic predators, which are not directly dependent on the epiphytic food resources available
in high velocity patches, would not be expected to show the same specific adaptations to high velocity
microhabitats. Winterbottom et al. ( 1997) found that large stoneflies of the species Leuctra nigra were
more susceptible to changes in discharge than were smaller individuals, and suggested that while

26

larger individuals were more likely to be eroded, smaller L. nigra were able to move more easily over
exposed areas of the stream bed at high velocities. Similarly, Lancaster and Hildrew (1993) reported
that the distribution of small , early instar L. nigra did not change with increases in flow, and were
generally associated with high velocities and coarse substrates. Thus, the distributions of stream
predators (which rarely demonstrate specific adaptations to counter the erosional effects of high
velocities; see below) may not be the result of a preference for low-velocity habitats. Rather, high
near-bed velocities may act to prevent stream invertebrate predators may from foraging in high
velocity habitats, where prey is often abundant.

Hydrodynamics, prey, and predation: eating in the fast lane
Given the ecological importance of morphological and behavioural adaptations to both stream
hydrodynamics (e.g. Quinn et al. 1996) and predator-prey interactions (e.g. Tikkanen et al. 1997), it is
of little surprise that research on the relationship between hydrodynamics and predator-prey
interactions is growing.

Basic ecological tenets state that harsh environments and physical

disturbances weaken interactions between species (see Menge 1976).

In freshwater lotic

environments, where hydrodynamics play a pivotal role across physical scales ranging from river-wide
zonations (Statzner and Higler 1986) to fine-scale respiratory currents over invertebrate gills (Wiley
and Kohler 1980), one would expect to find a wide range of ecological and behavioural responses
linking predator and prey (see Table 1-1).
Peckarsky et al. (1990) provided the first formal test of the harsh-benign hypothesis in the
stream environment, gauging the influence of hydraulic-habitat preferences on predator-prey
interactions.

The harsh-benign hypothesis, originally developed to explain the relative roles of

predation and competition in structuring marine rocky intertidal communities (Menge 1976), can be
used to assess the impact of predatory invertebrates on prey populations along a gradient of hydraulic
regimes . Peckarsky et al. {1990) found that densities of the predatory stonefly Dinocras cephalotes
peaked at FST hemisphere 11 (see Statzner and Muller (1989) for a description of the FST
apparatus), indicating a preference for medium-range flows and shear stresses. However, the prey
species of D. cephalotes (including Hydropsyche (Trichoptera), Baetis (Ephemeroptera), assorted
Chironomid species, and the amphipod Gammarus (Crustacea)) showed preferences for higher flow
conditions (average of FST hemisphere 18). When prey were distributed across a range of habitats,

27

---------

predation impacts on the prey were significant only in those patches with hydraulic reg imes favourable
to the predators . The flatworm predator Dugesia dorotocepha/a is similarly limited in its predation on
larval blackflies (Simulium vittatum) . Hart and Merz (1998) found that while D. dorotocephala will
readily attack black fly larvae when it encounters them, the flatworm is poorly adapted to high
velocities. As a result, high velocity sites not only acted as flow-mediated refuges from predation , but
also coincided with areas where the feeding rates of the food-limited, suspension-feeding black fly
larvae are maximised. Peckarsky et al. (1990) suggest that reductions in prey populations are not
entirely due to predation; predator avoidance behaviours, including drift (Wooster and Sih 1995) and
migration (Peckarsky and Cowan 1995) account for the majority of the observed community change.
However, the criteria by which predators and prey perceive hydrodynamic conditions as being either
'harsh' or 'benign' remain unknown for most species.
Microhabitat overlap between prey and predator species may determine encounter rates, but
this does not necessarily translate into prey preferences. Prey preferences are not only a function of
availability, but also species-specific handling times and attack success rates (Sih and Wooster 1994).
The perlodid stonefly Diura bicaudata demonstrates a strong preference for black fly larvae prey, even
though microhabitat overlap with black fly species is minimal. D. bicaudata has the highest degree of
microhabitat overlap with several species of mayfly, including Baetis, Ephemerella, and Heptagenia,
yet rarely utilised them as prey (Tikkanen et al. 1997). Tikkanen et al. ( 1997) also found that baetid
mayflies were able to risk interactions with predators due to highly effective escape manoeuvres.
Interestingly, however, Palmer (1995) reported that rates of predation by the stonefly predator
Megarcys signata on Baetis bicaudatis were higher at 10 em s-1 than 30 cm-1 , as the swimming

escape response of Baetis is less effective at lower velocities . It is possible that the higher shear
stress, greater turbulence intensity, or thinner boundary layers at higher velocities interferes with the
ability of M. signata to recognise the 'hydrodynamic signature' of the swimming baetids (see
Peckarsky et al. 1990). Alternatively, once the escape response is initiated, higher velocities may
reduce the rate of predation success by transporting prey out of the attack range of predators more
quickly.
Invertebrate predators also use hydrodynamic cues to detect prey and discriminate between
preferred and non-preferred prey species .

Peckarsky and Wilcox (1989) found that the perlodid

stonefly Kogotus modestus never attacked motionless mayfly prey, but was able to use the pressure

28

wave patterns associated with swimming escape behaviour of the mayfly Baetis bicaudatus to
recognise and attack this preferred prey species.

Based upon the pressure wave patterns, K.

modestus was also able to distinguish B. bicaudatus from Ephemerel/a infrequens , a non-preferred

mayfly prey species.
Black fly (simuliid) larvae increase the chance of surviving an attack by an invertebrate
predator by detaching into the drift or vigorously biting the head region of the predator (Tikkanen et al.
1997). These tactics are, unfortunately, relatively ineffective; black fly larvae are the preferred prey of
predators such as the stonefly Megarcys signata, which attack blackflies with a relatively high capture
probability when they are encountered . Yet rates of predation of black fly larvae are relatively low,
due to a separation of the hydraulic niches of predator and prey. Suspension-feeding simuliid larvae
aggregate in exposed microhabitats, preferably attaching in areas of low shear stress but relatively
high velocity (Lacoursiere 1992), while M. signata prefers low velocity microhabitats along the
periphery of streams . The microsite preferences of Simulium also moderate the predatory impacts of
freshwater triclad flatworms on aggregations of larval black flies (Hansen et al. 1991; Hart and Merz
1998).

Due to the incomplete microhabitat overlap between triclads and their simuliid prey, the

flatworms are unable to forage in areas of the bed that contain the highest densities of simuliids .
Muotka and Penttinen ( 1994) found that the stonefly predator /soper/a grammatica is similarly
excluded from the microhabitat preferred by the larval black fly Simulium sublacustre. They suggest
that /. grammatica either simply avoids the areas of high current velocity preferred by the black fly
larvae, or is competitively excluded from foraging in exposed areas by a species that moves more
efficiently over the surface of stones at high velocities. High velocity patches do not provide a refuge
from all stream predators, however. The microhabitat preferences of S. sublacustre and the predatory
caddisfly Rhyacophila obliterata overlap highly when local water velocities reach 40- 60 em s·1 ; this
results in high rates of consumption of the sedentary black flies by this large, active rhyacophilid
predator (Muotka and Penttinen 1994 ).
The importance of invertebrate predators to stream communities is also reduced during the
hydraulic restructuring that occurs during floods and spates. Increases and redistributions of velocity
and shear stress accompany increases in stream discharge. These often rapid changes can cause
losses and changes in community structure (Lancaster 1996).

Animals may be lost from

microhabitats subject to high erosional forces , while individuals in low-flow refugia may remain
29

essentially unaffected (Lancaster and Hildrew 1993; Winterbottom et al. 1997). The individuals in
these refugia, coupled with those that migrate into them during the disturbance event, will be available
to recolonise denuded areas of the streambed when the distributions and levels of velocity and shear
stress return to average. Lancaster ( 1996) found that the alderfly Sialis fuliginosa did not significantly
shift its microdistribution with changes in discharge. As a result, the total impacts of this predator
decreased during periods of hydraulic disturbance. Consumption of prey remained at pre-disturbance
levels for those individuals found in refugia, but was reduced throughout the remainder of the
streambed where prey populations had been reduced. Lancaster suggests that because S. fuliginosa
is an active forager, it may simply be unable to feed efficiently in fast currents.

However, flow

disturbance events have very different effects on the feeding ecology of the caddisfly Plectrocnemia
conspersa, a primarily lie-and-wait predator which spins silken nets in which prey become entangled .

Lancaster (1996) found that, unlike S. fuliginosa, rates of prey consumption by P. conspersa
increased during spate events. The microdistribution of P. conspersa remains similar to that of the
invertebrate species it preys upon; it is uniformly distributed across hydraulic habitats at low flows, and
becomes abundant in flow refugia during periods of high discharge. Increased rates of predation by
P. conspersa may be a function of the increased movements of migrating prey in the process of

searching for and moving into areas of lower velocity and shear stresses . However, Winterbottom et
al. (1997) noted that the mobility of P. conspersa was related to increases in discharge; this suggests
that changes in distribution of this caddisfly may be, in part, a result of dislodgement of individuals
caught in exposed areas of high shear stress during flood-associated increases in discharge. This
suggests that while the harsh-benign hypothesis may provide insight into the impacts a predator may
have on prey populations across a wide range of hydraulic conditions, its application may be limited to
those species that show well-defined hydraulic preferences. The predatory impacts of those species
that are able to modify their predatory behaviour to accommodate the microhabitats occupied by a
range of prey species may be less dependent on predator/prey microhabitat overlap.

30

Fine-scale flows: behavioural and morphological hydrodynamics
Fine-scale manipulations of flow
The benthic invertebrate taxa of lotic systems have evolved a range of adaptations to cope
with hydrodynamic forces. They have, in many cases, coupled complex behaviours with specialised
morphologies to utilise the energy of their environment (see Table 1-1 ). For example, the bodies of
the larvae of the aquatic beetle genus Sclerocyphon (Coleoptera: Psephenidae) locally modify the
thickness of the boundary layer to produce a suite of ecological and energetic benefits. Using dye
injection, Smith and Dartnall (1980) found that these unique larvae live in turbulent environments,
where turbulent boundary layers develop in high velocity flows . The streamlined, flattened bodies of
Sc/erocyphon (figure 1-4) modify the developing boundary layer, increasing the thickness of the local

viscous sublayer. A thickened viscous sublayer reduces the risk of dislodgement as Sclerocyphon
grazes the epiphytic diatoms and other algae found on the upper surfaces of boulders and cobbles
(Quinn et al. 1996). However, life within the viscous sublayer has disadvantages; the diffusion of
respiratory gases within the viscous sublayer relies upon the relatively slow rates of molecular
diffusion.

In situations where the viscous sublayer is sufficiently thick to impair respiration,

Sclerocyphon creates its own respiratory current using the anal tracheal gills located beneath the last

abdominal tergite. The larvae extrude, then actively pump these gills, creating an area of localised
turbulence at the rear of the body. This results in increased rates of respiration and waste removal
while minimising increases in overall drag on the body. However, as Reynolds numbers increase, the
viscous sublayer thins (Carling 1992), exposing the bodies of larval Sclerocyphon to a turbulent
boundary layer.

In these circumstances, Smith and Dartnall (1980) suggest that continued vortex

production at the rear of the body is possibly no longer of use in ventilation, but rather may be acting
to minimise pressure drag. In these high velocities, a secondary current that forces small amounts of
water between the lateral laminae produces a phenomenon known as boundary layer suction.
Boundary layers over the insect are further thinned, while boundary layer separation is delayed. As a
result, pressure drag produced by the body of Sclerocyphon is reduced .
As mentioned above, the bodies of the stream invertebrates that inhabit the exposed surfaces
of the streambed have long been regarded as being adapted to the hydraulic forces of their
environment. Ambuhl (1959) felt that dorso-ventral flattening could be regarded as an adaptation to
maximise the proportion of the body in the boundary layer, avoiding, to the greatest extent possible,
31

mean water velocities. This idea was generally accepted until Statzner and Holm (1982) re-examined
the fine-scale flow patterns around the body of the mayfly Ecdyonurus venosus, the same species that
Ambuhl had studied 23 years earlier. They found that Ecdyonurus , and presumably similar mayfly
species, are not simply "living a sheltered life in the boundary layer''. Moreover, the water moving
over the body of E. venosus is far less smooth than Ambuhl had originally reported. Statzner and
Holm ( 1982) found that water approaching the front of the mayfly slowed down substantially near the
substrate at the front of the animal. Though Statzner and Holm (1982) do not discuss this trend , this
pattern possibly indicates the presence of a horseshoe (solenoidal) vortex immediately in front of the
head of the insect. The isovels ('layers' of equal water velocity) are then compressed as water flows
up over the length of the thorax, then re-expand over the abdomen, suggesting that the lift and friction
forces are concentrated over the thorax, where the legs of E. venosus are in contact with the bottom .
Statzner and Holm (1982) regard this as an adaptation to maintain contact with the substrate in high
velocities . Weissenberger et al. (1991) found that the morphology of Ecdyonurus counters much of
the lift created by the steep velocity gradients over the thorax. The large head shield of this species
can be angled to create negative angles of tilt, which reduces lift forces .

The broad legs of

Ecdyonurus, which are shaped like aerodynamic 'spoilers', can be angled to press the body down

against the substrate. These morphological and behavioural adaptations can produce negative lift
forces , minimising the energy Ecdyonurus must expend to maintain contact with the substrate.
Alternatively, Weissenberger et al. found that Epeorus , another heptageniid mayfly, does not possess
similar lift-reducing adaptations. The airfoil-shaped body of Epeorus produced much higher lift forces ,
which it apparently counters by anchoring itself to the substrate.

Epeorus did, however, have the

lowest drag coefficient of the species examined (Weissenberger et al. 1991 ).
Of special interest is the relationship between lift, drag, and the morphology of the perlid
stonefly Perla bipunctata (Weissenberger et al. 1991 ). The lift forces experienced by this stonefly,
which does not exhibit any specialised adaptations for life or movement in high flow environments,
remain at very low levels until water velocities reach approximately 50 em s·1 • Above this velocity, the
lift forces experienced by an individual stonefly increase rapidly. This supports the results of Muotka
and Penttinen ( 1994 ), which suggest that a thresh hold velocity exists, beyond which stonefly
predators (generally 'hydrodynamic generalists') are prevented from seeking prey due to

32

hydrodynamic constraints. As lift forces increase, hydraulic generalists such as stonefly predators
may experience difficulties maintaining contact with the substrate.
Many benthic insect taxa (e.g. Diptera, Trichoptera) must enter a pupal stage before they can
become reproductively mature adults (Resh and Rosenberg 1984). During the pupal stage, aquatic
insects are generally immobile, and must therefore rely on the position adopted in the final larval instar
to provide a continuous source of oxygen. Voelz and Ward (1996) found that the pupal cases of the
caddisfly Brachycentrus occidentalis were clumped on the underside of the downstream side of
boulders, a highly turbulent site which should provide high levels of dissolved oxygen while minimising
shear stresses that might dislodge the pupating insects. A different approach is adopted by pupating
blepharicerid larvae. Pommen and Craig (1995) found that morphological features of blepharicerid
pupae produce respiratory vortices that interact with the pupal gills to create a unique, and efficient,
method of plastronic gas exchange (see Vogel1994).
The "scorpion posture", adopted by the mayfly Ephemerella when confronted by predators, is
characterised by a vertical flexion and extension of the terminal abdominal segments and terminal
filaments . Although recognised as a defensive behaviour by Peckarsky and Penton (1988), they
ascribe it a largely mechanical function . Rather, this distinctive behaviour may play one of several
hydrodynamic roles. The mayfly C/oeon dipterum positions its abdomen in a similar manner in order
to initiate sufficient thrust during escape manoeuvres, during which accelerations can reach as high as
6.5 m s·2 (Craig 1990).

This suggests that the scorpion posture may reflect a 'pre-swimming'

readiness on the part of Ephemerella. Alternatively, the extension of the abdomen and anal cerci into
the upper levels of the benthic boundary layer may distort the flow fields around the animal's body,
confusing predators as to the precise location and identity of the prey.

The fluid mechanics of suspension feeding and 'erosional hunting'
Several species of stream macroinvertebrate manipulate the hydrodynamics of their
immediate surroundings in order to enhance feeding efficiency. Benthic taxa variously create vortices
that enhance the efficiency of specialised suspension-feeding structures, erode soft sediments to
expose buried prey, and re-suspend deposited organic food particles from the stream bed (see Table
1-1 ).

The creation and manipulation of vortices requires specialised morphological features and

33

behavioural adaptations. Consequently, taxa that employ hydraulic manipulation to acquire food are
generally among the most unique denizens of the benthic community.
The predatory larvae of the mayfly species Pseudiron centra/is employ an unusual method of
hunting for and feeding on their preferred prey, larval chironomids that live concealed in the benthic
sands of large northern rivers. Pseudiron larvae adopt a peculiar posture, positioning themselves on
the upstream slopes of sand dunes, arching their thorax upward while holding their head and
mouthparts close to the substratum (figure 1-5). Soluk and Craig (1990) found that this directs flow
downward, resulting in the generation of a solenoidal vortex in front of the animal. The increased
velocity of this 'horseshoe' vortex erodes sand in front of the animal, exposing their chironomid prey.
As prey becomes exposed, it is seized and ingested by the mayflies.

The mayfly slowly moves

backward, excavating prey in a continuous trench.
The predatory activities of the mobile stonefly Dinocras cepha/otes also result in the erosion
of fine sediments (Statzner et al. 1996). The ecological importance of this stonefly approaches that of
a keystone species (Jones et al. 1994), due to the large influence this species has on erosional
processes in stream riffle habitats. The night-time activity levels of D. cephalotes larvae increase
when prey are scarce; it is this patrolling for prey that leads to increased bioturbation of sediments .
Increases in sediment transport in turn lead to increases in the depth of interstitial spaces . However,
increased foraging activities were noted only in hydraulic habitats preferred by the stonefly (optimal
shear stresses of 0.39 - 1.58 N m·2 for Dinocras; see Peckarsky et al. 1990). The authors suggest
that increases in sediment entrainment by this stonefly are merely incidental, a behavioural 'byproduct' of increased activity. If local shear stresses are already close to incipient motion thresholds
for sand, then the hydrodynamic disturbance provided by the presence of a stonefly will likely result in
local increases in entrainment (Statzner et al. 1996).

However, it is possible that increases in

interstitial space could directly benefit these predators if localised erosion exposed larval chironomids,
a known prey item of other stoneflies of the same family (Perlidae, see Stewart and Harper 1996).
Soluk and Craig (1988) also investigated a unique method of suspension feeding utilized by
the lotic mayfly Ametropus neavei. This mayfly also inhabits marginal regions of shifting sands in
rivers throughout the holarctic region , and are found most commonly in velocities ranging from 18- 30
em s·1 • In relatively slow flows, these mayflies simply bury themselves , leaving only their head and
caudal filaments exposed (Clifford and Barton 1979). However, when the flows of their environment
34

exceed approximately 8 em s·1 , the behaviour of Ametropus changes . They orient themselves facing
into the flow, and clear a shallow pit immediately in front of their head, and extend their prothoracic
legs above them into the flow (Figure 1-6).

They sweep their prothoracic legs towards their

mouthparts every few seconds, with the rate of 'sweeping' increasing with velocity. Soluk and Craig
(1988) found that modified microtrichia found on the forelegs were capturing food particles from the
water column . The microtrichia function as an aerosol (or "hydrosol") filter (Braimah 1987), rather
than a sieve (LaBarbera 1984), in that particles impact or collide with filter elements due to fluid
dynamic forces .

Aerosol filters capture particles much smaller than the spacing between filter

elements (LaBarbera 1984).

The minimal surface area required for aerosol capture presumably

confers on Ametropus the same benefits it provides for other suspension-feeding aquatic fauna such
as black fly (Diptera: Simuliidae) larvae {Chance and Craig 1986)- relatively low pressure drag. This
reduction in drag is especially important for Ametropus, as the sand dune-like habitats inhabited by
this species do not provide a source of firm attachment. In addition, Soluk and Craig (1988) found that
the forelegs, which are presumably held in the upper levels of the boundary layer, direct relatively
rapidly moving water downward in to the pit. In conjunction with the head and antennae, this creates
a stable, horizontally rotating solenoidal vortex. The increased velocities of the redirected flow entrain
food particles, while the vortex serves to keep the particles suspended and passes them over the
collecting apparatus {hairy coxal plates, labrum, and mouthparts) multiple times. The authors suggest
that the pit may also act as depositional trap, because shear stresses in pits are typically lower than
those of surrounding sediments (Nowell and Jumars 1984 ).
Another group of suspension feeders that manipulate vortices to enhance feeding efficiency
are the larvae of the dipteran family Simuliidae {the blackflies). The various species of the genus
Simulium enjoy a nearly cosmopolitan distribution, and are found in the running waters of streams

from the arctic to the tropics (Craig 1977; Merritt et al. 1996). Due to the medical importance and
pestiferous nature of simuliid adults, the feeding and positioning behaviour of the larvae have been
studied extensively (Craig and Chance 1982; Chance and Craig 1986; Hart et al. 1991; Lacoursiere
1992; Merritt et al. 1996), leading to some success in controlling populations using particulate,
ingestible insecticides. As suspension feeders, simuliid larvae are elegantly adapted to moving water,
manipulating local flow to create specialised patterns of vortices that gather food and bring it to the
mouth, remove wastes, and facilitate respiration.

35

Chance and Craig (1986) present one of the most detailed studies on the interaction between
the behaviour and hydrodynamics of black fly larvae.

Larvae orient themselves with their

posterolateral surface upstream, and twist their bodies 90 - 180°. This longitudinal twisting of the body
results in one labral fan being closer to the substratum, while the other is held up into the upper
boundary layer (Chance and Craig 1986}. The rotation also results in the adoral (concave) surface of
the labral fans being exposed to the flow. With few exceptions (Craig 1977), simuliid larvae capture
food from water using these modified labral fans, whose microtrichial structure is modified to present a
nearly hydrodynamically ideal structure to the flow, resulting in minimal drag. The body is deflected
from vertical by the flow, with angle of deflection increasing with velocity. The angle of deflection was
once thought to be passive, regulated by drag acting on the body of the insect (Maitland and Penny
1967; Chance and Craig 1986).

Lacoursiere (1992), however, found that posture is in part

behaviourally determined. The feeding stance is an outcome of feedback between flow forces and
behaviour that acts to maintain the labral fans in an optimal feeding position. When disturbed, black
fly larvae exhibit a characteristic "avoidance reaction"; they cease feeding and pull down closer to the
substrate into the decreased velocities of the lower boundary layer.
The adaptive significance of the positioning of the labral fans is only fully realised when it is
placed in context of the flow patterns generated by the unique morphology of this insect (Figure 1-7).
In addition to the longitudinal twisting of the body, black fly larvae 'yaw' their bodies across the mean
flow up to 20°.

This posture has a peculiar influence on the vortices generated by the modified

cylindrical form of these larvae.

Flow approaching near the substratum slows as it reaches the

stagnation point at the front of the body, then accelerates (by a factor of 1.6 to 1.7) as the flow
separates and passes on either side of the body just outside the 'horseshoe' vortex (Chance and
Craig 1986).

If the body approximated a symmetrical, but tilted, cylinder, the paired vortices

generated would rise up the downstream side of the body, and detach at the top in a classic von
Karman trail (Vogel 1994 ). However, due to the fact that the body is positioned across the flow as well
as being twisted, only one of the paired vortices rises up the body while the other simply slips to the
side and is carried, rotating, downstream. The upper vortex, which has greater angular velocity than
the lower vortex due to the greater distance it must travel around the larval body, resuspends fine
particulate organic matter that has been deposited on the substrate. This vortex then rises up the
downstream side of the larval form, passing through the lower labral fan. As mean current velocities

36

increase, the angle of deflection of the larval body increases, resulting in the larvae lying closer and
more parallel to the substrate. Although the cross-sectional area of the vortex discharging into the
lower labral fan decreases, overall discharge changes very little due to increased velocities and a
thinner boundary layer (Vogel 1994).
While the lower labral fan filters water only from the attached , rising vortex, the upper labral
fan captures food particles from an entirely different volume of water. The upper labral fan is held up
into the upper levels of the boundary layer, where it is exposed to velocities higher than that of the
body and lower labral fan . As velocities increase and angle of bodily deflection increases, the labral
fan is exposed to higher velocities, but is maintained at approximately the same position in the
boundary layer ( Lacoursiere 1992). The filtering efficiency of the labral fans is reduced as velocity
decreases. Larval blackflies decrease the aperture opening between the rays of the labral fan as
velocities decrease. At low velocities, Reynolds numbers are sufficiently low that a 'viscous zone'
(viscous effects predominating at low Reynolds numbers at the scale of individual filter elements; see
Braimah 1987) occupies most of the apertural space surrounding the rays and microtrichia.
Lacoursiere and Craig (1993) estimated that 97% of the apertural space was occupied by this 'viscous
zone' at velocities of 3.6 em s·1 . However, as Reynolds numbers increase, less of the apertural space
is occupied by these viscous effects (28% at 40 em s·1 ; Lacoursiere and Craig 1993).

Directions for future research
The behavioural and morphological adaptations of benthic invertebrates to the unidirectional
flow of stream environments are diverse.

Yet, due to the cosmopolitan nature of stream

hydrodynamics in lotic systems worldwide, stream ecologists can enjoy an unparalleled discourse on
the complex interactions between physics and biology in rivers and streams. The challenge that lies
before behavioural and physical ecologists is twofold . First, data relating the biology of stream taxa to
hydraulics must be collected and analysed in a manner appropriate to the question asked. Too often,
authors seem content with a level of discussion that approximates 'this species seems to prefer
medium velocities'. If a more complete understanding of the role that moving water plays in stream

ecosystems is ever to be achieved, stream ecologists must become more comfortable with the
concepts of shear, eddy production , and boundary layers (among others). Second, the nature of the
flow that is actually experienced by animals on the substrate must be more thoroughly examined. The
37

validity of studies of large-scale processes and patterns will continue to be limited unless we
understand the nature of flow in natural systems at organism-defined scales.

38
v '! ' '

Table 1-1. Selected examples of behavioural and morphological
adaptations of stream invertebrates to hydrodynamics
Category
Morphology I
flow interactions

Micropositioning

Taxon I Adaptation (reference)

Inferred functional advantage

1. Larvae of the caddisfly Sericostoma se/ysi
from high velocity areas had heavier, narrower
cases than those larvae from lower velocities
(Delgado and Carbonell 1997).

1. Larvae in high velocities are more easily
able to resist drag, reducing the risk of
becoming detached from the bed.

2. Water penny larvae (Sclerocyphon ;
Coleoptera: Psephenidae) were highly
streamlined , and controlled the boundary layer
over their bodies by suction through the lateral
laminae (Smith and Dartnall 1980).

2. Streamlining and boundary layer suction
allow larvae to maintain their position and
move over the substrate in high velocity,
highly turbulent flows .

3. Flow lines were compressed over the front
of the Ecdyonurus cf. venosus, and boundary
layer separation occurred over the bodies of E.
venosus and Ancy/us fluviatilus (Gastropoda)
(Statzner and Holm 1982).

3. Lift is concentrated over the thorax of E.
venosus (where the legs are in contact with
the substrate), and boundary layer
separation over both taxa suggests that the
boundary layer may not be thick enough to
provide substantial protection from high
near-bed velocities.

4. Steepest velocity gradients were found
close to the bodies of several benthic taxa
(Gastropods Ancylus, Acroloxus, and
Potamopyrgus, the amphipod Gammarus, the
larval caddisflies Anabolia , Micrasema, and
Silo), where parts of their bodies protruded
furthest into the flow (Statzner and Holm 1989).

4. Simultaneous morphological adaptations
to the forces of lift, pressure and friction
drag, erosion , diffusion are impossible;
adaptations will be a reflection of the
organisms size and Reynolds number, and
will change as the organism grows.

5. The mayfly Ecdyonurus tilted its head
shield, and had femurs shaped like 'spoilers',
while the mayfly Epeorus resisted lift by
anchoring itself to the substrate . The drag
forces experienced by the invertebrates were
proportional to the square of the velocity
(Weissenberger et al. 1991 ).

5. The morphological adaptations of
Ecdyonurus produce 'negative' lift, and press
its body against the substrate in highvelocity flows . Epeorus experiences high lift
due to its 'airfoil' shape, but counters this by
attaching itself firmly to the substrate.

1. Abundances of suspension-feeding black fly
(Simulium vittatum) larvae were significantly
related to velocity measured 2 mm above the
bed. Maximum near-bed acclerations
exceeded 1 x 104 em s·2 (Hart et al. 1996).

1. Particle interception rates are a function
of near-bed velocities and turbulence
intensities; S. vittatum larvae position
themselves in locales that maximise rates of
particle delivery. Forces due to acceleration
reaction may be greater than the forces
imposed by pressure drag on these
suspension-feeding larvae.

2. Simulium vittatum larvae gathered in
boundary layer separation zones, and avoided
areas of maximal shear stress (Lacoursiere
1992).

2. Larvae can detect (and migrate to) areas
of the bed with velocity profiles that will
maximise particle flux through their labral
fans while minimising drag on the bulbous
posterior portion of their abdomen .

3. The suspension-feeding caddisflies
Arctopsyche grandis and Brachycentrus
occidentalis preferred the bottoms and lower
lateral sides of rocks rather than the upper,
current-exposed, surfaces (Voelz and Ward
1996).

3. The microdistributions of A. grandis and
B. occidentalis may minimise predation and
the risk of accidental detachment while
allowing for 'acceptable' rates of particle
capture and respiration.

4. The suspension-feeding caddisfly
Brachycentrus occidentalis preferred
microhabitats with higher velocities, shallower
depths, higher water surface slopes, and
higher Froude numbers (Wetmore et al. 1990).

4. Larvae may choose sites with
accelerating flow to maximise particlecapture rates; however, as B. occidentalis
larvae must extend their legs into the flow to
suspension-feed, velocities must not be so
high that food particles are washed away
before they can be transferred to the mouth.

39

Habitat selection

1. The macroinvertebrate communities
associated with the front (solenoidal vortex) and
downstream (wake) regions of stream boulders
were distinctly different (Bouckaert and Davis
1998).

1. Benthic fauna associated with different
microflow regions around stream boulders
may not be responding to water velocity,
but rather to other hydrodynamic factors
(i.e. turbulence intensity) and/or factors
related to hydrodynamics that are directly
linked to the ecology of benthic organisms
(i.e. deposition of particulate organic
matter, exchange of dissolved gases).

2. Invertebrate abundance was positively

2. Invertebrate taxa with similar

correlated with roughness and negatively
correlated with slope. Associations of taxa with
similar adaptations to flow to specific velocities,
Reynolds numbers, roughness Reynolds
number, and shear velocity were noted (Growns
and Davis 1994 ).

morphological or behavioural adaptations to
flow can be classified as belonging to one
of three 'flow exposure groups': obligates
(found in turbulent, high shear habitats),
facultatives (found in habitats with greater
bed roughness), and avoiders (found in
high velocity habitats where fluxes of
dissolved oxygen and organic matter may
be maximised).

3. Densities of the mayflies Rhithrogena and
Baetis were greatest at depths of 1.5 m before
flooding , but shifted to depths of 0.5 m and 0.2

3. Many invertebrate taxa move from
deeper water to shallower water during
floods, possibly using the shore zone as a
flow refugium during flood events.

4. Distinct microcrustacean assemblages were

4. Areas of low shear stress ('refugia') may
enable the survival of epbenthic
microcrustacea during periods of high
discharge.

5. A relationship was found between the
microdistribution of several stream taxa and
various descriptors of the near-bed flow
environment (including the estimated thickness
of the laminar sublayer) (Statzner 1981 ).

5. The distribution of various taxa reflects
their degree of rheophily and dependence
on high-velocity microhabitats. Many
macroinvertebrates (e.g. suspensionfeeders) prefer habitats with specific
hydraulic characteristics (e .g. certain
ranges of turbulence, viscous sublayer
thicknesses, etc.)

1. Larger nymphs of the rheophilous mayflies
Rhithrogena semicolorata and Ecdyonurus gr.
venosus preferred rougher substrata (Buffagni et
al. 1995).

1. As nymphs increase in size, their
hydraulic preferences (i.e. for specific
velocities or turbulence intensities) change;
larger nymphs may be able to move and
graze successfully in high-velocity habitats
where smaller nymphs would be swept
from the bed.

2. Larger net-spinning hydropsychid caddisfly
larvae were found at higher velocities than
smaller, less mature larvae. However, net mesh
size did not increase with velocity among
species (Osborne and Herricks 1987).

2. Body size, rather than net mesh size,
limits the distribution of net-spinning
caddisflies.

m during periods of high discharge. Densities of
the suspension-feeding caddisfly Hydropsyche
were greatest at 1.5 m during all months;
however, the location of the 1.5 m depth shifted
laterally during floods (Rempel et al. 1999).

associated with areas of the streambed that
retained low shear stress during periods of high
discharge (Robertson et al. 1997).

Ontogenetic
shifts in habitat
selection

40

Suspension
feeding

1. The body of Simulium vittatum produced
downstream, paired vortices; one of the vortices
rose up the downstream side of the body, and
passed through the lower of the two Ia bra I fans .
The second , higher labral fan filtered water from
the upper boundary layer. Larvae positioned
side-by-side mutually enhanced flow between
them (Chance and Craig 1986).
2. The feeding stance of Simulium vittatum was

found to be the result of feedback between the
behaviour of the larvae and drag forces . The
aperture size of the labral fans (filtering
elements) increased with increases in velocity
(Lacoursiere and Craig 1993).

3. The mayfly Ametropus neavi suspension-fed
using modified foreleg trichia, and by inducing a
horizontally-rotating solenoidal vortex within a
specially constructed pit (Soluk and Craig 1988).
4. The larvae of four species of net-spinning

hydropsychid caddisflies (Hydropsyche betteni,
H. spama, H. cheilonis, and H. bronta) were
found primarily in areas with rapidly spiralling
vortices (Osborne and Herricks 1987).

Predator
avoidance

1. Black fly larvae abundance (Simulium

vittatum) was negatively related to flatworm
abundance (Dugesia dorotocephala, a predator)
and positively related to velocity (Hart and Merz

1998).

2. Densities of prey (Baetis rhodani and
Chironomidae) were reduced by the stonefly
predator Dinocras cephalotes only in hydraulic
regimes favourable to the predator (Peckarsky
et al. 1990).

Predation

Movement
(swimming)

1. The downstream vortex created by the
body of S. vittatum enhances feeding (by
entraining particles from the substrate).
The upper labral fan captures entrained
particles convected from upstream. The
morphology of these larvae also minimise
drag while maximising particle-capture
rates .

2. The interaction between behaviour and

morphology maximises the particle-capture
efficiency of the larvae (4- 26 times more
efficient than previously thought).

3. The vortex enhances the delivery of
seston to the particle-capture elements,
and allows larvae to inhabit the
noncohesive beds of sand-bottom rivers .
4. Particle-capture rates of net-spinning

caddisflies are dependent on flow patterns,
not just velocity. Larvae preferred faster,
turbulent flows to slower, more laminar
flows. Turbulence (especially spiralling
flows) may entrain particles or increase the
flux of particles through the net of
hydropsychid caddisflies.

1. S. vittatum larvae prefer faster velocities
not only to maximise rates of suspension
feeding, but also to minimise the risk of
predation by predators that are not adapted
to high velocity habitats.
2. Stream prey species can minimise the

threat of predation by moving into areas of
the streambed with abiotic regimes
unfavourable to predators (hydraulic
refugia).

1. Predation by the caddisfly predator
Rhyacophi/a obliterata on larval blackflies
(Simulium sublacustre) was greatest in areas of
greatest microhabitat overlap (where velocities
ranged from 40-60 em s·1 } (Muotka and
Penttinen 1994).

1. Prey acquisition rates will be greatest if
predators are able to successfully access
the hydraulic habitats inhabited by prey
species.

2. The stonefly predator Kogotus modestus
used the pressure wave patterns produced by
escaping mayfly prey to distinguish between
non-preferred (Ephemeral/a infrequens) and
preferred (Baetis bicaudatus) prey species
(Peckarsky and Wilcox 1989).

2. Use of hydrodynamic cues allows K.
modestus to selectively predate on
preferred prey.

3. Larvae of the mayfly Pseudiron centra/is
arched their bodies and lowered their heads to
create a solenoidal vortex , which eroded a pit in
front of their bodies (Soluk and Craig 1990).

3. The solenoidal vortex erodes
chironomid prey from the sandy sediments
that P. centra/is larvae inhabit; the mayfly
larvae seize any prey exposed in the
eroded pit.

1. The mayfly C/oeon dipterum was found to
have a highly streamlined shape, and small, but
projecting, gills (Craig 1990).

1. The streamlined shaped of C. dipterum
confers no advantage during swimming
(due to relatively low Re}, but becomes
important during 'acclerative escape
motions', as a streamlined shape gives a
low coefficient of added mass.

41

Respiration 1

1. Flow variation did not influence the
respiration rates of the mayfly Cinygmula
grandifo/a, but at lower velocities the respiration
rates of the stoneflies Stenopsyche mannorata
and Skwala pusilla were compromised
(Golubkhov et al. 1992).

1. The stonefly taxa generally inhabit highvelocity microhabitats, and as such, are
unable to respire efficiently unless gas
exchange is enhanced by water flowing
over their gills . C. grandifola , generally
found in interstitial spaces or low-flow areas
of the streambed , is not dependent on
flowing water to maintain respiration rates .

2. The pupa and gills of Blepharicerid larvae
acted as 'bluff bodies', and paired vortices (and
associated regions of low pressure) formed
downstream of the gills (Pommen and Craig
1995).

2. By generating vortices, the pupal gills of
Blepharicerid larvae enhance gas
exchange rates (due to the reduced
solubility of air in low-pressure areas).

3. At low dissolved oxygen levels, mayflies
moved into positions on the substrate that were
more exposed to the current (Wiley and Kohler
1980).

3. Stream mayflies are able to
behaviourally compensate for low oxygen
levels by moving into high velocity
microhabitats; presumably, respiration in
these locations is enhanced by thinning
boundary layers over their gills, and
subsequently increasing rates of diffusion
of dissolved gases.

42

Acknowledgements
Special thanks goes to all researchers whose work motivated me, whose publications I found
to be riveting accounts of the intimate relationship between moving water and the lives of organisms,
and whose obvious love and knowledge of physical ecology continue to inspire me. This list, though
far from complete, includes Mike Church, Hugh Clifford, Doug Craig, Jenny Davis, Mark Denny, Chris
Finelli, Charlotte Gjerlov, David Hart, Alan Hildrew, Jill Lancaster, Bob Newbury, John Richardson,
Bobbi Peckarsky, B. Statzner, Steven Vogel, and Yixin Zhang.

43

References
Abelson , A. 1997. Settlement in flow: upstream exploration of substrata by weakly swimming larvae.
Ecology 78 : 160-166.
Ackerman, J.D., L. Wong, C. R. Ethier, D.G . Allen, and J.K. Spelt. 1994 . Preston-static tubes for the
measurement of wall shear stress. J. Fluids Eng . 116: 645-649 .
Ackerman , J.D., and T.M . Hoover. 2001 . Measurement of local bed shear stress in streams using a
Preston-static tube. Limnol. Oceanogr. 46: 2080-2087.
Allan , J.D., A.S . Flecker, and N.L. McClintock. 1986. Diel epibenthic activity of mayfly nymphs, and its
nonconcordance with behavioural drift. Limnol. Oceanogr. 31 : 1057-1065.
Ambuhl, H. 1959. Die Bedeutung der stromung als okologischer faktor. Schweizerische Zeitschrift fur
Hydrologie 21 : 133-264.
Bergeron, N.E., and A.D. Abrahams. 1992. Estimating shear velocity and roughness length from
velocity profiles. Water Resour. Res. 28: 2155-2158 .
Bouckaert, F.W., and J. Davis. 1998. Microflow regimes and the distribution of macroinvertebrates
around stream boulders. Freshwater Bioi. 40: 77-86.
Braimah, S.A. 1987. The influence of water velocity on particle capture by the labral fans of larvae of
Simulium vittatum (Diptera: Simuliidae). Can . J. Zool. 65: 2359-2399.
Buffagni, A., G. Crosa, and R. Marchetti. 1995. Size related shifts in the physical habitat of two
mayfly species (Ephemeroptera). Freshwater Bioi. 34: 297-302.
Carling, P. 1992. The nature of the fluid boundary layer and the selection of parameters for benthic
ecology. Freshwater Bioi. 28: 273-284.
Chance, M.M. , and D.A. Craig . 1986. Hydrodynamics and behaviour of Simuliidae larvae (Diptera).
Can. J. Zool. 64 :1295-1309.
Clifford, H. F., and D. R. Barton. 1979. Observations on the biology of Ametropus neavei
(Ephemeroptera: Ametropododae) from a large river in northern Alberta, Canada. Can.
Entomol. 111 : 855-858 .
Collier, K. 1994. Influence of nymphal size, sex and morphotype on microdistribution of De/eatidium
(Ephemeroptera: Leptophlebiidae) in a New Zealand river. Freshwater Bioi. 31: 35-42.
Collier, K.J ., G.F. Croker, C.W. Hickey, J.M. Quinn, and B.S. Smith . 1995. Effects of hydraulic
conditions and larval size on the microdistribution of Hydrobiosidae (Trichoptera) in two New
Zealand rivers. New Zea . J. Mar. Fresh . Res. 29(4): 439-451.
Connell, J. H. 1978. Diversity in tropical rain forests and coral reefs. Science 199: 1302-1310.
Craig, D.A. 1977. Mouthparts and feeding behaviour of Tahitian larval Simuliidae (Diptera :
Nematocera). Quaest. Entomol. 13: 195-281 .
Craig, D.A. 1990. Behavioural hydrodynamics of Cloeon dipterum larvae (Ephemeroptera: Baetidae).
J. N. Am. Benthol. Soc. 9: 346-357 .
Craig, D.A. 1993. Hydrodynamic considerations in artificial stream research. J. N. Am. Benthol. Soc.
12: 324-328 .

44

Craig, D.A. 1996. Some of what you should know about water, or K.I.S .s : for hydrodynamics
('Keeping It Stupidly Simple). http://www.inhs.uiuc.edu./nabs/resrc/kiss.htm (reprinted from
Bull. N. Am. Benth . Soc. 35: 178-182).
Craig, D.A., and M.M. Chance. 1982. Filter feed ing in larvae of Simuliidae (Diptera: Culicomorpha):
aspects of functional morphology and hydrodynamics. Can. J. Zool. 60: 712-724 .
Davis, J.A. 1986. Boundary layers, flow microenvironments, and stream benthos. in Limnology in
Australia, eds . P. De Deckker and W.O. Williams, Dr W. Junk Publishers, East Melbourne,
Australia.
Davis, J.A., and L.A. Barmuta. 1989. An ecologically useful classification of mean and near-bed flows
in streams and rivers. Freshwater Bioi. 21: 271-282.
Delgado, J.A., and R. Carbonell. 1997. Case features of caddisfly larvae (Sericostoma selysl) as
related to water velocity and potential to drift. J. Freshwater Ecol. 12: 193-197.
Forrester, G.E. 1994. Diel patterns of drift by five species of mayfly at different levels of fish
predation. Can . J. Fish . and Aquat. Sci. 51: 2549-2557.
Fuller, R.L., and R.J . Mackay. 1980. Feeding ecology of three species of Hydropsyche (Trichoptera:
Hydropsychidae) in southern Ontario. Can J. Zool. 58: 2239-2251 .
Golubkhov, S.M., T.M. Tiunova, and S.L. Kocharina. 1992. Dependence of the respiration rate of
aquatic insects upon the oxygen concentration in running and still water. Aquatic Insects 14:

137-144.

Gore,

J .A.
1978.
A technique for predicting in-stream flow
macroinvertebrates. Freshwater Bioi. 8: 141-151.

requirements

of benthic

Growns, 1.0., and J.A. Davis . 1994. Longitudinal changes in near-bed flows and macroinvertebrate
communities in a Western Australian stream . J. N. Am. Benthol. Soc. 13: 417-438.
Hansen, R.A. , D.O. Hart, and R.A. Merz. 1991. Flow mediates predator-prey interactions between
triclad flatworms and larval black flies. Oikos 60: 187-196.
Hart, D.O., and C.M. Finelli. 1999. Physical-biological coupling in streams : the pervasive effects of
flow on benthic organisms. Ann . Rev. Ecol. Syst. 30: 363-395.
Hart, D.O., and R.A. Merz. 1998. Predator-prey interactions in a benthic stream community: a field
test of flow mediated refuges . Oecologia 114: 263-273.
Hart, D.O., R.A. Merz, S.J. Genovese, and B.D. Clarke. 1991. Feeding posture of suspensionfeeding larval black flies : the conflicting demands of drag and food acquisition . Oecologia
(Berlin) 85: 457-463.
Hart, D.O., B.D . Clark, and A. Jasentuliyana. 1996. Fine-scale field measurements of benthic flow
environments inhabited by stream invertebrates. Limnol. Oceanogr. 41 : 297-308 .
Hildrew, A.G., and C.R. Townsend . 1982. Predators and prey in a patchy environment: a freshwater
study. J. Anim. Ecol. 51 : 797-815.
Holomuzki, J.R., and S.H. Messier. 1993. Habitat selection by the stream mayfly Paraleptophlebia
guttata. J. N. Am . Benthol. Soc. 12: 126-135.
Jones, C.G ., J.H. Lawton, and M. Shachak. 1994. Organisms as ecosystem engineers. Oikos 69:

373-386.

Jowett, I.G., J. Richardson, B.J.F. Biggs, C.W. Hickey, and J.M. Quinn . 1991 . Microhabitat
preferences of benthic invertebrates and the development of generalized De/eatidium spp .

45

habitat suitability curves, applied to four New Zealand Rivers . New Zea. J. Mar. Fresh . Res .
25: 187-200.
LaBarbera, M. 1984. Feeding currents and particle capture mechanisms in suspension feeding
animals . Am . Zool. 24: 71-84.
Lacoursiere, J.O. 1992. A laboratory study of fluid flow and microhabitat selection by larvae of
Simulium vittatum (Diptera: Simuliidae). Can. J. Zool. 70: 582-596 .
Lacoursiere, J.O., and D.A. Craig . 1990. A small flume for studying the influence of hydrodynamic
factors on benthic invertebrate behaviour. J. N. Am . Benthol. Soc. 9: 358-367.
Lacoursiere, J.O., and D.A. Craig . 1993. Fluid transmission and filtration efficiency of the labral fans
of black fly larvae (Diptera: Simuliidae): hydrodynamic, morphological , and behavioural
aspects. Can . J. Zool. 71: 148-162.
Lancaster, J. 1996. Scaling the effects of predation and disturbance in a patchy environment.
Oecologia 107(3): 321-331 .
Lancaster, J., A.G. Hildrew, and C.R. Townsend . 1990. Stream flow and predation effects on the
spatial dynamics of benthic invertebrates. Hydrobiologia 31: 337-349 .
Lancaster, J., and A.G . Hildrew.
1993.
Flow refugia and the microdistribution of lotic
macroinvertebrates. J. N. Am. Benth. Soc. 12: 385-393 .
Lancaster, J., A.G. Hildrew, and C. Gjerlov. 1996. Invertebrate drift and longitudinal transport
processes in streams. Can. J. Fish . Aquat. Sci. 53: 572-582.
Lancaster, J., and A. L. Robertson. 1995. Microcrustacean prey and macroinvertebrate predators in
a stream food web. Freshwater Bioi. 34: 123-134.
Maitland, P. S., and Penny, M. M. 1967. The ecology of the Simuliidae in a Scottish River. J. Anim .
Ecol. 36 : 179-206.
· Malmqvist, B. , and G. Englund. 1996. Effects of hydropower-induced flow perturbations on mayfly
(Ephemeroptera) richness and abundance in north Swedish river rapids . Hydrobiologia 341 :
145-158.
Menge, B.A. 1976. Organization of the New England rocky intertidal community: role of predation,
competition, and environmental heterogeneity. Ecol. Monogr. 46: 355-369.
Merritt, R. W. 1987. Do different instars of Aedes triseriatus feed on particles of the same size? J.
Am. Mosq. Control Assoc. 3: 94-96.
Merritt, R.W., D.A. Craig, R.S. Wotton , and E.D. Walker. 1996. Feeding behaviour of aquatic insects:
case studies on black fly and mosquito larvae. Invert. Bioi. 115: 206-217.
Muotka, T., and A. Penttinen . 1994. Detecting small-scale spatial patterns in lotic predator-prey
relationships: statistical methods and a case study. Can. J. Fish. Aquat. Sci. 51 : 2210-2218 .
Nowell, A.R.M., and M. Church . 1979. Turbulent flow in a depth-limited boundary layer. J. Geophys.
Res . 84: 4816-4824.
Nowell, A.R.M ., and P.A. Jumars. 1984. Flow environments of aquatic benthos. Ann. Rev. Ecol.
Syst. 15: 303-328.
Osborne, L.L. , and E.E. Herricks. 1987. Microhabitat characteristics of Hydropsyche (Trichoptera :
Hydropsychidae) and the importance of body size. J. N. Am. Benthol. Soc. 6: 115-124.

46

Palmer T.M. 1995. The influence of spatial heterogeneity on the behaviour and growth of two
herbiverous stream insects. Oecologia 104: 476-486 .
Peckarsky, B.L. 1996. Alternative predator avoidance syndromes of stream-dwelling mayfly larvae.
Ecology 77: 1888-1905.
Peckarsky, B.L., and M.A. Penton. 1988.
stoneflies . Ecology 70: 1203-1218.

Mechanisms of prey selection by stream-dwelling

Peckarsky, B.L. , and R.S. Wilcox. 1989. Stonefly nymphs use hydrodynamic cues to discriminate
between prey. Oecologia 79: 265-270.
Peckarsky, B.L., S.C. Horn, and B. Statzner. 1990. Stonefly predation along a hydraulic gradient: a
field test of the harsh-benign hypothesis. Freshwater Bioi. 24: 181-191.
Peckarsky, B.L., and C.A. Cowan. 1995. Microhabitat and activity periodicity of predatory stoneflies
and their mayfly prey in a western Colorado stream . Oikos 74(3): 513-521.
Poff, N.L., and J.V. Ward . 1991 . Drift responses of benthic macroinvertebrates to experimental
streamflow variation in a hydrologically stable stream. Can . J. Fish . Aquat. Sci. 48 : 19261936.
Poff, N.L., N.J. Voelz, J.V. Ward, and R.E. Lee. 1990. Algal colonization under four experimentallycontrolled current regimes in a high mountain stream . J. N. Am. Benthol. Soc. 9: 303-318.
Pommen, G.D.W., and D.A. Craig. 1995. Flow patterns around gills of pupal net-winged midges
(Diptera: Blephariceridae): possible implications for respiration . Can. J. Zool. 73: 373-382 .
Prantl, L. 1952. Essentials of Fluid Dynamics. Hafner, New York, NY.
Preston, J.H. 1954. The determination of turbulent skin friction by means of Pitot tubes . J. Royal
Aeronaut. Soc. 58 : 109-121.
Quinn, J.M., and C.W. Hickey. 1994. Hydraulic parameters and benthic invertebrate distributions in
two gravel-bed New Zealand rivers . Freshwater Bioi. 32 : 489-500.
Quinn, J.M., C.W. Hickey, and W. Linklater. 1996. Hydraulic influences on periphyton and benthic
macroinvertebrates: simulating the effects of upstream bed roughness. Freshwater Bioi. 35:
301-309.
Rader, R.B., and J. V. McArthur. 1995. The relative importance of refugia in determining the drift and
habitat selection of predaceous stoneflies in a sandy bottomed stream . Oecologia, 103: 1-9.
Rempel, L.L. , J.S. Richardson , and M.C. Healey. 1999. Flow refugia for benthic macroinvertebrates
during flooding of a large river. J. N. Am . Benthol. Soc. 18: 34-48 .
Resh, V. H., and D. M. Rosenberg.
Publishers.

1984. The Ecology of Aquatic Insects. New York: Praeger

Resh, V. H., and D. M. Rosenberg. 1989.
insects. Can. Ent. 121 : 941 -963.

Spatial-temporal variability and the study of aquatic

Richardson, W. B. 1992. Microcrustacea in flowing water: experimental analysis of washout times and
a field test. Freshwater Bioi. 28 : 217-230 .
Robert, A., A.G . Roy, and B. De Serres . 1996. Turbulence at a roughness transition in a depth
limited flow over a gravel bed . Geomorphology 16: 175-187.

47

Robertson, A.L., J. Lancaster, L.R. Belyea, and A.G. Hildrew. 1997. Hydraulic habitat and the
assemblage structure of stream benthic microcrustacea. J. N. Am. Benthol. Soc. 16: 562575.
Sagar, P.M., and G.J. Glova. 1992. Diel changes in the abundance and size composition of
invertebrate drift in five rivers in South Island, New Zealand . N. Zea. J. Mar. Freshwater Res .
26:103-114.
Shiozawa, O.K. 1991. Microcrustacea from the benthos of nine Minnesota streams.
Benthol. Soc. 10: 286-299.

J. N. Am.

Sih, A., and D.E. Wooster. 1994. Prey behaviour, prey dispersal, and predator impacts on stream
prey. Ecology 75:1199-1207.
Smith, J.A., and A.J . Dartnall. 1980. Boundary layer control by water pennies (Coleoptera:
Psephenidae). Aquat. Insects 2: 65-72.
Soluk, D.A., and D.A. Craig. 1988. Vortex feeding from pits in the sand : a unique method of
suspension feeding used by a stream invertebrate. Limnol. Oceanogr. 33: 638-645.
Soluk, D.A., and D.A. Craig . 1990. Digging with a vortex: flow manipulation facilitates prey capture by
a predatory stream mayfly. Limnol. Oceanogr. 35: 1201-1206.
Statzner, B. 1981. The relationship between "hydraulic stress" and microdistribution of benthic
macroinvertebrates in a lowland running water system, the Schierenseebrooks (North
Germany). Arch . Hydrobiol. 91 : 192-218.
Statzner, B., U. Fuchs, and L.W. Higler. 1996. Sand erosion by mobile predaceous stream insects:
implications for ecology and hydrology. Water Resour. Res . 32: 2279-2287 .
Statzner, B. , J.A. Gore, and V.H . Resh. 1988. Hydraulic stream ecology: observed patterns and
potential applications. J. N. Am. Benthol. Soc. 7: 307-360 .
Statzner, B., and B. Higler. 1986. Stream hydraulics as a major determinant of benthic invertebrate
zonation patterns. Freshwater Bioi. 16: 127-139.
Statzner, B., and T.F. Holm. 1982. Morphological adaptations of benthic invertebrates to stream flow:
an old question studied by means of a new technique (Laser Doppler Anemometry).
Oecologia 53 : 290-292.
Statzner, B., and T.F. Holm. 1989. Morphological adaptation of shape to flow: microcurrents around
lotic macroinvertebrates with known Reynolds numbers at quasi-natural flow conditions .
Oecologia 78: 145-157.
Statzner, B., and R. Muller. 1989. Standard hemispheres as indicators of flow characteristics in lotic
benthos research. Freshwater Bioi. 21: 445-459.
Stewart, K.W., and P.P Harper. 1996. Plecoptera . in An Introduction to the Aquatic Insects of North
America. Eds. R.W. Merritt and K.W. Cummins. Kendai/Hunt Publishing Co., Dubuque, lA.
Tikkanen, P., T. Muotka, A. Huhta, and A. Juntunen . 1997. The roles of active predator choice and
prey vulnerability in determining the diet of predatory stonefly (Piecoptera) nymphs. J. Anim .
Ecol. 66: 36-48.
Townsend, C. R., and A. G. Hildrew. 1979. Form and function of the prey catching net of
Plectrocnemia conspersa larvae (Trichoptera). Oikos 33:412-418.
Townsend, C.R., M.R. Scarsbrook and S. Doledec. 1997. The intermediate disturbance hypothesis,
refugia, and biodiversity in streams . Limnol. Oceanogr. 42: 938-949.
48

Voelz, N.J., and J.V. Ward . 1996. Microdistributions of filter-feeding caddisflies (Insecta: Trichoptera)
in a regulated Rocky Mountain river. Can. J. Zool. 74: 654-666 .
Vogel , S. 1994. Life in moving fluids . Princeton, NJ. : Princeton University Press.
Wallace, J.B., and R.W. Merritt. 1980. Filter feeding ecology of aquatic insects. Ann. Rev. Entomol.
25: 103-132.
Walsh , G., R. Morins, and R.R. Naimans. 1988. Daily rations, diel feeding activity and distribution of
age-0 brook trout, Sa/velinus fontinalis, in two subarctic streams. Environ . Bioi. Fishes 21 :
195-205.
Weissenberger, J., H.-Ch. Spatz, A. Emanns, and J. Schwoerbel. 1991 . Measurement of lift and
drag forces in the mN range experienced by benthic arthropods at flow velocities below 1.2
m·s·1 • Freshwater Bioi. 25:21 -31 .
Wetmore, S.H., R.J. Mackay, and R.W. Newbury. 1990. Characterization of the hydraulic habitat of
Brachycentrus occidentalis, a filter-feeding caddisfly. J. N. Am . Benth . Soc. 9: 157-169.
White, F.M . 1999. Fluid Mechanics- 41h Edition. WCB McGraw-Hill, Boston.
Wiley, M.J., and S.L. Kohler. 1980. Positioning changes of mayfly nymphs due to behavioral
regulation of oxygen consumption . Can . J. Zool. 58 : 618-622.
Wiley, M., and S.L. Kohler. 1984. Behavioural adaptations of aquatic insects. in The Ecology of
Aquatic Insects. Eds. V.H . Resh and D.M. Rosenberg . Praeger Publishers, New York.
Winterbottom, J.H., S.E. Orton, and A.G. Hildrew. 1997. Field experiments on the mobility of benthic
invertebrates in a southern English stream. Freshwater Bioi. 38: 37-4 7.
Wooster, D., and A. Sih . 1995. A review of the drift and activity responses of stream prey to predator
presence. Oikos 73: 3-8.
Young, W.J . 1992. Clarification of the criteria to identify near-bed flow regimes. Freshwater Bioi. 28:
383-391.
Young, W.J. 1993. Field techniques for the classification of near-bed flow regimes. Freshwater Bioi.
29: 377-383.

49

Chapter 1 - Figures

-

10

.

fully turbulent

,....

en
E
..._...
I

::J
>.

·c:;
0

Q)

>

0.1
0.01
0.001

laminar

0.0001
0.00001
0.001

0.1

0.01

10

depth D (m)
Figure 1-1 . Reynolds number conditions for the occurrence of turbulent flows in open channels (see
text for description of flow conditions in each case).

N
Q)

.~
......
en

outer layer

-------------------------------------- --------------

..c
::::l

en-.

~~

0

log layer

..c

ro

------viscous-su&Ta-er__ _
velocity U (m s- 1 )

A.

50

-

..,....

.C/)

0.5 ,...-- - -- - - - - - - -- - ----,
0.4

E
'-"' 0.3
0.2
0.1

0

,. ,.

--------------~--------------~

0.1

0.01

0.001

height above substrate z (m)

B.
Figure 1-2. (A) Representation of the boundary layer in shallow water. The flow conditions of the
bed layer may be turbulent or laminar, depending on hydraulic conditions. (B) The logarithmic
boundary layer plotted on log-normal axes . The shear velocity (U*) is inversely proportional to the
slope of the profile. The characteristic roughness length (Zo) can be estimated as the x-intercept of
the regression line. U* and z0 can be estimated from the profile only if the profile is log-normal
(modified from Carling 1992).

I

u+

........................................:......... cS
...

I

...

I

... :
... ,.
I

I

IJJ-'

----t••,. ,//

I

'tw

Figure 1-3. Diagram of the forces acting on a particle within a velocity gradient (velocity gradient
shown as vectors on the left side of the diagram). The force due to lift (F1) acts to 'pull' the particle
upwards, the force due to drag (Fd; integrated pressure drag and skin friction) acts to 'pull' the particle
downstream, while shear stress (tw) acts to 'roll' the particle in the downstream direction. The
thickness of the boundary layer (the height above the substrate at which velocities reach 90% of
mainstream velocities) is denoted as 8. Flow is from left to right.

51

Figure 1-4. (a) Dorsal and (b) ventral views of a Sclerocyphon (Coleoptera : Psephenidae) larvae,
showing the streamlined body form and lateral laminae. 1. Operculum covering gills; 2. Slot between
lateral laminae. (reproduced from Smith and Dartnall 1980).

Figure 1-5. Diagram of the patterns of flow used by Pseudiron centra/is (Ephemeroptera:
Pseudironidae) to excavate prey from sand beds. This mayfly positions its body to create the
necessary conditions for the formation of a solenoidal ("horseshoe") vortex, which it then uses to
excavate prey from the sediment (reproduced from Vogel 1994; redrawn from Soluk and Craig 1990}.

52

Figure 1-6. Diagram of feeding posture adopted by larvae of the river mayfly Ametropus neavi
(Ephemeroptera: Ametropodidae). Arrows indicate the direction of flow (reproduced from Soluk and
Craig 1988).

Maln1tr11m
Flow

General flow

dJrection

Figure 1-7. Diagram of the flow patterns around the body of the black fly larva (Simulium vittatum,
Diptera: Simuliidae). The flow in the diagram is from left to right, and the larva is yawing toward the
viewer. Length of larva= 6.0 mm (reproduced from Merritt et al. 1996).

53

Chapter 2

Environmental influences on macroinvertebrate distribution at
reach and watershed scales

Summary
1.

The

relationship

between

within-watershed

and

within-stream

distributions

of

benthic

invertebrates was investigated in the tributaries of the lower Torpy River watershed in east-central
British Columbia.
2. The relationship between invertebrate community structure and several environmental variables
was analysed using canonical correspondence analysis (CCA), a direct ordination technique.
Aquatic macroinvertebrates were primarily distributed along a physical gradient (related to
streambed substrate composition and discharge) and secondarily along a hydrochemical gradient
(related to conductivity and dissolved oxygen).
3. The CCA ordination divided the sampled invertebrates into four functional assemblages . Each
assemblage had well-defined habitat requirements (erosional vs. depositional) and trophic
relationships (e.g. scraper, collector-gatherer, etc.).
4.

Based on the results of the CCA, the habitat preferences of the heptageniid mayfly Epeorus, a
representative torrential invertebrate, were further determined in two high discharge streams in
the Torpy watershed .

5. Densities of late instar Epeorus larvae were significantly and negatively correlated to velocity (U),
depth (D), channel Reynolds number (Re), and relative roughness (D/kave). Habitat requirements
of this mayfly may be related to food resource (periphyton) availability, predation, ontogenetic
shifts (e.g. pre-emergence behaviour), and/or physiological requirements.

Introduction
One of the primary objectives of ecology is to understand the ways in which environmental
factors influence the distribution of organisms. In streams, the structure of the benthic community
and the distribution of organisms are influenced by water temperature (Hawkins et al. 1997), water

54

chemistry (Pienet et al. 1996; Williams et al. 1997), food resource availability (Hearnden and Pearson
1991; Shannon et al. 1994 ), bed roughness (Growns and Davis 1994), substrate composition (De
March 1976) and an array of biotic interactions (e.g. Hart and Merz 1998).

Stream hydraulics,

however, are generally thought to be the single most important physical factor influencing the
structure of benthic invertebrate communities .

The forces generated by moving water have

substantial effects on the feeding, habitat selection, and inter-specific interactions of benthic fauna
(Peckarsky and Wilcox 1989; Wetmore et al. 1990; Quinn et al. 1996). Hydraulic variables related to
community structure include discharge and discharge stability (Malmqvist and Englund 1996), mean
(Hearnden and Pearson 1991) and near-bed (Growns and Davis 1994) velocity, water surface slope
(Danehy et al. 1999), thickness of the viscous sublayer (Statzner and Higler 1986), and turbulence
intensity (Bouckaert and Davis 1998).
Several authors have noted that changes in faunal assemblages occur from the headwaters
to the mouth of a stream.

The river continuum concept (Vannote et al. 1980) attributes such

longitudinal gradients in macroinvertebrate diversity to system-wide shifts in metabolic, nutrient, and
energetic factors. Subsequently, Statzner and Higler (1986) demonstrated that flow characteristics
(i.e. 'stream hydraulics') are the most important factor controlling the zonation of stream invertebrate

taxa. Growns and Davis (1994) suggested that the apparently contradictory conclusions of Vannote
et al. (1980) and Statzner and Higler (1986) might, in fact, be complementary, as hydrodynamic
factors strongly influence numerous components of the stream environment, and , as a result, have
pervasive effects on ecological processes in streams. Several studies (e.g. Growns and Davis 1994)
have demonstrated that the headwater-to-valley gradient along which macroinvertebrate taxa are
distributed is defined by a number of hydraulic and channel factors, including stream width, water
surface slope, Froude number, Reynolds number, bed roughness, and shear velocity. Hydrodynamic
variables are also related to the distribution of macroinvertebrates between streams and within
watersheds.

Danehy et al. (1999) found that macroinvertebrates were distributed along an

environmental gradient related to stream width and water surface slope, while Robertson et al. ( 1997)
demonstrated that, in addition to hydrochemical variables, near-bed flow characteristics influenced
the community structure of benthic microcrustacea in several streams.
Clear patterns in the microhabitat preferences of individual macroinvertebrate taxa have been
demonstrated by numerous studies . Stream invertebrates respond to small-scale variations in light

55

and UV intensity (Donahue and Schindler 1998), predation (Muotka and Pentinnen 1994), substrate
roughness (Way et al. 1995), and food availability (Vaughn 1986). Near-bed hydraul ic conditions also
influence habitat preferences by determining the levels of lift, drag, and shear stress experienced by
organisms on the streambed (Statzner and Holm 1989; Weissenberger et al. 1991 ). Hydraulic forces
may also influence benthic organisms indirectly, by mediating levels of predation (Hart and Merz,
1998) and the availability of food resources (e.g. periphyton, see Vaughn 1986; Quinn et al. 1996), as
well as ecologically important physical factors such as substrate particle size composition and oxygen
availability (Golubkov et al. 1992; Carling et al. 1998).
Several studies have demonstrated that simple hydraulic parameters (e.g. mean velocity,
depth) are correlated with the distribution of individual taxa (e.g. Collier 1993). Generally, however,
the habitat preferences of stream invertebrates are closely correlated with various combinations of
hydrodynamic parameters (i.e. Froude number, Reynolds number, etc.) that better define the nearbed flow environment. For example, Collier et al. (1995) found that the distribution of larvae of the
hydrobiosid caddisflies Costachorema and Hydrobiosis were positively related to water velocity,
Froude number, boundary layer Reynolds number, and inferred shear velocity. Similarly, Quinn and
Hickey (1994) found that of all hydraulic parameters estimated, boundary Reynolds number was most
strongly correlated with the distributions of several benthic invertebrate taxa in two New Zealand
rivers .
Many stream invertebrates that live in high velocity microhabitats exhibit adaptations to
hydraulic stress. Black fly larvae reduce drag by modifying body posture (Lacoursiere 1992), while
the mayfly Ecdyonurus alters femoral angle in order to counter the typically high lift forces of highvelocity habitats (Weissenberger et al. 1991 ).

Invertebrates that live on the upper surfaces of

streambed stones exposed to high water velocities often exhibit morphological adaptations to
maintain contact with the substrate. Blepharicerid larvae possess a row of ventral suckers (Vogel
1994), while the abdominal gills of several mayfly species (e.g. De/eatidium var. myzobranchia
(Collier 1994), Rhithrogena, and Epeorus (Vogel 1994)) are also arranged to form a ventral "suckerlike" disk. Delgado and Carbonell (1997) found that the cases of the caddisfly Sericostoma se/ysi
collected in high velocity habitats were heavier and more narrow than those from individuals collected
in slower habitats, and interpreted this as an adaptation to counter lift and minimise drag, thus
decreasing the probability of accidental entrainment. Studies by Robertson et al. ( 1995; 1997) and
56

Rempel et al. (1999) suggest that invertebrates lacking morphological adaptations to high velocities
may instead rely on behavioural adaptations, and move into microhabitats characterised by low
hydraulic stresses (i.e. low shear stress areas of the bed (including interstitial spaces), low velocity
reaches , and shore zones).
Many studies have also shown that the habitat preferences of stream invertebrates are
related to large-scale environmental gradients such as substrate characteristics, discharge, mean
velocity, hydrochemistry, and groundwater inputs (e.g. De March 1976, Growns and Davis 1994;
Malmqvist and Englund 1996; Williams et al. 1997).

As many of these factors are related to

geographical or geomorphological features of the watershed, changes to the catchment of a stream
are inevitably reflected in the stream itself. For example, forestry activities adjacent to streams are
known to increase turbidity, increase inputs of coarse particulate organic matter, change temperature
regimes, and alter primary production (Gurtz et al. 1980; Noel et al. 1986; Growns and Davis 1991 ).
The structure of the benthic community changes rapidly in response to logging-induced alterations of
the stream environment (Newbold et al. 1980; Wallace and Gurtz 1986), and changes in the benthic
invertebrate community structure may persist nearly two decades later (Carlson et al. 1990; Stone
and Wallace 1998).
Given these observations, it is surprising that the majority of studies of stream invertebrate
habitat preferences have examined the distribution of organisms at a single spatial scale.
Distributions of biota and related environmental factors are often quantified at watershed, reach, or
even smaller (patch or stone) scales; but rarely is more than one spatial scale considered in the same
study. Remarkably, few studies have attempted to examine how the factors that directly influence the
microdistributions of stream invertebrates at small scales are linked to the distribution of the same
organisms within the stream system or watershed . For example, water velocity has been identified as
a factor that influences the distribution of stream invertebrates within watersheds, stream systems,
reaches, and on the surface of individual stones (Quinn and Hickey 1994; Hart et al. 1996; Danehy et
al. 1999). However, only a few authors have attempted to determine if large-scale, community-level
responses to variation in velocity are closely related to the microhabitat preferences of the constituent
taxa with respect to velocity (but see various works by B. Statzner and others, including Statzner and
Higler (1986) and Statzner and Holm (1989), see also Muotka and Pentinnen (1994)).

57

The goal of the present study is to examine the community-level and taxa-specific responses
of stream macroinvertebrates to hydrochemical, physical, and riparian factors .

A two-part

experimental design was employed to accomplish this . First, canonical correspondence analysis
(CCA) was used to examine community-level responses to a suite of commonly measured
environmental factors (e.g. dissolved oxygen , temperature, bed composition , etc.) in gravel-, cobble-,
and boulder-bed tributaries of a mountain watershed in north-central British Columbia. Second , the
fine-scale

habitat preferences of torrential

mayfly larvae (Epeorus

spp.) (Ephemeroptera :

Heptageniidae) were measured in two high-discharge streams in the study area (the preferred habitat
type of Epeorus larvae, based on the results of the CCA). Epeorus was chosen as a model organism
as it shows watershed-wide preferences that can also be examined at within-stream spatial scales .
The measurement of the habitat characteristics of this torrential mayfly genus at the two spatial
scales (within-watershed and within-stream) allowed for an assessment of the linkages between the
environmental factors that produce distributions of torrential fauna at large and small scales.

Methods and Materials
Watershed study area
The Torpy River watershed is located approximately 90 km east of Prince George, in the
MacGregor Range of east-central British Columbia. The study area consisted of the southern portion
of the watershed, and encompassed the 116 tributaries that flow into the lower section of the Torpy
River. This portion of the watershed is bounded on the west by the confluence of the Torpy and West
Torpy Rivers (54° 02' 12" N, 121 o 25' 51 " W), and on the east by the furthest extent of forestry roads
which provide access to the region (53° 49' 58", 120° 57' 49" W), near the confluence of the Torpy
River with the Fraser River. This portion of the Torpy River valley has been logged continuously
since 1963, and rates of disturbance due to forestry activities have increased in the last decade due
to salvage harvests of conifer stands killed by western hemlock looper (Lambdina fiscellaria
lugubrosa (Hulst) (Lepidoptera : Geometridae).

58

Watershed-scale sampling
During the late summer and fall of 1996, a biological , geomorph ic, and hydrochemical survey
was conducted on all permanent (non-ephemeral) tributaries of the Torpy River within the study area .
At each stream sampling site, several geomorphic, hydrochemical , and riparian factors were
measured, including conductivity, pH , temperature, dissolved oxygen, substrate composition ,
catchment size, riparian vegetation composition, stream surface slope, discharge, and the presence
of logging activity within the catchment.

Conductivity and dissolved oxygen (DO) were measured

using a multimeter (Corning Checkmate 90; Corning, Acton , Maine), and pH was measured using a
Canlab pH meter (model 607; VWR Canlab, Mississauga, Ontario). Temperature was measured
using an alcohol thermometer (+/- 0.2°C}, and stream surface slope (%) was measured using an
inclinometer.
Substrate composition was assessed at each sampling site by visually estimating the areal
coverage of various sizes of substrate particles (using methodology modified from Allan (1995)).
Substrate size was defined by the diameter of the substrate elements (streambed stones): boulders >
300 mm, cobbles 75 - 300 mm, gravels 4 - 75 mm, fines < 4 mm . The bed of each stream sampling
site was categorised as belonging to one of five classes, based on the areal coverage of the
predominant particle size class: (1) boulder/cobble substrate (areal coverage by boulders > 30%), (2)
cobble/gravel substrate (areal coverage by cobbles > 30%), (3) gravel/fines substrate (areal coverage
by gravels > 30%), (4) inorganic/organic fines (areal coverage by fine sediments > 40%), or (5)
organic fines/organic debris (areal coverage by organic debris> 40%).
The riparian vegetation adjacent to each stream was classified as being dominated by one of
the following plant species assemblages and/or functional plant types ; (1) closed coniferous canopy,
dominated by white spruce (Picea glauca), subalpine fir (Abies lasiocarpa), and western hemlock
(Tsuga heterophylla), (2) closed mixed canopy, dominated by conifers in combination with several
deciduous species including Populus spp., paper birch (Betula papyrifera), and alder (Alnus spp.),
and having shade tolerant species such as devil's club (Oplopanax horridus) beneath the canopy, (3)
closed deciduous canopy, (4) open canopy, where riparian vegetation was dominated not by tree
species but rather by various herbaceous plants including thimbleberry (Rubus parviflorus), goat's

59

beard (Aruncus diocius), and various fern species, and (5) wetland areas dominated by grasses and
wetland species such as skunk cabbage (Lysichiton americanum) .
The discharge (Q) of each stream was determined using one of two methods. In larger
3

1

streams (Q < 0.01 m s- , approximately), velocity was measured at 60% of depth across a relatively
uniform stream section at regular intervals. The product of the velocity and area (section width x
depth) of each section were summed to estimate discharge (as per Danehy et al. 1999). In smaller
streams, which flowed under the access road via a culvert rather than a bridge, discharge was
estimated by multiplying the cross-sectional area of the flow through the culvert (calculated from
culvert diameter and maximum water depth) by the average water velocity (measured at 60% of
maximum culvert water depth).
An index of the catchment area size of each stream was obtained by averaging the distances
from each stream to the two nearest streams.

Between-stream distances were measured

perpendicular to the stream, along the forestry access road in the lower Torpy River valley, using the
vehicular odometer (the road paralleled the Torpy River). This variable was included in the ordination
analyses as the factor AVEDIST (AVErage DISTance) as it is an index of the relative width of the
catchment basin of a stream at the sampling point, rather than a true measure of the area of a
stream's catchment.
All sampling sites were located upstream of access roads or other disturbances (a minimum
of 3 m) in order to minimise the influence of local anthropogenic activities on the biotic and abiotic
components of the stream environment.

Watershed-scale macroinvertebrate sampling
Of the 116 streams in the lower Torpy River watershed, 28 streams had beds composed of
gravels, cobbles, and/or boulders, and 88 streams had beds composed of fine organic material and/or
inorganic silt and clay. Invertebrate samples were collected in streams whose beds were composed
of coarse bed material.

Of these 28 streams, the substrates of 15 were comprised of gravel or

gravel/cobble riffles (riffle habitat), and the substrates of 2 were comprised primarily of boulders
(boulder cascade habitat). Three invertebrate samples were obtained in each of these streams (i.e. 3
replicate samples in representative riffle or boulder reaches). The remaining 11 streams had reaches
of both riffle and boulder cascade habitats. In each of these 11 streams, 6 invertebrate samples were
60

obtained (i.e. 3 replicate samples in riffle habitat, 3 replicate samples in boulder cascade habitat). In
total, invertebrates were sampled at 39 benthic sampling sites (i.e. 15 riffle streams (one sampling
site each) + 2 boulder cascade streams (one sampling site each) + 11 streams with diverse
substrates (two sampling sites per stream) = 39 sampling sites). As three invertebrate samples were
collected at each site, a total of 117 invertebrate samples were obtained.
A modified Surber sampler (sampling area = 0.090 m , modified by the addition of a
2

removable sample 'trap') was used to obtain invertebrate samples at all sampling sites . At sampling
sites whose beds were composed primarily of gravel or gravel/cobble riffles, the area within the
sampler frame was disturbed to a uniform depth of 5 em, and any larger stones {diameter> 10 em)
were cleaned by hand to remove any attached invertebrates. At sites where the streambed was
composed almost entirely of boulders and large cobbles, the Surber sampler was placed over a
representative boulder, and the boulder was scrubbed by hand to dislodge attached invertebrates.
The boulder was then removed, and the remaining gravels were disturbed to a uniform depth in order
to collect the remaining invertebrates. Samples were preserved and fixed in the field using 70%
ethanol, sorted under a dissection microscope (2 - 6X), and identified to the lowest practical
taxonomic level, usually family or genus.

Within-stream macroinvertebrate sampling
Over a two-week period in 1997 (July 23 - August 5) 50 benthic samples were obtained from
two adjacent tributaries of the Torpy River using a modified Surber sampler (sampling area = 0.090
m2 ) . The two streams (streams 48 and 56) were hydrochemically similar, but differed with respect to
physical parameters (Table 2-1 ).

As such , the two streams represent a range of physical

microenvironments available to stream macroinvertebrates. In order to prevent bias in the selection
of sampling sites and to ensure that a representative range of benthic microenvironments were
sampled in each stream, Surber samples were taken in the centre of the channel (rather than the
thalweg), regularly (rather than randomly) every 10 m. Macroinvertebrates were preserved, fixed ,
and separated from debris using the methods detailed above. Abundance of the heptageniid mayfly
Epeorus was determined for each sample. The microhabitat preferences of Epeorus were examined

in further detail as this mayfly genus was thought, based on the results of the canonical
correspondence analysis, to represent a typical member of the "torrential fauna" (Vogel 1994 ).

61

Before each sample was collected, several environmental variables were measured at each
sampling site, including mean velocity, near-bed velocity, stream depth, and stream width . Mean
velocity (U) was measured at 60% of total depth using a propeller velocimeter (propeller diameter =
3.0 em; Swoffer current velocity meter model 2100; Swoffer Instruments Inc. Seattle, Washington).
An integrated measurement of near-bed velocity was obtained by deploying the propeller velocimeter
2 em above the bed, the minimum distance from the bed the velocimeter can be deployed without the
propeller striking the surface (nominally, this velocity was measured 2 em above the bed , hence

Ua.o2m). After benthic samples had been obtained, the lengths of the four largest stones contained
within the sampling area were measured .

Table 2-1. Comparison of environmental factors between streams sampled in
microhabitat preference analysis
Hydrochemical Factors
Conductivity (!lS)

PH

Stream 48 a

Stream 56 a

218 .0
8.30

247.0
7.75

Physical Factors
3
1
Discharge (m s" )
0.191
0.332
Average Surface Slope (%)
4.1
3.3
0.50
0.95
Average Distance (km) b
a stream 48 (km 22.0 Lower Torpy Road), stream 56 (km 25.8 Lower Torpy Road)
b average of distance to the two adjacent streams (an indicator of catchment size)

For each sampling site, several substrate and hydraulic parameters were calculated,
including Froude number (Fr), channel Reynolds number (Re), average stone length (Lave; calculated
as the mean of the longest-axis diameters of the four largest stones), standard deviation of stone
length (L50 ), and relative roughness (D/kave). Channel Reynolds number and Froude number are
determined as

Re =Ul/u

(2-1)

and

Fr = U/(gD)

05

(2-2)
62

respectively, where U is velocity, I is the characteristic length scale (in this case, channel depth), u is
2

the kinematic viscosity of water, g is gravitational acceleration (9 .81 m s" ), and D is channel depth.
See appendix 1 for notation and formulae .

Data Analysis
Macroinvertebrate community structure was related to the measured environmental variables
using canonical correspondence analysis (CCA). CCA is a gradient analysis technique that directly
relates species abundances to environmental gradients (ter Braak 1986); such forms of ordination are
appropriate to summarise the relationship of taxa to environmental variables when species show
unimodal response curves to environmental gradients (see Palmer 1993; ter Braak 1986). CCA
ordination analysis of the 117 benthic samples was conducted using CANOCO version 3.2 (ter Braak
1991 ).
In CCA, the weighted average indicates the 'centre' of the species distribution along a given
environmental variable (ter Braak 1986). The difference in the weighted average of several species
indicates the differences in the distribution of those species along the environmental variable.
Species that place highest along an environmental variable ('vector') will have the highest weighted
average with respect to that environmental variable . The significance of the extracted CCA axes was
tested by an unrestricted Monte Carlo simulation (999 permutations).
In order to minimise the influence of 'rare' taxa on the ordination, several taxa were excluded
from the analyses (Gauch 1982). 'Rare' taxa were identified as those taxa whose 95% confidence
interval of the pooled abundances included zero. This method removes taxa with relatively few
occurrences; it is, however, biased towards removing those taxa that are relatively abundant at the
few sites where they occur. The 24 remaining taxa were included in the analysis, and were given
equal weighting. Throughout the analysis outliers were not removed ; all samples were examined ,
and thought to represent biologically valid observations .
Two transformations of the taxa abundance and environmental data were undertaken. (1) As
the distribution of individuals of a species are often highly contagious and have highly variable
absolute abundances (see Williams et al. 1997), taxa abundances were ln{x+1) transformed, and

63

related to environmental variables using canonical correspondence analysis (CCA), as described by
ter Braak (1986). (2) Since all environmental variables were measured on different scales , they were
normalised prior to analysis to a mean of zero and a variance of one (ter Braak 1986; ter Braak
1995).
A preliminary principal components analysis (PCA; using Statistica 5.1, Statsoft, Oklahoma)
was undertaken to assess whether there was collinearity between environmental variables . The PCA
showed that there was a high degree of collinearity between the geographic variable KM (distance
along the long axis of the watershed) and the logging (LOGPROX) and defoliation (DEFOL) variables.
In order to account for the variation in invertebrate community structure due to either naturally
occurring or anthropogenically induced geographical variation, and to account for seasonal variation
in macroinvertebrate community structure (order of stream sampling along the watershed axis was
sequential rather than random), KM was included as a covariate in the CCA ordination.
To ensure that analysis of the macroinvertebrate community incorporates the distinct species
assemblages that exist at small spatial scales, macroinvertebrate abundances at each sampling site
(rather than stream averages) were included in the CCA ordination.

Univariate analyses were

conducted on 'reach-averaged' rather than 'stream-averaged' data (i.e. analyses conducted on
stream sampling site data). This approach was adopted to ensure that habitat-specific responses in
community and taxa-specific variables were not lost when riffle and boulder-cascade were combined
to produce an average value.
2

2

Coefficients of determination (R ) of regression analyses are reported as percentages (R x

100).

Results
Distribution within the watershed
In total, ten environmental variables were measured and included in the CCA ordination
analysis. A summary of the major variables is presented in Table 2-2 .
Macroinvertebrates were present in all 117 benthic samples collected at the 39 stream
sampling sites. A total of 55 macroinvertebrate taxa were collected, with the taxonomic richness of
the sampling sites ranging from 1 to 16 taxa. No single macroinvertebrate taxon was collected at

64

every sampling site. Several genera were, however, relatively widespread, including the stonefly

Zapada and the predatory caddisfly Rhyacophila, which were found at 94.9% and 92.3% of the sites
sampled , respectively.

Several mayfly genera, including Drunella, Rhithrogena, and Baetis, were

also relatively common in the study area. Taxonomic richness tended to decrease with increasing
discharge (ANOVA, F(1,115) = 11.67, p < 0.001) (Figure 2-1}.

However, total invertebrate

abundance was not related to discharge. This suggests that the benthic invertebrate community in
larger {i.e. higher discharge) streams was dominated by fewer species. Stream hydrochemistry was
also related to discharge. While the conductivity of low discharge streams was variable, the
conductivity of higher order streams was relatively uniform at about 250 !lS (Figure 2-2) .

Table 2-2. Major environmental characteristics of the tributaries of the lower
Torpy River watershed
Benthic Sample Sites a
(n=39)
Mean (SO)
Ranqe
Environmental variables
1.51 (0.75)
1-3
Substrate composition °
3 1
oe- 0.63
0.01-0.63
Discharge (m s- }
0.070 (0.15)
0.20 (0.20)
oe -1.20
Average distance c (km)
0.59 (0.31)
0.10-1 .20
0.41 (0.15}
6.30-8.40
8.13 (0.24)
7.35-8.40
7.76 (0.42)
pH
50- 100
93.23 (5.35)
69-100
Dissolved Oxygen(%)
86 .34 (9.19)
Riparian Vegetation b
2.23 (0.53}
2-4
oe- 46 .50
1.2-19.0
Stream Surface Slope (%)
5.46 (5.38)
6.02 (3.86)
241 .8 (82 .1)
30.6-404
258.0 (39.5)
195-400
Conductivity (!lS)
Riparian Defoliation d
0.28 (0.45)
0-1
Logging d
0.28 (0.45)
0-1
a benthic macromvertebrates samples were obtained at all stream s1tes which had gravel, gravelAll Streams in the lower Torpy
River watershed (n=116)
Ranqe
Mean (SO)

-

-

-

-

-

cobble, or boulder substrates.

ban index (see text for description)
c average distance to the two adjacent streams, measured perpendicular to the stream channel
d dichotomous variables
e approximately zero - unable to measure accurately using methods employed
- not calculated for sites where invertebrate samples not obtained

Taxonomic richness was related to degree of stream catchment disturbance (ANOVA,
F(2, 114) = 11.24, p = 0.023) (see also Table 2-3). Streams with relatively recent logging activity
(within the decade preceding the study; mean richness per sample = 6. 79 taxa, SE = 0.51) had
significantly fewer taxa than streams with unlogged catchments (mean richness = 8.92 taxa, SE =
0.53) {Tukey's HSD, p = 0.020). The taxonomic richness of streams with less recent logging activities
65

(10- 36 years before the study was conducted; mean richness = 8.79 taxa, SE = 0.48) was not
significantly different than streams with either recently logged (Tukey's HSD, p = 0.68) or unlogged
(Tukey's HSD, p = 0.37) areas .

CCA ordination
Canonical correspondence analysis ordination of the 117 benthic samples distinguished the
influence of chemical (pH, conductivity, dissolved oxygen), physical, (stream surface slope,
discharge, substrate size, and average distance to adjacent tributaries), and riparian (riparian
vegetation

and

defoliation

indices,

and

logging

activities

within

catchment)

factors

on

macroinvertebrate community composition. CCA ordination separated the stream sampling sites first
by physical variables related to stream size and then by hydrochemical factors (Table 2-3). Axes I
and II explained 59.5% of the total variance in the 'species-environmental variables' relationship (see
Table 2-4 for a summary). The Monte Carlo test showed that axis I was significantly related to the
species data (p = 0.01 ). CCA axis I was most strongly correlated with substrate composition (r1 =
0.8696) and discharge (r1 = -0 .7515), while CCA axis II was most strongly correlated to conductivity (r11
= -0.4783) and dissolved oxygen (r11 = 0.4257); in addition, pH had a high intraset correlation with axis
II (r11 = 0.3898).

The weighted averages of each macroinvertebrate taxa on the measured

environmental variables are displayed in a biplot (Figure 2-3), split for clarity into taxonomic and
environmental variable components.
The CCA biplot (Figure 2-3) and axis loadings (Table 2-3) indicate the existence of four
functional assemblages of stream invertebrates (Table 2-5). The first assemblage (AS1) includes the
taxa in the upper left quadrant of the biplot, and includes the trichopterans Agapteus, Lepidostoma,
and Arctopsyche, as well as the stoneflies Kogotus and Zapada. AS1 taxa were generally erosional
zone clingers, and include several shredder-detritivores (see Table 2-5). The preferred habitats of
these genera include high discharge streams with high surface slopes, coarse substrates, and low
conductivity.

The riparian areas of streams whose benthic communities were dominated by AS1

species were typically closed coniferous and deciduous canopies. The second assemblage (AS2)
includes taxa located in the upper right quadrant of the biplot, and includes the stonefly Visoka, the
caddisfly Rhyacophila, the mayfly Yoraperla, larval coleoptera of the fam ily Elmidae, and dipteran
larvae of the families Psychodidae, Empididae, and Chironomidae.
66

Genera included in the AS2

assemblage were trophically diverse with leaf/litter associations (collector-gatherers and shredders).
The AS2 assemblage included both erosional and depositional zone taxa that are categorised as
clingers and burrowers (see Table 2-5).

AS2 taxa were found at sites with low pH and fine

substrates, but were poorly associated with discharge and riparian variables .

Taxa of the third

functional assemblage (AS3) are found in the lower right quadrant of the biplot, and are separated
from AS2 taxa primarily on the basis of stream surface slope and conductivity. The AS3 assemblage
includes the mayfly Cinygmu/a, larvae of the dipteran families Tabanidae, and Tipulidae, the Uenoid
trichopteran Neothremma, and Oligochaeta . AS3 taxa were generally depositional zone burrowers,
and were largely collector-gatherers (see Table 2-5).

Benthic habitats dominated by the AS3

assemblage included small, closely spaced streams with high conductivity, low slopes, and fine
substrates.

Riparian vegetation at AS3 sites tended to consist of open canopies, grasses, and

wetland plant species . The fourth functional assemblage (AS4) includes the mayfly genera Epeorus,
Rhithrogena, Drunel/a, and Baetis, as well as the caddisfly Glossosoma and the stoneflies Megarcys

and Chloroperlidae.

AS4 taxa were generally clingers and scrapers associated with erosional

habitats (see Table 2-5). Sites preferred by these taxa had typically high discharge, large substrates,
and high pH .
The Baetid mayfly Baetis had high, negative loading on both axes I and II .

This

demonstrates a strong association with sites characterised by large substrates (cobbles and
boulders). Baetis also has a high weighted average with respect to pH, and a low weighted average
with respect to slope. Baetis abundance was only weakly associated with discharge, suggesting that
stream size is not a strong habitat determinant for this genus. The weighted averages of Tipulidae
and the caddisfly Neothremma (Uenoidae) demonstrate a strong preference of these two taxa for
very small streams with low slope, high conductivity, and open , wetland-type riparian vegetation.
Abundance of the heptageniid mayfly Epeorus in tributaries of the Torpy River was more
closely related to the physical stream gradient (axis I} than the hydrochemical gradient (axis II} (Table
2-6). This indicates that the habitat preferences of Epeorus include high discharge streams with
coarse substrates.

67

Table 2-3. lntraset correlation coefficients of environmental variables with the
first two axes of the Canonical Correspondence Analysis (CCA) a
lntraset correlation coefficients
Label
Axis I
Axis II
Environmental variables c
Substrate composition
SUBSTR
86 .96
7.45
Discharge
Q
-75.15
33 .53
AVEDIST
-70.77
39.36
Averaged distance
pH
pH
-62.28
-38.98
Dissolved Oxygen
DO
-49.15
-42.57
Riparian Vegetation
RIPARIAN
-38.48
-27.55
31 .09
Stream Surface Slope
SLOPE
22.20
Conductivity
COND
9.81
-47.83
Riparian Defoliation
DEFOL
-5.94
27.52
Logging
LOGPROX
3.33
-13.11
a environmental variables standardised to unit mean and variance (species data ln(x+1) transformed).
b intraset correlation coefficients = 100 x r
c variables ordered with respect to loadings on axis I.

Table 2-4. Summary of Canonical Correlation Analysis (CCA) of lower Torpy
River watershed data a
II
CCA axis
Eigenvalues
0.258
0.099
0.878
0.729
Species-environment correlation
Cumulative % variation explained
Species data
12.1
16.8
Species- env. relationship
43.0
59.5
a geographical variable KM entered as covariate in CCA
* Monte Carlo test of significance of extracted axes: overall test F
0.60

Ill
0.076
0.685

:E of Eigenvalues
2.221*

20.4
72 .2

=4.13, p =0.01 , Trace Statistic =

Distribution within the stream
The ordination analysis (CCA) demonstrated that hydraulic and substrate factors appeared to
be relatively important in determining the habitat preferences of the torrential mayfly Epeorus.
Subsequently, the densities of Epeorus (ln(x+1) transformed densities) at 50 sites in two adjacent
high-discharge streams (stream 48, n =25; stream 56, n =25) were measured and related to a series
of hydraulic and substrate variables using single and multiple regression analyses (see Table 2-7). In
independent regression analyses, channel depth (D) and mean water velocity (U) each accounted for
2

a significant amount of the variation in Epeorus abundance (R = 16.73% and R2 = 25.36%,
respectively; see Figures 2-4 and 2-5). Depth and velocity accounted for 33.20% of the variation in
Epeorus abundance in multiple regression analysis, while channel Reynolds number (Re) accounted

68

for 16.83% (see Figure 2-6). Relative roughness (Dikave) was also significantly related to Epeorus
abundance (Figure 2-7).

Near-bed velocity (Uo.o2m). average stone length (Lave). maximum stone

length (Lmax). stone length variation (L50 ), and Froude number (Fr) did not explain significant amounts
of the variation in mayfly abundance.

Table 2-5. Summary of ecological data a for common macroinvertebrate taxa
collected in tributaries of the lower Torpy River b
Assemblag_e I Taxa
AS1
Agapteus
Arctopsyche
Kogotus
Zapada
Lepidostoma

Habitat

Habit

Troe_hic Relationship_

Erosional
Erosional
Erosional
Erosional
Erosional - Depositional

Clingers
Clingers
Clingers
Sprawlers I clingers
Climbers - Sprawlers
-Clingers

Sc, C-G
C-F
Pr
Sh-D
Sh-D

AS2
Visoka
NA
NA
NA
Yoraperla
NA
Sh- D, Sc
NA
Rhyacophila
Erosional
Clingers
Pr, C- G, Sh- H
Elmidae
Erosional
Clingers
C, G, Sc
C-G
Psychodidae
Depositional
Burrowers
Chironomidae
Erosional - Depositional
Burrowers
C- G, F, Pr
AS3
Depositional
Sprawlers I burrowers
Pr
Tabanidae
Sc, C-G
Cinygmula
Erosional
Clingers
Oligochaeta c
Burrowers
Depositional
NA
Erosional - Depositional
Sprawlers I burrowers
Pr (some C- G)
Empididae
Sh-D , C-G
Tipulidae
Erosional - Depositional
Burrowers
Neothremma
Erosional
Clingers
Sc, C-G
AS4
Baetis
Erosional- Depositional
Climbers I clingers
C- G, Sc
Chloroperlidae
Erosional
Clingers
Pr, Sc, C - G
Megarcys
Erosional
Clingers
Pr
Drunella
NA
Clingers I sprawlers
Sc, Pr
G/ossosoma
Erosional
Clingers
Sc
Rhithrogena
Erosional
Clingers
C - G, Sc
Ee_eorus
Erosional
Clingers
C - G, Sc
a ecological information from Merritt and Cummins (1996), except where otherwise noted
b see Figure 2-3 for ordination diagram
c ecological information from Thorp and Covich (1991)
C
collector, D
detritivore, F
filterer I suspension-feeder, G
gatherer, H
herbivore, Pr
predator, Sc scraper, Sh shredder, NA information not available

=

=

=

=

=

=

=

=

=

In streams 48 and 56, densities of Epeorus were generally highest in shallow flows (D < 0.20
1

m), at low velocities (U < 30 em s- ) , and at relatively low channel Reynolds numbers (Re < 100,000).
Epeorus densities appeared to decrease with increases in all three hydraulic factors (Figures 2-4, 2-5,
and 2-6). There was, however, a weakly significant relationship between channel depth and velocity
69

(R 2=7.73%) (Figure 2-8). Regression analysis revealed a strong relationship between mean velocity

(U) and near-bed velocity (Uo.o 2m) (R2 = 45.47%) (Figure 2-9). In addition , densities of Epeorus were
significantly higher in stream 48 than stream 56 (t = -5.47, p < 0.001 ).

Table 2-6. Cumulative fit of Epeorus (Heptageniidae) abundance data against
environmental axes (canonical correspondence analysis a)
Axis I
19.13%

Cumulative fit
(as percent of total variance)
a for summary see Table 2-4

Table 2-7.

Axis II
21 .80%

Total variance explained
31.38%

Simple linear and standard multiple regressions of Epeorus

microdistribution a and various hydraulic and substrate characteristics
R"
Parameter (simple regress.)
16.73%
Depth (D)
25.36%
Mean velocity (U)
4.74%
Near-bed velocity (Uo.o2m)
Maximum stone length (Lmax) 2.34%
5.31%
Average stone length (Lave)
0.080%
Stone length SO (Lso)
23.17%
Relative roughness (0/kave)
Froude number (Fr)
3.63%
Reynolds number (Re)
31 .19%
Parameter (mutt. regress.)
Depth + Mean U
33.20%
a Epeorus densities ln(x+1) transformed

ANOVA(df) = F
(1 ,48) = 9.64
(1,48) = 16.31
(1 ,48) = 2.39
(1,48)=1 .15
(1 ,48) = 2.69
(1 ,48) = 0.040
(1 ,48) = 14.47
(1 ,48) = 1.81
(1 ,48) = 21.80

p (slope)

-8.29
-4.42
-2.63
4.85
13.70
2.30
-0.56
-2.21
-0.000013

p-value
0.0032
< 0.001
0.13
0.29
0.11
0.84
< 0.001
0.19
< 0.001

(2,47) = 11.68

-5.91' -3.71

0.023, 0.0014

Discussion
The

measured

environmental factors

were

important

in

determining

the

benthic

macroinvertebrate community structure in the 39 streams examined in the lower Torpy River
watershed.

Once geographical variation was accounted for, hydraulic and substrate variables

appeared to be of primary importance in determining the macroinvertebrate species assemblages ,
while hydrochemical

factors

were

of secondary

importance.

CCA ordination

separated

macroinvertebrate assemblages first by substrate composition, discharge, and catchment size, and
second by conductivity and dissolved oxygen (see Tables 2-3 and 2-4). While several of these
parameters (e.g. discharge, catchment size) may not be exactly those experienced by benthic
70

invertebrates, they do reflect the overall habitat structure and environmental gradients to which the
benthic community responds .

This conclusion is supported by similar findings elsewhere (e.g.

Malmqvist and Englund 1996; Danehy et al. 1999).
Rather than measuring macroinvertebrate diversity and abundance at the stream scale (i.e.
sampling large areas of the stream bed or obtaining mean values of species richness and abundance
by averaging several replicated samples), ordination analyses were conducted at the 'patch' scale

(i.e. the area encompassed within the frame of the Surber sampler). Although several studies have
shown that even relatively small areas may encompass several substantially different flow
microhabitats (e.g. Hart et al. 1996; Robert et al. 1996), the size of the 'patch' (Surber sampler area)
2

utilised in the present study was small enough (approximately 0.09 m ) that the benthic community
enclosed therein would be exposed to a similar range of environmental factors . This 'patch-scale'
analysis was used because the structure of the benthic community, rather than invertebrate
abundance or richness , was the objective of the ordination analysis . This patch size is large enough
to include taxa commonly found within the reach, yet small enough that invertebrates collected within
a single sample should belong to a characteristic assemblage that is influenced by a common set of
environmental and biotic factors .

Taxonomic richness
The results indicate that taxonomic richness generally decreases with increasing discharge
within the range measured . This result conflicts with the observations of Growns and Davis (1994)
and Statzner and Higler (1986) who reported that taxonomic richness increased with increasing
discharge .

However, Figure 2-1 demonstrates that the relationship between benthic community

composition and discharge is complex, and may, in part, be dependent on spatial scale (i.e. small3

scale habitat characteristics). Among small streams (Q < 0.1 m s·\ taxonomic richness is highly
variable.

3

As discharge increases (Q > 0.1 m s·\ the variation in taxonomic richness between

streams decreases substantially, and fewer invertebrate taxa are present.

This suggests that

discharge and discharge-related parameters (including substrate compos ition, hydraulic variables,
channel morphology, etc.) may have the strongest influence on community composition in large, highorder streams.

The substantial variation in taxonomic richness among low discharge, low-order

streams suggests that as stream size decreases, factors other than discharge are more important in

71

structuring the benthic invertebrate community. Although an inverse relationship between discharge
and taxonomic richness was unexpected, there are several possible explanations to account for these
results.
First, small-scale habitat complexity or chemical factors may be more important in influencing
the structure of the benthic community in small streams than in large streams .

For example,
3

1

conductivity, which is relatively consistent (- 250 J.LS) at discharges greater than 0.1 m s- , is highly
variable in streams with low discharge.

Similarly, Growns and Davis (1994) found that while

conductivity is variable in small streams, it is more consistent (about 250 J.LS) in streams of greater
discharge.

As such, if invertebrate diversity were strongly influenced by hydrochemical factors,

taxonomic richness would be expected to be highly variable in low discharge streams, and more
uniform in larger streams.

This explanation agrees with the observed pattern .

The relative

importance of conductivity is supported by the results of the ordination analysis; intraset correlation
coefficients of the CCA (see Table 4-3) reveal that conductivity is the most important hydrochemical
factor influencing community composition . Similarly, Williams et al. (1997) found that the invertebrate
community structure in eastern Canadian springs was dependent on water chemistry, especially on
ionic composition and concentration. However, while both groundwater and surface water inputs may
play an important role in determining invertebrate community structure, their individual effects are
difficult to discern (Jones and Holmes 1996).
There may also be a shift in the importance and nature of small-scale hydraulic factors such
as boundary layer thickness and substrate-flow interactions (Carling 1992; Young 1992; Hart et al.
1996) as streams increase in size (see Newbury 1984). However, the present study measured only a
single, large-scale hydraulic factor (discharge) in each stream within the watershed-scale study area.
The influence of small-scale hydraulic factors on the community structure in first-order streams is
poorly understood, and requires further study.
In addition, inter-stream variation in streambed disturbance regime could, in part, account for
the observed pattern of decreased taxonomic richness with increased discharge. Surface runoff-fed
streams tend to be 'flashy' and subject to frequent bed disturbance, while spring-fed streams are
more stable and, as a result, subject only to infrequent bed disturbance.

Input sources of small

streams in the Torpy watershed are diverse, and include snowmelt, surface runoff, and groundwater
(springs) (personal observation). As such, disturbance intensity and frequency may be substantially
72

more variable between first-order streams than between higher-order streams, wh ich have multiple
inputs from diverse sources .

Townsend et al. (1997), applying the intermediate disturbance

hypothesis, found that species richness was low both in streams with frequent, intense bed
disturbance (where relatively few, rapidly colonising species dominate the assemblage) and very
infrequent bed disturbance (where relatively few, competitively superior species dominate the
assemblage). As such, inter-stream variation in invertebrate diversity among small streams within the
study area may be due, in part, to variable disturbance regimes , with streams with the highest
taxonomic richness having intermediate frequencies of bed disturbance.

Similarly, higher order

streams with greater discharge could have increasingly stable beds, which leads to a community
dominated by fewer, competitively superior invertebrate species.
The pattern of decreasing invertebrate taxa richness with increasing discharge may also be
the result of a community-level response to an unmeasured physical factor related to discharge. For
example, mean channel velocity, average substrate particle size, and channel width increase with
discharge (Newbury 1984; Hubert and Kozel 1993). This corroborates the observation of Townsend
et al. (1997) who noted that richness increased with the percentage of the bed made up of small
particles (8- 32 mm in diameter). While Townsend et al. (1997) attributed this trend to a decrease in
bed stability, it seems likely that it was due, at least in part, to the interaction between near-bed flow
and bed roughness . However, the results of the present study and those of Townsend et al. (1997)
conflict with the results of several studies relating channel characteristics to community structure .
While the present study found that faunal richness was greatest in smaller streams, Statzner and
Higler ( 1986) and Growns and Davis ( 1994) found that richness was greatest in the mid-reaches of
streams. Both studies suggest that the highest species richness should occur in the areas of the
highest hydraulic variation, where several species assemblages overlap due to habitat heterogeneity.
This result also contradicts, at least superficially, the river continuum concept proposed by Vannote et
al. (1980), which states that species richness is highest in mid-order reaches where environmental
variation in greatest. It is important to note, however, that faunal richness in the present study was
measured at the patch scale. In very small streams , a single patch (0.090 m2 , as defined by this
study) can encompass a relatively large proportion of the streambed, and as such , may contain a
number of microhabitat types (e.g. near-shore and thalweg zones, or high-velocity and low-velocity
zones). In larger streams, where stream width was far greater than the width of the sampler, each
73

patch enclosed fewer microhabitat types (i.e. relatively uniform benthic habitat was sampled within
the quadrat represented by the Surber sampler frame) . As such, it is important to limit interpretation
of the results to the patch scale. It is likely that total invertebrate diversity is greater in high discharge
streams due to greater habitat complexity, more spatially diverse interstitial spaces, and greater
hydraulic variability (Growns and Davis 1994; Statzner and Higler 1986; Townsend et al. 1997). If
this assessment (greater habitat variation per patch in smaller streams) is accurate, then the
observed pattern of decreasing richness with increasing discharge may actually support the river
continuum concept, even though the taxonomic richness per reach may actually be greater. More
research is needed to determine if the relationship between species richness and stream order is
dependent on the spatial scale at which invertebrate fauna are sampled .
Forestry activities in the watershed also appear to have an impact on the benthic invertebrate
community. Streams whose catchments were disturbed by relatively recent logging activities had
significantly fewer taxa than streams with undisturbed catchments . Logging activities are known to
affect stream invertebrate communities; Newbold et al. (1980) found that macroinvertebrate diversity
was lower in streams without riparian buffer strips than in those whose banks were protected with
buffer strips. The decreased taxonomic richness in streams with adjacent logging may be the result
of a shift in community structure . Stone and Wallace (1998) found that clearcutting reduced the
proportion of scraper taxa in a mountain stream and increased the proportion of shredders. The
proximity of the conductivity vector to the logging vector on the ordination diagram (Figure 2-3)
suggests that logging activities increase water conductivity. This concurs with the findings of Growns
and Davis (1991) who found that logging within the catchment of a stream significantly increased
streamwater conductivity.

They further suggested that increased conductivity was, in part,

responsible for shifts in invertebrate community structure. In the present study, however, the weak
relationship of the logging variable with either environmental axis suggests that the influence of
forestry activities on invertebrate community composition in the Torpy watershed is limited.

Community structure- CCA ordination
The underlying physical gradient represented by the first CCA axis is associated with physical
stream factors ; large discharge streams with coarse substrates lie on the left of the ordination
diagram, while small streams with beds composed of relatively fine particles lie on the right. The
74

eigenvalues and intraset correlation coefficients suggest that this environmental gradient has the
greatest influence on the structure of the macroinvertebrate communities in tributaries of the Torpy
River watershed . Ordination analyses of benthic commun ity structure by several other authors have
also found that a physical gradient dominates benthic community structure.

Danehy et al. (1999)

found that the first axis was most strongly correlated with water surface slope and mean channel
width, and attributed this to the dominant effect of headwater-to-valley environmental gradients on the
macroinvertebrate community structure. Marchant et al. (1985) and Rossaro and Pietrangelo (1993),
in separate studies of large-scale variation in benthic communities, both found that the underlying
environmental gradient represented by the first axis was related to substrate particle size.
CCA ordination of the benthic community divided the common invertebrate fauna into four
functional assemblages, each associated with a distinct habitat type defined by physical and
hydrochemical factors .

Notably, the invertebrates of both the AS1 and AS4 assemblages were

positively correlated with stream hydraulics (CCA axis 1), and are categorised as clingers inhabiting
erosional habitats (Merritt and Cummins 1996). However, the taxa included in assemblage AS1 are
generally dependent on allochthonous inputs as a food resource (two shredder-detritivore genera and
a collector-gatherer), while the taxa of AS4 are almost exclusively scrapers, dependent on
autochthonous production (periphyton) as a food source. This suggests that AS4, which includes the
heptageniid mayfly Epeorus, may represent a functional assemblage that inhabits the exposed upper
surface of streambed stones, and is found in bouldery, high velocity habitats {i.e. a boulder-torrent
guild). Likewise, AS1 may represent an assemblage (i.e. a cobble/gravel riffle guild) that inhabits
riffle habitats of high-discharge streams.
The positive relationship between discharge and the suspension-feeding hydropsychid
caddisfly Arctopsyche (AS1) supports earlier observations that the local abundances of hydropsychid
and simuliid suspension-feeding taxa are related to high water velocities (Osborne and Herricks 1987;
Growns and Davis 1994). However, while Arctopsyche was categorised as a flow obligate by Growns
and Davis, it was not grouped with the other flow obligate taxa (e.g. Epeorus, Rhithrogena, etc.) in the
ordination diagram (Figure 2-3). This suggests that other variables, such as channel slope, may be
important in the microhabitat preferences of this genus. Hubert and Kozel (1993) reported that water
surface slope was significantly and positively related to environmental factors that have been found to

75

be related to the habitat preferences of suspension-feeding taxa, such as the proportion of the bed
composed of boulders (see Wetmore et al. 1990; Voelz and Ward 1996).
Invertebrates included in the third functional assemblage (AS3) were negatively correlated
with stream hydraulics.

This functional assemblage is composed largely of collector-gatherer,

burrower taxa. Hearnden and Pearson (1991) found that benthic macroinvertebrates with trophic
relationships of this type were strongly associated with detrital deposits in pools with coarse
substrates . Leaf litter deposits and current intensity are strongly related (Minshall 1984), and Growns
and Davis (1994) suggest that detrital material (FPOM and CPOM), the food resource of collectorshredder taxa, is unlikely to accumulate in areas of the stream bed with high current velocities.
Wiggins (1996) classified most AS3 taxa (including psychodid , tabanid, tipulid, and empidid dipterans,
as well as oligochaetes) as being depositional zone taxa. Several AS3 taxa have distinct adaptations
for burrowing rather than movement on the exposed surface of the streambed, suggesting that these
taxa belong to the flow avoider functional group. This supports the findings of Growns and Davis
(1994) who found that flow avoiders are generally shredders. Taxa lying on the right side of the
ordination diagram could be expected to be associated with depositional rather than erosional
habitats, where low currents enhance the deposition of both CPOM and FPOM (Hearnden and
Pearson 1991 ).
Interestingly, the clinger mayfly Cinygmula was included in assemblage AS3, even though it
is generally classified as an erosional zone genus (see Edmunds and Waltz 1996). This suggests
that Cinygmula may not be as dependent on high velocity microhabitats as other erosional zone
species. This result supports the findings of Golubkhov et al. (1992), who observed that Cinygmula
grandifola was not as sensitive to the decreases in oxygen uptake associated with low flow velocities

as other rheophilous species.
Four of the six taxa included in the fourth functional assemblage group (AS4) have distinct
adaptations that allow them to live in high velocity habitats and minimise the probability of being
detached from the substrate . The heptageniid mayflies Epeorus and Rhithrogena have an abdominal
gill arrangement that forms a sucker-shaped ventral disk, which is thought to function as a "suction
cup" which prevents accidental detachment (Vogel 1994 ).

The hairy ventral disk of Drunella is

thought to function in similar fashion . The grazing caddisfly Glossosoma cements its case to large
streambed stones, preventing it from being eroded from the bed . All four taxa would be categorised

76

as flow obligates, based on the criteria of Growns and Davis (1994) who placed stream invertebrates
into three flow exposure groups (obligates, facultatives, and avoiders) based on behavioural and
morphological adaptations to flow.

These four taxa are only weakly associated with axis II,

suggesting that the habitat preferences of flow obligates are determined almost entirely by physical
(flow and substrate characteristics) rather than hydrochemical (e.g. conductivity) factors. AS4 taxa
predictably show strong associations with coarse substrates.

Erosional habitat grazers should be

strongly associated with boulder and cobble substrates, as larger stones provide a relatively stable
substrate for the growth of periphyton; smaller bed particles tend to be eroded and entrained at lower
velocities (Minshall1984).
Comparing the positions of the macroinvertebrate taxa within the ordination diagram to the
known trophic relationships of the taxa demonstrates the ecological validity of CCA ordination.
Organisms with similar habitat requirements and ecological roles have strong associations in Figure
2-3. There are invertebrate predators associated with each of the four functional assemblages,
suggesting some degree of resource partitioning among common predatory taxa.

The predatory

caddisfly Rhyacophila, one of the most ubiquitous invertebrate taxa in the Torpy River watershed, had
a very weak weighted average with respect to axis II, demonstrating that it may be relatively
insensitive to variations in water chemistry. The mayfly Rhithrogena, another commonly occurring
benthic insect, was also very weakly associated with axis II.

This suggests that while common

invertebrate taxa may demonstrate preferences to certain ranges of physical habitat types, they may
be relatively tolerant of a wide range of hydrochemical conditions . Additionally, the ordination sheds
light on the ecological status of several taxa whose habitat preferences are poorly understood. For
instance, the mayfly Drunefla, included in AS4 and shown to be strongly associated with axis I, is
likely a erosional zone species (along with several other AS4 taxa including Epeorus and
Rhithrogena) .

The microdistribution of Epeorus
Large-scale environmental factors (e.g. discharge, substrate composition, disturbance
regime) determine the availability of niches within stream systems, and thus determine if certain taxa
will be present. However, small-scale factors are of greater importance in determining the distribution
of invertebrates within each reach or riffle.

Increasingly, stream hydraulics are thought to be the

77

single most important factor determining the distribution of benthic organisms (see Davis and
Barmuta 1986; Carling 1992; Hart et al. 1996}. So, while CCA ordination demonstrates that the
mayfly Epeorus was associated with physical variables related to stream hydraulics (discharge,
substrate composition), the spatial scale was too coarse to clearly define the microhabitat
preferences of this genus.

However, correlation of several physical stream factors with the

distribution of Epeorus within two adjacent streams sheds light on the relationship between
watershed-scale and small-scale patterns of distribution of benthic invertebrates. Neither simple
parameters (U, U0.o2 m. D, various substrate factors) nor ratios of parameters (Re, Fr, D/kave) provided
the strongest correlations with Epeorus abundance (Table 2-7).

Moreover, while a multiple

regression model incorporating both velocity and depth explained 33.2% of the variation in Epeorus
density, interpretation may be difficult. Larval Epeorus preferred a distinct range of depth and velocity
1

conditions, namely depths less than 20 em, mean velocities less than 30 em s- , channel Reynolds
number less than 100,000, and relative roughness (D/kave) less than 3.

In other words, larvae

preferred shallow, slow flows over rough beds.
Local densities of Epeorus were significantly and negatively related to mean velocity and
channel depth. This result was unexpected, as the distinct morphological adaptations of Epeorus, a
known rheophilous, torrential macroinvertebrate, suggest that larvae should display a marked
preference for high velocity flows . Collier (1994) found that De/eatidium var. myzobranchia, which has
large overlapping gills in a sucker-like arrangement similar to that of Epeorus, was restricted to
velocities greater than 0.90 m s- 1 .

Collier speculated that the morphological adaptations of

Deleatidium allowed it to withstand high velocities and thus access areas of higher quality food

resources . Weissenberger et al. (1991) note that Epeorus, though subject to substantial lift forces at
high velocities (due to its 'airfoil' shape), is able to attach itself firmly to the surface of the bed,
preventing accidental dislodgement from the substrate.
It is difficult to separate the effects of velocity, depth, and bed morphology (see Minshall
1984; Newbury 1984; Carling 1992), and as such, there are several possible explanations for this
apparent discrepancy. Stream insects exist almost exclusively in a zone that extends only a few
millimetres above the surface of the substrate. Epeorus is no exception; the microhabitat preferences
of this genus are undoubtedly determined by near-bed rather than mean flows. The results, however,
show that velocities near the bed (Uo.o2m} are correlated to mean velocities. If velocities measured at
78

2 em above the bed (U 0 .o2 m) are, in turn, correlated to velocities immediately adjacent to the bed (e.g.
Uo.oo 2 m). then it would appear that local densities of Epeorus are related to low velocities. However,

the relationship between mean velocity and velocity immediately adjacent to the bed at the scale of
the larvae (- 2 mm) may be not be consistent across the range of relative roughnesses (see Hart et
al. 1996).

As depth decreases and substrate roughness increases (primarily due to increased

substrate particle size), velocities adjacent to the bed approach (or, in some cases, exceed) mean
velocity due to flow constriction and the development of pressure gradients (see chapter 3). Thus, in
shallow channels with low relative roughness, near-bed velocities may be higher than predicted by
conventional boundary layer theory (see White 1999), and the distribution of Epeorus may be in
response to some as yet unmeasured variable (e.g. near-bed turbulence intensity u'0002 m ) .
Substantial research into the distribution and nature of near-bed hydraulic parameters is needed
before results of this kind can be interpreted with certainty.
A more probable explanation is that Epeorus larvae are not, in fact, responding to velocity,
but rather to other factors related to velocity. Depth, relative roughness, and substrate composition
are significantly related to mean velocity (Nowell and Jumars 1984; Minshall 1984; Newbury 1984).
As such, the preferences of Epeorus may be for shallow sites with low relative roughness, rather than
low velocity. Larval Epeorus graze periphyton (mainly diatoms) from the exposed upper surfaces of
streambed stones.

Food resource availability may be higher in shallower areas; growth rates of

periphyton are higher in shallow flows, as photosynthetically active radiation (PAR) is attenuated to a
lesser degree. This interpretation is supported by Wellnitz et al. (1996), who concluded that light is
the most important abiotic factor regulating algal abundance in streams . In addition, streambeds
composed of larger particles are more stable (Minshall 1984 ), providing a site for long-term
accumulation of algae. As such, beds with low values of 0/kave should provide a more suitable site for
periphyton growth, and have, as a result, greater resources available to grazing invertebrates. A
positive relationship between grazer abundance and periphyton abundance was demonstrated by
Vaughn ( 1986), who found that densities of the trichopteran grazer Helicopsyche borealis were
highest in areas of high algal chlorophyll a; larvae drifted from food-depleted areas of the substrate.
Similarly, Kerans (1996) found that the hydropsychid grazer Hydropsyche s/ossonae used periphyton
availability to assess microhabitat quality.

79

Oxygen availability may also be a factor in microhabitat selection.

Epeorus larvae are

intolerant of low oxygen concentrations, and are dependent on flow over their abdominal gills to
obtain sufficient dissolved oxygen to survive (personal obseNation) . When the height of the substrate
roughness is less than three times the total depth (D/kave < 3), flow is considered "chaotic" (Davis and
Barmuta 1989; Young 1992; 1993). Flow conditions immediately adjacent to the bed in chaotic flows
are poorly understood (see Hart et al. 1996; chapter 3}, but may provide microflow conditions that
enhance gas exchange.

However, this is not likely the sole factor determining the microhabitat

preferences of the genus, as Epeorus larvae often migrate to the underside of stones during the day
(see chapter 4) where current velocities are substantially lower than the upper surface of streambed
stones.
Chaotic flows may also act to reduce the threat of predation .

Bull trout (Salvelinus

confluentus), common in tributaries of the Torpy River (G . Cho, UNBC grad student, unpublished

data) may be unwilling to move into shallow, chaotic flows to feed . Similarly, American Dippers
(Cine/us americanus) may be unable to forage efficiently in chaotic flows . However, while several

studies (e.g. Hart and Merz 1998; Peckarsky et al. 1990) have demonstrated that high velocities can
reduce invertebrate predation on stream insects, few studies have examined the effect of chaotic
flows on predation intensity. More research on the ecological importance of flow regime is needed.
An alternative explanation for the observed pattern of decreasing Epeorus density with
increasing velocity is a possible ontogenetic shift in microhabitat preferences. Knopp and Cormier
(1997) report that prior to emergence (which generally occurs in July and August), larval Epeorus in
western Canada move from high velocity habitats into slower areas of the streambed . As samples
were collected in late July I early August, pre-emergence behaviours may be a factor. Collier (1994)
found that last instar males of the rheoph ilous mayfly Deleatidium var. myzobranchia preferred lower
velocity habitats than females , possibly reflecting pre-emergence movement into suitable emergence
sites . Increases in larval size may also , in part, explain the apparent preferences of larval Epeorus
for relatively low-velocity habitats. Larger, later instar larvae of the caddisflies Arctopsyche grandis
and Brachycentrus occidentalis are found in less current-exposed locations, possibly due to the
greater "current forces" to which larger individuals are subjected (Voelz and Ward 1996).
Preferences of larval Epeorus for areas of greater relative roughness may also be related to larval
development. Buffagni et al. (1995) found that nymphs of the torrential mayfly species Rhithrogena
80

semicolorata and Ecdyonurus venosus moved into rougher areas in their last instar, possibly to find

suitable sites for emergence. However, in the present study, larval abundance was correlated with
relative bed roughness but not with substrate particle size. Although seemingly contradictory, this
result can be explained if relative roughness is an important predictor of the flow conditions
immediately adjacent to the surface of the substrate (i.e. turbulence intensity and structure, etc.) to
which larval mayflies are exposed . Substrate roughness influences several relevant flow parameters
including turbulence intensity and boundary layer conditions (Davis and Barmuta 1989; Young 1992;
Robert et al. 1996).
Mean velocity (measured at 60% of the total depth) proved to be more closely related to the
distribution of Epeorus than measurements of near-bed velocity (recorded 2.0 em above the bed).
This suggests that mean velocity may be a better predictor of the habitat preferences of Epeorus.
Similarly, mean velocity may be a better indicator of flow conditions at the "patch" scale (i.e. in this
2

case, 0.090 m ) than a single measurement of near-bed velocity obtained in the centre of the patch .
This is undoubtedly due to the heterogeneous nature of near-bed flows. Hart et al. (1996) found that
near-bed velocities differed as much as fourfold over distances as small as a few centimetres.
Interactions between streambed geometry and the current can produce variable local flows , and
measurements of near-bed velocity are likely to be dependent on bed geometry and variation in
microflow regime at the measurement point. This result supports the findings of Quinn and Hickey
(1994) who found that in cobble-bedded channels, mean velocity was approximately as good a
predictor of invertebrate variables as complex hydraulic parameters, while in more variable
environments (especially over gravel beds), complex variables were more closely associated with
invertebrate distributions.
Contrary to expectations, Froude number (Fr) was not a good predictor of Epeorus
distribution.

Orth

and

Maughan

(1983) found

macroinvertebrate taxa and Froude number.

significant

relationships

between

several

Similarly, Wetmore et al (1990) found that Froude

number was the best predictor of feeding sites of the caddisfly Brachycentrus occidentalis and black
fly larvae (Simulium vittatum). However, positioning with respect to Froude number may be important
only in suspension-feeding taxa . B. occidentalis and S. vittatum are both suspension-feeding taxa,
and Wetmore et al. (1990) suggest that the microhabitat preferences of these larvae (Fr =- 0.6 and 0.7, respectively) are the result of larvae positioning themselves in zones of converging flow lines in
81

order to maximise seston delivery (often, larger Froude numbers are associated with accelerating
flows where distances between adjacent, parallel streamlines are decreasing, which in turn increases
the discharge per unit of cross-sectional area).

If the primary ecological significance of Froude

number is as an indicator of particle capture efficiency, it may be relatively unimportant to grazing
macroinvertebrate taxa. Jowett et al. (1991) found that Froude number was not a good predictor of
abundances of Deleatidium, a grazing mayfly. Rather, they found that mean velocity, depth, and
substrate characteristics acted as better predictors of local larval abundance. In addition, Wetmore et
al. (1990) found that the distribution of the grazing caddisfly G/ossosoma intermedium, though poorly
related to Froude number, was significantly and negatively related to depth.

The results of the

present study similarly demonstrate that Froude number was not significantly related to Epeorus
distribution, while mean velocity, depth, Reynolds number, and relative roughness (a substrate
characteristic) were. This suggests that the physical factors to which grazing and suspension-feeding
invertebrates respond may be different. Suspension-feeding organisms may position themselves in
microflow regimes that maximise particle capture rates , while herbivorous macroinvertebrate taxa
may prefer flow regimes that maximise algal growth and gas exchange while minimising predation
and the probability of accidental detachment from the substrate.
It is important to note, however, that the environmental factors measured in this study may
not be those to which stream invertebrates are responding proximally. Several authors, including
Wetmore et al. (1990) and Davis and Barmuta (1989), note that conventional samplers and field
techniques often sample across several microhabitats.

As such, using these techniques (i.e.

propeller velocimeter, Surber sampler), it is difficult to discern the nature of the factors that Epeorus
larvae respond to. Even a relatively small patch of cobble streambed may contain diverse flow types
that vary substantially with respect to important flow descriptors. While many studies have found
significant relationships between invertebrate ecology and a variety of flow characteristics, the nature
of the hydrodynamic parameters to which benthic invertebrates actually respond remains largely
unknown. These proximal hydrodynamic parameters may be related to micro-scale factors . For
example, Hart et al. (1996), using a hot-film anemometer to measure near-bed flows in a natural
stream channel, found that black fly larvae (Simulium vittatum) abundance was related to current
velocity measured 2 mm above the bed (approximately the height of the labral fans).

82

However,

studies of this kind are rare, and research on the relationship between macroinvertebrate behaviour
and flows immediately adjacent to the bed (at the scale of the organisms of study) is needed.

Conclusions
The results of the present study support the hypothesis that hydrodynamic forces are the
underlying ecological gradient that determines the distribution of benthic stream invertebrates at both
the community and taxa-specific levels.

Ordination divided the benthic fauna into distinct

assemblages based primarily on physical factors (substrate, substrate composition, etc.) and
secondarily on hydrochemical factors (pH, conductivity, etc.).

However, while this result provides

insight into the relative ecological importance of the various parameters, it does little to elucidate the
proximal mechanisms that produce these functional assemblages.
Further research is required to identify the parameters to which benthic organisms respond.
However, this may prove difficult as many flow variables are intimately related (e.g. near-bed velocity
and shear stress). For example, the results of the present study reveal that the abundance of the
mayfly Epeorus was negatively related to channel Reynolds number. However, channel Reynolds
number is a ratio that incorporates both channel depth and mean velocity. While it is possible that
Epeorus, and other benthic invertebrate taxa, may respond to flow characteristics associated with low

Reynolds number (e.g. specific regimes of eddy generation), or may have direct preferences for low
velocities or shallow depths, it is possible that they are also responding to additional, unmeasured
environmental parameters that are associated with either depth or mean velocity (e.g. near-bed
velocity, bed roughness, UV radiation, etc.). To adequately describe the habitat preferences of any
species of stream invertebrate, a combination of substrate and hydraulic parameters should be
measured throughout the range of habitat types inhabited by the species.

83

Acknowledgements
Special thanks to those who assisted me in the extensive field work and invertebrate
identification required in the course of this study; Steve Bruce, Michelle Dube, Terry Fortune, and
Shelley Hoover. I gratefully acknowledge Colleen Cassady St. Clair who provided statistical advice.
A special thank you goes to the members of the UNBC Aquatic Sciences Lab for their
encouragement and entertainment. This work was supported by NSERC and FRBC grants to J.D.
Ackerman .

84

References
Allan, J.D. 1995. Stream Ecology: Structure and Function of Running Waters. Chapman and Hall,
London, UK.
Bouckaert, F.W., and J. Davis. 1998. Microflow regimes and the distribution of macroinvertebrates
around stream boulders. Freshwater Bioi. 40 : 77-86.
Buffagni, A. , G. Crosa, and R. Marchetti. 1995. Size-related shifts in the physical habitat of two
mayfly species (Ephemeroptera). Freshwater Bioi. 34 : 297-302.
Carling, P .A. 1992. The nature of the fluid boundary layer and the selection of parameters for
benthic ecology. Freshwater Bioi. 28: 273-284.
Carling, P.A., J.J. Williams, A .. Kelsey, M.S. Glaister, and H.G. Orr. 1998. Coarse bedload transport
in a mountain river. Earth Surf. Proc. Landforms 23: 141-157.
Carlson, J.Y., C.W. Andrus, and H.A. Froehlich. 1990. Woody debris, channel features, and
macroinvertebrates of streams with logged and undisturbed riparian timber in northeastern
Oregon, U.S.A Can. J. Fish. Aquat. Sci. 47:1103-1111 .
Collier, K.J . 1993. Flow preferences of larval Chironomidae (Diptera) in Tongariro River, New
Zealand . N. Zea. J. Mar. Freshwater Res. 27 : 219-226.
Collier, K. 1994. Influence of nymphal size, sex, and morphotype on microdistribution of Deleatidium
(Ephemeroptera: Leptophlebiidae) in a New Zealand River. Freshwater Bioi. 31: 35-42.
Collier, K.J., G.F. Croker, C.W. Hickey, J.M. Quinn, and B.S. Smith. 1995. Effects of hydraulic
conditions and larval size on the microdistribution of Hydrobiosidae {Trichoptera) in two New
Zealand rivers. N. Zea. J. Mar. Freshwater Res. 29: 439-451 .
Danehy. R.J., N.H. Ringler, and R.J. Ruby. 1999. Hydraulic and geomorphic influences on
macroinvertebrate distribution in the headwaters of a small watershed. J. Freshwater Ecol.
14: 79-91.
Davis, J. A., and L. A. Barmuta . 1989. An ecologically useful classification of mean and near-bed
flows in streams and rivers . Freshwater Bioi. 21 : 271-282.
Delgado, J.A., and R. Carbonell. 1997. Case features of caddisfly larvae (Sericostoma selysl) as
related to water velocity and potential to drift. J. Freshwater Ecol. 12: 193-197.
De March, B.G.E. 1976. Spatial and temporal patterns in macrobenthic stream diversity. J. Fish .
Res. Board Can. 33: 1261-1270.
Donahue, W.F., and D.W. Schindler. 1998. Diel emigration and colonization responses of blackfly
larvae (Diptera: Simuliidae) to ultraviolet radiation. Freshwater Bioi. 40 :357-365 .
Edmunds, G.F. Jr., and R.D. Waltz. 1996. Ephemeroptera. in An Introduction to the Aquatic Insects
of North America - Third Edition . Eds. R.W. Merritt and K.W. Cummins. Kendall/Hunt
Publishing Company, lA.
Gauch, H.G. 1982.
Cambridge.

Multivariate Analysis in Community Ecology. Cambridge University Press,

Golubkhov, S.M., T.M. Tiunova, and S.L. Kocharina. 1992. Dependence of the respiration rate of
aquatic insects upon the oxygen concentration in running and still water. Aquatic Insects 14:
137-144.
85

Growns, 1.0., and J.A. Davis . 1991 . Comparison of the macroinvertebrate communities in streams in
logged and unlogged catchments 8 years after harvesting. Aust. J. Mar. Freshwater Res. 42 :
689-706.
Growns, 1.0., and J.A. Davis. 1994. Longitudinal changes in near-bed flows and macroinvertebrate
communities in a Western Australian stream . J. N. Am. Benthol. Soc. 13:417-438.
Gurtz, M.E., J.R. Webster, and J.B. Wallace. 1980. Seston dynamics in southern Appalachian
streams: effects of clearfelling . Can. J. Fish. Aquat. Sci. 37: 624-631 .
Hart, D.D., B.D. Clark, and A. Jaentuliyana. 1996. Fine-scale field measurement of benthic flow
environments inhabited by stream invertebrates. Limnol. Oceanogr. 41 : 297-308.
Hart, D.D., and R.A. Merz. 1998. Predator-prey interactions in a benthic stream community: a field
test of flow-mediated refuges. Oecologia 114: 263-273.
Hawkins, C.P., J.N. Hogue, L.M. Decker, and J.W. Feminella. 1997. Channel morphology, water
temperature, and assemblage structure of stream insects. J. N. Am . Benthol. Soc. 16: 728749.
Hearnden, M.N., and R.G. Pearson . 1991. Habitat partitioning among the mayfly species
(Ephemeroptera) of Yuccabine Creek, a tropical Australian stream. Oecologia 87: 91-101 .
Hubert, W.A., and S.J. Kozel. 1993. Quantitative relations of physical habitat features to channel
slope and discharge in unaltered mountain streams. J. Freshwater Ecol. 8: 177-183.
Jones, J.B., and R.M . Holmes. 1996. Surface-subsurface interactions in stream ecosystems. Trends
Ecol. Evol. 11: 239-242.
Jowett, I.G., J. Richardson, B.J.F. Biggs, C.W. Hickey, and J.M . Quinn . 1991. Microhabitat
preferences of benthic invertebrates and the development of generalised Deleatidium spp .
habitat suitability curves, applied to four New Zealand rivers. N. Zea. J. Mar. Freshwater
Res . 25: 187-199.
Kerans, B.L. 1996. The influence of periphyton and rock texture on the diel drift periodicity of a
Hydropsychid caddisfly. J. Freshwater Ecol. 11: 163-169.
Knopp, M., and R. Cormier. 1997. Mayflies: An Anglers Study of the Trout Water Ephemeroptera .
Greycliff Publishing, Helena, MT.
Lacoursiere, J.O. 1992. A laboratory study of fluid flow and microhabitat selection by larvae of
Simuliium vittatum (Diptera : Simuliidae). Can. J. Zool. 70 : 582-596 .
Malmqvist, B., and G. Englund . 1996. Effects of hydropower-induced flow perturbations on mayfly
(Ephemeroptera) richness and abundance in north Swedish river rapids. Hydrobiol. 341 :
145-158.
Marchant, R., L. Metzeling, A. Graesser, and P. Suter. 1985. The organization of macroinvertebrate
communities in the major tributaries on the LaTrobe River, Victoria, Australia. Freshwater
Bioi. 15: 315-331.
Merritt, R.W., and K.W. Cummins. 1996. An Introduction to the Aquatic Insects of North AmericaThird Edition . Kendall/Hunt Publishing Company, lA.
Minshall, G.W. 1984. Aquatic insect-substratum relationships . in The Ecology of Aquatic Insects.
Eds. V.H. Resh and D.M. Rosenberg. Praeger Publishers, New York, NY.
Muotka, T., and A. Pentinnen . 1994. Detecting small-scale spatial patterns in lotic predator-prey
relationships : statistical methods and a case study. Can . J. Fish . Aquat. Sci. 51: 2210-2218.
86

Newbold, J.D. , D.C . Erman, and K.B. Roby. 1980. Effects of logging on macroinvertebrates in
streams with and without buffer strips. Can . J. Fish . Aquat. Sci. 37: 1076-1085.
Newbury, R.W. 1984. Hydrologic determinants of aquatic insect habitats. in The Ecology of Aquatic
Insects. Eds. V.H. Resh and D.M. Rosenberg. Praeger Publishers, New York, NY.
Noel, D.S., C.W. Martin, and C.A. Federer. 1986. Effects of forest clearcutting in New England on
stream macroinvertebrates and periphyton. Env. Manag. 10: 661-670 .
Nowell, A.R.M., and P.A. Jumars. 1984. Flow environments of aquatic benthos. Ann . Rev. Ecol.
Syst. 15: 303-328.
Orth, D.J., and M.E. Maughn. 1983. Microhabitat preferences of benthic fauna in a woodland
stream . Hydrobiol. 106: 157-168.
Osborne, L.L. , and E.E. Herricks. 1987. Microhabitat characteristics of Hydropsyche (Tricoptera :
Hydropsychidae) and the importance of body size. J. N. Am . Benthol. Soc. 6: 115-124.
Palmer, M.W.
1993.
Putting things in even better order: the advantages of canonical
correspondence analysis. Ecology 74: 2215-2230 .
Peckarsky, B.L., and R.S . Wilcox. 1989. Stonefly nymphs use hydrodynamic cues to discriminate
between prey. Oecologia 79 : 265-270.
Peckarsky, B.L., S.C. Horn, and B. Statzner. 1990. Stonefly predation along a hydraulic gradient: a
test of the harsh-benign hypothesis. Freshwater Bioi. 24: 181-191 .
Plenet, S. H. Hugueny, and J. Gilbert. 1996. Invertebrate community responses to physical and
chemical factors at the river/aquifer interaction zone II . Downstream from the city of Lyon .
Arch . Hydrobiol. 136: 65-88 .
Quinn, J.M., and C.W. Hickey. 1994. Hydraulic parameters and benthic invertbrate distributions in
two gravel-bed New Zealand rivers. Freshwater Bioi. 32: 489-500.
Quinn, J.M., C.W. Hickey, and W. Linklater. 1996. Hydraulic influences on periphyton and benthic
invertebrates: simulating the effects of upstream bed roughness. Freshwater Bioi. 35: 301309.
Rempel, L.L., J.S. Richardson, and M.C. Healey. 1999. Flow refugia for benthic macroinvertebrates
during flooding of a large river. J. N. Am . Benthol. Soc. 18: 34-48 .
Resh, V.H., and D.M. Rosenberg (Eds) . 1984. The Ecology of Aquatic Insects. Praeger Publishers,
New York, NY.
Robert, A. , A.G . Roy, and B. De Serres. 1996. Turbulence at a roughness transition in a depth
limited flow over a gravel bed . Geomorphology 16: 175-187.
Robertson, A.L. , J. Lancaster, and A.G . Hildrew. 1995. Stream hydraulics and the distribution of
microcrustacea: a role for refugia? Freshwater Bioi. 33: 469-484 .
Robertson, A.L., J. Lancaster, L.R. Belyea, and A.G . Hildrew. 1997. Hydraulic Habitat and the
assemblage structure of stream benthic microcrustacea. J. N. Am . Benthol. Soc. 16: 562-575.
Rossaro, B., and A. Pietrangelo. 1993. Macroinvertebrate distribution in streams : a comparison of
CA ordination with biotic indices. Hydrobiol. 263: 109-118.
Shannon, J.P., D.W. Blinn, and L.E . Stevens. 1994. Trophic interactions and benthic animal
community structure in the Colorado River, Arizona, U.S.A. Freshwater Bioi. 31 : 213-220.
87

Statzner, B., and T.F. Holm. 1989. Morphological adaptation of shape to flow: microcurrents around
lotic macroinvertebrates with known Reynolds numbers at quasi-natural flow conditions.
Oecologia 78: 145-157.
Statzner, B., and B. Higler. 1986. Stream hydraulics as a major determinant of benthic invertebrate
zonation patterns . Freshwater Bioi. 16: 127-139.
Stone, M.K., and J.B. Wallace. 1998. Long-term recovery of a mountain stream from clear-cut
logging: the effects of forest succession on benthic invertebrate community structure.
Freshwater Bioi. 39: 151-169.
ter Braak, C.J.F. 1986. Canonical correspondence analysis: a new eigenvector technique for
multivariate direct gradient analysis. Ecology 67: 1167-1179.
ter Braak, C.J.F. 1991 . CANOCO- A FORTRAN program for Canonical Community Ordination.
Version 3.12. Microcomputer Power, Ithaca, New York.
ter Braak, C.J.F. 1995. Ordination . in Data Analysis in Community and Landscape Ecology. Eds.
R.H.G. Jongman, C.J.F. ter Braak, and O.F.R. Tongeren . Cambridge University Press.
Thorp, J.H., and A.P. Covich. 1991. Ecology and Classification of North American Freshwater
Invertebrates. Academic Press, San Diego, CA.
Townsend, C.R., M.R. Scarsbrook, and S. Doledec. 1997. The intermediate disturbance hypothesis,
refugia, and biodiversity in streams. Limnol. Oceanogr. 42: 938-949.
Vannote, R.L., G.W. Minshall, K.W. Cummins, J.R. Sedell, and C.E. Cushing.
continuum concept. Can. J. Fish. Aquat. Sci. 37: 130-137.

1980. The river

Vaughn, C.C. 1986. The role of periphyton abundance and quality in the microdistribution of a
stream grazer, Helicopsyche borealis {Trichoptera: Helicopsychidae). Freshwater Bioi. 16:

485-493.

Vogel, S. 1994. Life in Moving Fluids, 2"d Edition . Princeton University Press, Princeton, NJ.
Voelz, N.J., and J.V. Ward .
1996.
Microdistributions of filter-feeding caddisflies {Insecta:
Trichoptera) in a regulated Rocky Mountain river. Can . J. Zool. 74: 654-666 .
Wallace, J.B., and M.E. Gurtz. 1986. Response of Baetis mayflies (Ephemeroptera) to catchment
logging. Am. Midland Nat. 115: 25-41.
Way, C.R., A.J. Burky, C.R. Bingham, and A.C. Miller. 1995. Substrate roughness, velocity refuges,
and macroinvertebrate abundance on artificial substrates in the lower Mississippi River. J. N.
Am. Benthol. Soc. 14: 510-518.
Weissenberger, J., H.-Ch . Spatz, A. Emanns, and J. Schwoerbel. 1991 . Measurement of the lift and
drag forces in the mN range experienced by benthic arthropods at flow velocities below 1.2 m
s-1 . Freshwater Bioi. 25: 21-31 .
Wellnitz, T.A., R.B. Rader, and J.V. Ward . 1996. Importance of light and nutrients in structuring an
algal community in a Rocky Mountain stream . J . Freshwater Ecol. 11 : 399-413.
Wetmore, S.H., R.J . Mackay, and R.W. Newbury. 1990. Characterization of the hydraulic habitat of
Brachycentrus occidentalis, a filter-feeding caddisfly. J. N. Am. Benthol. Soc. 9: 157-169.
1

White, F.M. 1999. Fluid Mechanics- 4 h Edition . WCB McGraw-Hill, Boston .

88

Wiggins, G.B. 1996. Trichoptera Families. in An Introduction to the Aquatic Insects of North
America - Third Edition . Eds. R.W. Merritt and K.W. Cummins . Kendall/Hunt Publish ing
Company, lA.
Williams, D.O ., N.E. Williams, andY. Cao. 1997. Spatial differences in macroinvertebrate community
structure in springs in southeastern Ontario in relation to their chemical and physical
environments. Can . J. Zool. 75: 1404-1414.
Young , W.J. 1992. Clarification of the criteria to identify near-bed flow regimes . Freshwater Bioi. 28:
383-391.
Young, W.J. 1993. Field techniques for the classification of near-bed flow regimes. Freshwater Bioi.
29: 377-383.

89

Chapter 2 - Figures

-

..........

--.. .....

c: 18
16 ••
(/)
(/)
•
Q) 14
c: 12
..c:::
(.)
·c 10 ~
(.)
.E 68 liM•
~
0
c: 4 ••• •
0
•
>< 2
ro
0
.000

• ••••

---- ••

...•• ••

-

•

•

•
• ••
•• •• ••
••
• •

.200

2

R = 9.20%

•• •
• -.
• ••
•

• ••
•- •
•• ••
.600

.400

.800

discharge Q (m 3 s- 1)

A.

- ,_.
..........

c: 25

( /)
(/)

Q)

20

c:
..c::: 15
(.)
·c
(.)
10

E
0

c:

0

><
ro

5

•
t.._•

·:, i

•• •

0
.000

2

R = 9.50%

. - • •• •• ••-••
••• •

.200

.400

.600

.800

discharge Q (m 3 s- 1)
B.
Figure 2-1. Taxonomic richness of macroinvertebrate taxa in relation to stream discharge. A.
Taxonomic richness measured per 'patch' (per Surber sample, - 0.090 m\ B. Taxonomic richness
measured per stream (total taxa collected in all three benthic sample replicates per stream).
Macroinvertebrates identified to lowest convenient taxonomic level, usually genus.

90

,, •

450
-400
~ 350
.....>. 300
·:;: 250
:;::;
u 200
::J
"0 150 ~
c 100
0
u 50 It~
0
.000

••
•
••
•

••

• ••

.200

• •
•
•
•
•

.400

.600

.800

discharge Q (m 3 s- 1)
Figure 2-2.
Columbia.

Conductivity versus stream discharge in tributaries of the lower Torpy River, British

Figure 2-3. (next page) Canonical correspondence analysis (CCA) biplot for the 24 benthic
macroinvertebrate taxa and the 11 environmental variables of the lower Torpy River, British
Columbia. For clarity, the biplot has been divided into constituent taxonomic (top) and environmental
(bottom) variable components. Taxa scores labelled with open circles, environmental variables
(vectors) labelled with arrows. A ln(x+1) transformation was performed on the taxa data. The
following environmental variables are indices: DEFOL (riparian vegetation defoliation index; 0 = intact
riparian vegetation, 1 = defoliated riparian vegetation), SUBSTR (substrate composition; higher
values indicate increasingly fine bed material), LOGPROX (logging activities in proximity of stream
catchment; 0 = no logging activity in catchment, 1 = logging in catchment), and RIPARIAN (riparian
vegetation index; higher values indicate increasingly open streamside vegetation). See Methods and
Materials and Tables 2-4 and 2-5 for further clarification of abbreviations used .

91

.

Axis II
1-

Agapteuso

Arctopsyo
Kogotuso

ZapadaC

OVisoka

LepidostO

Axisl-1 .5----~~
G/ossosoo

0

Epeoruso

Yorape

Rhyacoph
Elmidae

--~~~~
~~~~~-----Rhithrog Chilll'nom
PsyChodi
I
Cin

mu/0

DrunellaO
Megarr:yso

0

Chlorope

0

Oligocha

OTabanida

1.5

OEmpidida

OBaeffs
Tipulida()

NeothremO

-2 -

Axis II

SLOPE
Q

"'..

Axis I
-1 .5

~

~

DEFOL

I

I LOGPROX
/DO
pH
COND

1.5

RIPARIAN

-2 ...J

92

Cl)

4

Q)

2

.. •.•• • •• •
• •
••
..••• .

0
0.00

0.10

-

8

7
+ 6
><
c:: 5

-.r-

2
3
0

Lfr 1

0.30

0.20

R2 = 16.73%

•

0.40

0.50

depth D (m)
Figure 2-4. Relationship between abundance of Epeorus larvae and channel depth (D) in streams 48
and 56 in the lower Torpy River watershed, British Columbia.

-

8

2

R = 25.36%

7
+ 6
><
c:: 5

-.r-

Cl)

2

0

Q)

Lfr

4

3
2
1
0
0.25

•

• • •

0.50

0.75

1.00

1.25

velocity U (m s- 1)
Figure 2-5. Relationship between abundance of Epeorus larvae and mean velocity (U, measured at
0.60 - see text for details) in streams 48 and 56 in the lower Torpy River watershed, British
Columbia.

93

8 ~--------------------------------

+
><
~

c

Cl)

2

0

•

7
6

5

4
3

~ ~

R2 =31.19%

0

------ ----- - --------~~ ---~

0

50000 100000 150000 200000 250000

channel Reynolds number Re
Figure 2-6. Relationship between abundance of Epeorus larvae and channel Reynolds number (Re)
in streams 48 and 56 in the lower Torpy River watershed , British Columbia.

-

8
7
+ 6
><
c 5
Cl) 4
20 3
Cl> 2
1

•

~

tfr

R =23.17%
2

•

0
0

1

2

3

4

5

6

7

8

9

D/kave
Figure 2-7. Relationship between abundance of Epeorus larvae and relative roughness (D/kave) in
streams 48 and 56 in the lower Torpy River watershed , British Columbia .

94

-

.,...
I

(/)

1.20
1.00

•

E 0.80
..........

::J 0.60
>.
......

'(3

0

•

••

••••
••• •

0.40

Q)

> 0.20

R =7.73%
2

••• •
•

• •

0.00 +------,-------,------.-------,-----_,
0.30
0.40
0.50
0.10
0.20
0.00

depth D (m)
Figure 2-8. Relationship between mean velocity (U) and depth (D) measured in streams 48 and 56,
located in the lower Torpy River watershed, British Columbia.

-

~

' (/)

E

E

1.00
0.80

R2 = 45.47%

N
0

•

ci

-

::J 0.60
>.

' (3

0

Q)

>

0.40

"0

Q)
.cI 0.20

......
(IJ
Q)

c

•

•

0.00
0.20

•• •

•

0.40

0.60

0.80

1.00

1.20

1

mean velocity U (m s- )
Figure 2-9. Relationship between near-bed velocity (Uo.o2m- velocity measured 2 em above the bed)
and mean velocity (U - measured at 0.60). See text for details.

95

Chapter 3
Near-bed hydrodynamics above boulders in shallow torrential
streams

Summary
1.

The near-bed hydraulic environment inhabited by torrential stream fauna was characterised by
recording velocity profiles, near-bed velocities, and shear stress distributions over the upper
surface of eight streambed boulders in a mountain stream located in east-central British
Columbia. Flow at the study site was rough-turbulent.

2.

Velocity profiles above the boulders regularly deviated from a semi-logarithmic shape. Velocity
profiles were often found to be 'wedge-shaped', with velocities increasing from the free surface
downward, reaching a maximum a few millimetres above the stream bed; velocities below this
point decreased rapidly and logarithmically towards the bed surface.

Near-bed acceleration

resulting from flow constriction by the curved leading edge of the boulder protruding upward is
proposed as the mechanism producing this distinctive velocity distribution .
3.

Wall shear stress (-rw). measured using a Preston-static tube (PST) was generally lowest over the
leading edge of the boulders, and increased over the upper surface, reaching a maximum near
the rear of the stone (upstream of the point of flow separation, where the shear stress dropped off
abruptly}. High shear stresses were also recorded on the sides of the boulders.

4. Wall shear stress measurements made using the PST were similar to shear stress estimates
made using the law of the wall (i.e. the velocity gradient method; estimating -rw based on the slope
of the regression of U on In z) only at locations where velocity profiles were log-normally
distributed.

Where velocity profiles were wedge-shaped, the velocity gradient method

underestimated 'tw compared to the direct measurements made using the PST.
5. Bed geometry was found to be an important influence on near-bed flows in shallow streams with
low relative submergences. Possible mechanisms by which bed geometry may influence nearbed flow parameters are discussed.

96

Introduction
The hydrodynamic forces created by moving water are perhaps the most important factors
governing the behaviour, distribution, and ecology of benthic invertebrates in stream environments
(Carling 1992; Hart et al. 1996; Hart and Finelli 1999). Benthic organisms living on the exposed
surfaces of boulders and gravels are subjected to a suite of hydraulic forces that are variable in both
time and space (Hart et al. 1996; Buffin-Belanger and Roy 1998).

While at first glance this

environment may seem hostile, there are benefits to living in habitats of this type. For example, high
near-bed velocities decrease the risk of predation (Peckarsky et al. 1990; Hart and Merz 1998),
increase the respiratory efficiency of the gills of stream benthic insects (Wiley and Kohler 1980;
Pommen and Craig 1995), and increase the rate of seston delivery to suspension-feeding benthic
invertebrates (Wetmore et al. 1990; Lacoursiere 1992). However, high velocities also increase the
lift, drag, and shear stress acting to dislodge invertebrates from the substrate (Weissenberger et al.
1991; Vogel 1994; also see Denny 1994; Delgado and Carbonell 1997).
In addition to their role in regulating the biology of streams, hydrodynamics influence a
number of physical processes, including dissolved oxygen levels (Moog and Jirka 1999), near-bed
flow patterns (Jarrett 1990; Bergeron 1994), the level and structure of turbulence (Robert et al. 1996;
Buffin-Belanger and Roy 1998; Sukhodlov et al. 1998), bed roughness (Nikora et al. 1998) and
bedload movement (Blizard and Wahl 1998). Yet, despite the numerous studies that have been
conducted within streams, few descriptions of stream flow have been made at spatial scales
important to stream organisms (i.e. within millimetres of the surface of the bed; see Hart et al. 1996).
Moreover, the generation of eddies, wall shear stress, and other boundary layer phenomena, which
take place at these spatial scales, are likely to have direct effects on stream organisms .
A number of recent studies have stressed the need to study the biology and environment of
benthic invertebrates at the small spatial scales at which they experience flow (Carling 1992; Hart et
al. 1996).

However, due to the difficulty of characterising flow at organism-defined scales, most

research efforts continue to examine the biology of stream macroinvertebrates at reach, or even
watershed scales. It is reasonable to suggest that an understanding of the proximate and ultimate
forces that shape the behaviour, ecology, and morphology of stream macroinvertebrates will be
gained by examining the hydraulic processes of streams at biologically relevant spatial scales. This

97

is especially true of torrential streams where benthic organisms depend on high velocity habitats for
nutrient delivery, protection from predation, and gas exchange (Vogel1994).
Given the importance of small scale fluid phenomena to stream organisms, and the relative
rarity of studies at these scales, it seems timely to describe the near-bed hydraulic environment
experienced by larval insects in torrential streams with steep slopes and coarse substrates.

To

achieve this end, this study has three specific goals: (1 ), to examine the distribution of velocities over
the upper surfaces of torrential stream boulders; (2), to measure the range and distribution of shear
stress on stream boulders; and (3), to characterise the local channel hydraulics of torrential stream
environments. This will be achieved through description of the flow environment of a shallow stream
at high spatial resolution, as well as measurements of wall shear stress in a natural channel.

Methods and materials
Study Area and Experimental Approach
The study site was located on a tributary of the Torpy River, located approximately 90 km
east of Prince George, British Columbia, Canada (Figure 3-1 ). The 20 m study reach was a single
thread, generally straight channel, with an average stream surface slope of 3.2%, although local
water surface slopes regularly exceeded 70% . The bed was composed primarily of boulders and
cobbles with patches of coarse gravel in the interstices (pockets) between the larger substrate
elements (see chapter 2 for a definition of stone size classes). Streambed gravels were poorly
cemented, due to the high gradient and seasonally variable discharge of the study stream. The
maximum discharge of the stream recorded at the site during the study period (Aug .15 to Oct.15,
3

1

1998) was 0.43 m s· on October 2, 1998.
In order to examine the relationship between bed geometry, bulk flow, near-bed flow, and the
benthic biota, a series of stones (also referred to as boulders depending on size; see below) were
deployed in the stream channel at the site. Thirty stones (ranging in size from 22 - 36 em in
diameter; i.e. small boulders, referred to below as stones or boulders, depending on size) were
collected from the banks of streams within the Torpy River watershed, with care taken to obtain
stones from above the high-water line. Obtaining stones in this fashion ensured that they would not
have prior accumulations of dried biofilm from previous immersions, a factor that might influence

98

subsequent growth of periphyton . Light coloured (off-white to tan), naturally rounded stones were
chosen in order to facilitate the observation of insect and algal distributions, as well as minimise the
influence of substrate colour on insect behaviour. The stones were subsequently removed to the
Aquatic Sciences Laboratory at UNBC, where they were cleaned with a stiff nylon brush and marked
with an identification number and a 5 x 5 em grid of points (Figure 3-2). The reference grid provided
an accurate and reliable system to facilitate the recording of the distribution of hydraulic and biotic
parameters over the surface of each stone. The reference points were distributed in a modified
Cartesian grid, the centre of which was a datum located at the intersection of the longest and second
longest orthogonal axes (i.e. the centre of the upper surface of the stone). Both the grid and the
identification number were applied to the stone using an indelible black felt marker (Sharpie®
Permanent Marker; Sanford, USA).
The stones were deployed in the active channel at the study site on August 20, 1998. In
order to ensure that the stones were placed in 'natural' orientations, where possible, naturally
occurring boulders were removed from the stream bed and replaced with stones of similar shape and
volume. In addition, to minimise hydraulic interactions between substrates (see Davis and Barmuta
1989; Young 1993), the stones were placed in the channel at approximately 1 m intervals in an offset
pattern, such that no stone was immediately upstream of another. After deployment, the stones were
left undisturbed for 10 days in order to accumulate the biofilm, periphyton, and benthic invertebrates
characteristic of the stones of the surrounding stream bed. During this time the densities of larval
Epeorus longimanus (a representative high velocity, torrential habitat species) were monitored on

each of the deployed stones . After the acclimatisation period had ended, a series of measurements
were made on the five stones that had the highest densities of E. longimanus. The measurements,
which were used to estimate the preferred microhabitat of this torrential mayfly, included stream bed I
water surface profiles, wall shear stress distributions, midline and midpoint velocity profiles, and
identification of basic flow patterns through flow visualisation using fluorescein dye (see below). For
comparison and verification, the flow fields of three additional naturally occurring boulders with
similarly high densities of E. longimanus were also examined in the same manner.
Some of the techniques were tested under field and laboratory conditions prior to their use in
the field.

Specifically, to compare the hydraulic conditions measured at the torrential stream site

(stream 48) with the hydraulic conditions of relatively simple channels, a velocity profile, shear stress
99

estimate, and stream bed I water surface profile were obtained over a gravel bed riffle in an adjacent
stream of lower average slope (1 .5%) but similar discharge.

Three velocity profiles were also

collected in a laboratory flume at UNBC (20 em x 20 em x 200 em long) at different velocities (0.242,
0.382, and 0.465 m s·\

Apparatus
The water surface and stream bed topography were measured in transects above each of the
deployed and natural stones using a stream hydraulics profiling device, similar to the Wilson Creek
flow profiler used by Wetmore et al. (1990). Whereas the Wilson Creek profiler consisted of a point
gauge mounted on a triangular tripod table, the apparatus used in this study consisted of a
rectangular, field deployable stage (table) with an adjustable leg at each corner. A cylindrical PVC
socket (3" diameter) was mounted vertically in the centre of the field stage, permitting the quick and
easy interchange of the two types of apparatus used to quantify the stream environment; (1) a point
gauge to measure water surface I stream bed topography, and (2) a Preston-static tube (see below,
and Appendix 4) to measure water velocity and shear stress.

The field stage, while minimising

disturbance to local flows, also allowed the Preston-static tube and point gauge to be fixed at any
height with a minimum of flow-induced vibration.

The stage consists of a rectangular aluminium

frame (40 x 152 em) which could be levelled above the stream using the adjustable aluminium legs
and a circular liquid bubble level. Positioning of the Preston-static tube in the streamwise (x) direction
was made possible by a smaller (40 x 40 em) aluminium stage which could be adjusted on the rails
formed by the long axis of the larger frame. The second, smaller frame in turn supports an acrylic
sheet (36 x 40 x 1 em), in the centre of which the point gauge was fixed. The acrylic sheet, capable
of sliding over the second aluminium frame on polyethylene runners, provided positioning in the
cross-stream (y) direction, while a rack and pinion gauge fitted with a vernier scale provided
positioning in the vertical (z) plane. Using this method, the Preston-static tube or point gauge could
be positioned accurately at any point above the surface of the stream bed.
Measurements of near-bed velocity profiles were made using a small-scale Preston-static
tube (PST), essentially a Pitot-static constructed of a 90° bevelled tip syringe needle fixed to a sidebored syringe needle (tube diameter = 0.88 mm; see Ackerman et al. (1994) for a detailed

100

description) .

The Preston-static tube differs from a standard Preston tube in that the dynamic

pressure port (which faces into the flow) is coupled with a static pressure port (on the sidewall). As a
result, the difference between static pressure Ps and dynamic (or stagnation) pressure Po can be
measured, and velocity or shear stress can be determined, depending on deployment (see below). A
Validyne model CD379-1-2 membrane-type differential pressure transducer (Validyne Engineering ;
Northridge, CA), connected to the Preston-static tube via 2.0 m of flexible pressure lines (inside
diameter= 3 mm), was used to measure differential pressure. Before each set of measurements was
taken, the pressure lines connecting the Preston-static tube and pressure transducer were visually
inspected for the presence of any introduced air bubbles that would potentially alter pressure
transmission . Both the pressure transducer and the pressure lines were periodically flushed in order
to minimise the possibility of the presence of trapped air, which would bias the differential pressure
measurements. The pressure difference (which is related to velocity; see equation 3-2 below) was
expressed as a voltage by a Validyne portable indicator (model CD379).

Fifteen voltages were

recorded over the course of approximately 30 seconds, due to the relatively slow response time of the
Preston-static tube I pressure lines I pressure transducer system. These voltages were converted to
differential pressure, then averaged .
One of the primary disadvantages of both Preston tubes and Pitot tubes is that they must be
aligned parallel to the flow; there are substantial errors in both Ps and Po associated with yaw angles
greater than 10 degrees (White 1999). Preliminary diagnostic tests in both the lab flume and the
natural stream channels confirmed that the inaccuracies associated with positioning errors less than
10 degrees are negligible (Ackerman and Hoover 2001 ).

When deployed in the field, flow

visualisation using sodium fluorescein injected upstream of the Preston tube was used to ensure that
the Preston tube was oriented parallel to the flow. To record velocity at a given height above the
substrate, the Preston tube was deployed at the correct depth using the rack and pinion gauge.
Then, using a long, thin needle (length = 200 mm , diameter = 1 mm), a filament of dye was injected
approximately 10 mm upstream of the tube.

The orientation of the tube with respect to the dye

filament was then checked with a magn ifying, subsurface periscope, which was oriented in the water
perpendicular or downstream of the Preston-static tube. If the tube and dye trail were not parallel (i.e.
the Preston tube was incorrectly oriented), the tube was readjusted in order to ensure that it was
oriented directly into the flow.
101

The Preston-static tube used in this study is configured such that it can be used to directly
measure wall shear stress when deployed on a surface (Ackerman et al. 1994 ). The Preston-static
tube (PST} was developed for measuring wall shear in situations where the incorporation of a wallmounted static pressure tap is impractical {i.e. around small-scale objects where flow lines may be
curved), but has been recently tested for use in a variety of field and lab conditions (Ackerman and
Hoover 2001 ).
The relationship between the voltage output of the pressure transducer and pressure head
was determined by calibrating the pressure transducer by varying the pressure, produced by a
column of water of known height, across the transducer membrane. A regression of head (height}
against the voltage output provided the following relationship:

Head= -0.00164 + (0.0508 x voltage)

27

2

(n = 15, R = 1.00; ANOVA F{1,13) = 5.29 x 10

, p < 0.001).

(3-1)

Bernoulli's equation, which relates

pressure, velocity, and height of the fluid {head}, is

2

P + 1/2pU + pgh =constant,

(3-2)

where P is the pressure in the moving fluid, p is the density of the fluid, g is gravitational acceleration
2

(9.81 m s" ), and h is the height of the fluid . Given equation 3-2, the relationship between pressure
{head) and velocity is

U=

~

(3-3)

where the variables are as defined above.

Preston ( 1954) demonstrated that the non-

dimensionalised pressure difference, x*, can be expressed as

(3-4)

102

where ~

is the pressure difference, d is the diameter, and u is the kinematic viscos ity of water.

Similarly, the non-dimensionalised shear stress, y*, can be expressed as

(3-5)

where 'tw is wall shear stress.

Preston (1954) (also see Ackerman et al. 1994) showed that the

relationship between x* and y* in the wall layer is

y* = -1 .394 + 7/8 x* ,

(3-6)

which can be represented as

(3-7)

where the constant A is given as

(3-8)

and the pressure difference ~

~

=

is given by

(3-9)

~

The constant A is fixed at a given water temperature and density (density p and kinematic viscosity u
are dependent on temperature; see appendix A in White (1999)), while the pressure difference ~
a function of the density of the medium (p), acceleration due to gravity (g; 9.81 m s·\ and head ~
Channel conditions were characterised using Froude number (Fr).

Froude number, an

important descriptor of flow conditions in open channels, is the ratio of mean channel velocity to the
05

propagational velocity of a surface wave (Fr = U I (gD) . ) . Where Froude numbers are less than
unity (Fr < 1; subcritical flow), flows which are characterised by deep, smooth flowing water. Where
103

Froude numbers exceed unity (Fr > 1; supercritical flow), flows are characterised by broken , white
water (see White 1999 and chapter 1 for further discussion).

Sampling
The shape of the water surface and stream bed were measured along downstream transects
crossing the centre of the stones studied using the point gauge and vernier scale. Generally, water
surface and bed elevations were obtained at 10 mm intervals over the boulders and at 20 mm
intervals over the stream bed in front and rear of the boulder. The total length of each transect varied
with the centreline diameter of the boulder.
Velocity profiles were recorded above the centre (datum) of each of the five deployed and
two of the naturally located boulders . In addition, velocity profiles were recorded above the surface of
a smaller stone (N03), and above a gravel riffle (GR). Each profile consisted of velocities taken at
incrementally increasing distances from the surface of the stone (0, 1, 2, 3, 5, 7, 9, 12, 15, 18, 23, 28,
and 33 mm). In instances where the water surface was greater than 33 mm from the surface of the
stone, additional velocity measurements were taken at 10 mm intervals. These heights were chosen
in order to provide the resolution necessary to describe both the near-bed and bulk flow fields. In all
cases, velocity measurements could not be obtained in a zone extending approximately 3- 10 mm
down from the water surface, due to the periodic exposure of the PST tip to air during frequent
fluctuations in depth of the highly turbulent flow. Velocity profiles were obtained at 50 mm intervals
along the centre line of the upper surface of three of the deployed boulders (stones 7, 31, and 25) in
order to assess the changes in flow that occur as water crosses the upper surface of stream
boulders.
Distributions of wall shear stress over the surface of the five deployed boulders (stones 7, 31 ,
25, 23, and 10) and two of the three naturally occurring boulders (N01 and N02) were measured
using the PST. In each case, the spatial distribution of shear stress was measured by recording
shear stress in perpendicular transects . A streamwise transect of shear stress measurements was
taken along the centreline of each boulder in the direction of flow at 25 mm intervals, while the
second, cross-stream transect was taken perpendicular to the first at 50 mm intervals, crossing the
streamwise transect at the centre of each boulder.

The location of each shear measurement

corresponded to points of the reference grid. In addition, single measurements of shear stress were
104

made on the upper surface of the third naturally occurring stone (N03), the gravel bed riffle (GR), and
in the UNBC lab flume at the three different velocity settings.

Data Analysis
The stream bed I water surface data were plotted as profiles showing the upper and lower
boundaries of the flow measured over each of the boulders. These profiles are hereafter referred to
as "surface/bed contours" in order to distinguish them from the vertical velocity profiles obtained
above the bed, which are termed "velocity profiles".
Conventionally, mean velocity is determined as the velocity at 60% of the total depth (U 0 .60 ).
However, Jarrett (1990) recommends that, in the case of mountain rivers, mean velocities are more
accurately approximated by mid-depth velocities (Uo.so) than velocities at 60% of the total depth
(U 0 .60 ). As such, for the purposes of this study, 'mean' or mid-depth velocities were estimated using
Jarrett's (1990) 0.50 guideline. Near-bed (Uo.oo 2m) and mid-depth (U 0 .50 ) velocities were estimated
from the velocity profiles by interpolation using the velocities at the two depths of the profile that
bracket the depth for which a velocity is required. The error associated with this method was thought
to be minimal due to the high spatial resolution of the velocity profiles obtained (as noted above).
Velocity profiles were categorised as belonging to one of three categories; (1) log-normal
(displaying a typical semi-logarithmic shape); (2) wedge-shaped (displaying a distinctive shape in
which velocities increase in logarithmic fashion downwards from the water surface, reaching a
maximum value at a point near the bottom, and then decreasing); or (3) transitional (velocity profiles
showing some near-bed distortion, and appear to be intermediate between log-normal and wedgeshaped profiles). Tests of linearity (e.g. XLOF (Minitab 1985; Minitab Inc., State College, PA); see
Bergeron 1994) were not used to determine if velocity profiles were log-linear.

Rather, a vertical

velocity variation index (0) of the relative magnitude of the difference between mid-depth and nearbed velocities (equation 3-1 0) was used to categorise the velocity profiles, because the vertical
distribution of near-bed velocities is a primary focus of the study. The index 0 is given as

0 = (Uo.so- Uo.oo2m) I Uo.so

(3-1 0)

105

where the subscripts 0.50 and 0.002m refer to 50% of the total channel depth and 2 mm from the
bottom, respectively. In cases where 0 > 0.15, velocity profiles were categorised as being relatively
log-normal (i.e. mid-depth velocities were substantially lower than near-bed velocities; category 1). In
cases where 0 < -0.15 in magnitude (i.e. near-bed velocities substantially exceeded mid-depth
velocities), velocity profiles were categorised as wedge-shaped (i.e. category 2).

When the

distribution of velocities was such that -0.05 < 0 < 0.05 (i.e. near-bed and mid-depth velocities were
approximately equal}, velocity profiles were categorised as transitional (i.e. category 3). When the
magnitude of the difference between mid-depth and near-bed velocities was intermediate between
categories (-0.15 < 0 < -0.05 and 0.05 < 0 < 0.15), the functional type of the profile was determined
by visually inspecting the plot of depth z against velocity u.
All statistical analyses (regressions, ANOVA, ANCOVA) were conducted using Statistica 5.0
(Statsoft, Tulsa, Oklahoma). Where the number of data points available for regression analyses were
2

2

limited, R adi (R adjusted for a small sample size) was used as an indicator of the goodness of fit.

Results
Stream bed and water surface topography
The stream bed and water surface topographies along the centre line of each of five
deployed boulders (Figures 3-3 to 3-7), three naturally occurring boulders (Figures 3-8 to 3-10), and a
gravel-bed riffle (Figure 3-11) are shown below. While the surface/bed contours collected over the
boulders were quite variable in shape, they shared several common features. The elevation in water
surface associated with deceleration of flow (i.e. hydraulic drop) at a stagnation point was seen at the
upstream margin (i.e. front) of each of the stones, although the magnitude of the hydraulic drop was
limited in several cases, and influenced by phenomena upstream. Subsequent decreases in water
depth associated with accelerating and partially deflected flow were seen downstream along the
upper surface of the boulders (Figures 3-6, 3-7, 3-8). Increases in water surface elevation associated
with a hydraulic jump were also visible downstream of the boulders (Figures 3-3, 3-4). In addition to
the downstream patterns noted in the centreline transects, as the height of streambed boulders
approached the depth of the water (i.e. relative submergences decreased), water was forced to flow
around the bed elements rather than over them (i.e. flow became increasingly three-dimensional).

106

The flow around stone 7 provides a representative example of these flow features, as the
stream bed was composed of gravels and cobbles of substantially smaller diameter than the
deployed boulder (see Figure 3-3).

The changes in water surface elevation associated with the

hydraulic drop, acceleration, and re-expansion in the hydraulic jump were seen over the front, upper
surface, and rear of the stone, respectively. Similarly, stone 31 was located in a relatively smooth
section of the study reach, and flow features similar to those noted above were also visible (Figure 34). The same flow features were associated with stone 23 (Figure 3-6), although the surrounding
substrate was composed of coarser bed material. Unlike the other deployed substrates, stone 25
was located below an 80 mm cascade produced by water flowing over an upstream obstruction
composed of boulders and cobbles (Figure 3-5). As a result, a substantial portion of the flow plunged
onto the front edge of the stone. The angle of the approaching flow and the proximity of the stone
immediately upstream interacted to create a zone of recirculating, interactive flow between the two
stones. Of the three naturally positioned boulders, N01 was located in a high velocity location (Figure
3-8), and provided little resistance to the flow due to bed material piled up against its leading edge.
Boulder N02 was more similar to stone 7 in that it protruded into the flow, and the slope of the
approaching flow was relatively flat (Figure 3-9). N03 was located in an exposed location on the
stream bed, and had a substantial change in water surface slope over its surface (Figure 3-1 0).
Compared to flow above the boulders, relatively little variation in water surface or bed
elevation was recorded above the gravel riffle (GR; Figure 3-11 ).

Velocity profiles
The vertical distribution of velocity and the near-bed flow regimes over the upper surface of
the torrential stream boulders were characterised through the measurement of 14 velocity profiles
over the five deployed boulders and three profiles over the upper surface of three naturally occurring
boulders. These profiles were compared to a profile measured over a gravel riffle and three profiles
obtained at various velocity settings in a laboratory flume.
Four vertical velocity profiles, separated by a distance of 50 mm, were measured above the
upper surface of stone 7 (Figure 3-12). None of the profiles conformed to a logarithmic velocity
profile model (see Table 3-1 ), although the profile obtained over the rear of the stone was distorted to
a lesser extent than the profiles measured over the front of the stone. For purposes of comparison,
107

two additional velocity profiles were obtained over the centre of stone 7 during periods of elevated
stream discharge (Aug . 7 and Sept. 30). The three velocity profiles obtained above stone 31 are
shown in Figure 3-13, the four velocity profiles measured over the upper surface of stone 25 are
shown in Figure 3-14, and the single velocity profiles were collected over the centres of stone 23 and
stone 10 are shown in Figure 3-15. The upper surface of stone 10 extended nearly to the water
surface. As a result, much of the flow was diverted around the sides of stone 10 rather than over its
upper surface. One velocity profile was collected over the centre of the upper surface of stones N01,
N02, and N03 (Figure 3-16). A single velocity profile was also obtained above a small stone in the

gravel bed riffle site GR (Figure 3-16), located in an adjacent stream . As mentioned above, the bed
of the gravel riffle site was composed of well-sorted, relatively fine gravels (see Figure 3-11 ).
Flow immediately above the upper surfaces of the stones exhibited a variety of patterns. In
many cases, velocity profiles deviated from a semi-logarithmic velocity distribution (where velocity
increases with the logarithm of distance from the bed) in that they showed substantial near-bed
distortion, often in the form of high velocity 'wedges' (e.g. Figure 3-12).

This was especially

pronounced where the leading edge of the stone was exposed to the flow and the geometry of the
stone was curved or inclined (rather than square-edged) (Table 3-1 ). In these instances (stones 7,
31, 23, N03), flow was constricted , and near-bed velocities (Uo.oo 2m) substantially exceeded mid-

stream (U 0 .50 ; see Table 3-2) or near-surface velocities. Although measured at various points over
the upper surface of different boulders, the distorted velocity profiles shared several common
features . In each case, maximum velocities were recorded near the boundary (3 -10 mm above the
stone surface); above this depth, velocities decreased in a logarithmic manner towards the free
surface. In each case, velocities increased logarithmically from the bed to the depth at which the
maximum velocity was recorded .
Figure 3-12 shows that the velocity profile over the front ( 10 em upstream of centre) of stone
7 was 'wedge' shaped (see Table 3-3 for a summary of velocity profile shapes at all stone locations).
Velocity profiles obtained 5 em upstream of the stone's centre and stone's centre were transitional
(intermediate between distorted, wedge-shaped profiles and logarithmically distributed profiles), while
the velocity profile 5 em downstream of the stone's centre was approaching log-normal. While the
inclined, curved leading edge of stone 7 created near bed acceleration over the front of the stone, the
upper surface of the stone was generally flat. Surface flow conditions, as indicated by the Froude
108

number, also changed as flow crossed stone 7 (Table 3-2). Local Froude number, determined at the
position of the velocity measurement, increased (10 em upstream of centre Fr = 0.98, 5 em
downstream of centre Fr = 1.82) and became supercritical (i.e. Fr > 1) as flow crossed the boulder.
1

The mean velocity also increased along the downstream transect, from U0 .50 = 0.713 m s· to Uo.so =
0.950 m s-1 . The height above the substrate at which maximum velocities were found increased in
the downstream direction above stone 7 as well (see Figure 3-12) . The boundary layer thickness o
(estimated as the height above the substrate at which velocities attain 90% of maximum) was
approximately 3.5 mm at the front of the stone, and approximately 5 mm at the most downstream
point measured (5 em downstream of centre). However, assigning a precise value of o to these
locations above the boundary is difficult, due to the non-logarithmic nature of the velocity distribution.

Table 3-1. Stone surface geometry and velocity profile shape categories for all
sampled stone surface locations
Stone

Stone curvature

7
31
25
23
10
N01
N02
N03
GR (riffle)

Leading edgec
Entire
Littled
Entire
Leading edge
Little
Little
Entire•
Little

Profile location• I Profile shape0
+100 mm
+50 mm
Wedge
Transitional
Transitional
Wedge

0 (centre)
Transitional
Wedge
Wedge
Wedge
Log-normal
Transitional
Log-normal
Wedge
Log-normal

-50mm
Log-normal
Wedge
Wedge

a locations at which velocity profiles were obtained

are recorded with respect to the centre of the
stone (e.g. +100 mm denotes that the profile was obtained 100 mm upstream of the stone's centre, 0
denotes the centre of the stone)
b see text for definition
c curvature restricted to the leading edge of the stone; the centre of the stone was relatively flat
d the upper surface of stone 25 was relatively flat; however, the stone was inclined slightly in the
upstream direction (see Figure 3-5)
e the upper surface of stone N03 was relatively flat; however, the stone was inclined steeply in the
upstream direction (see Figure 3-1 0)
The leading edge of stone 31 (Figure 3-4) was continuously curved, unlike stone 7. As the
approaching flow crossed the front of stone 31, the rising bed elevation forced the flow near the bed
to accelerate (Figure 3-14). As flow crossed the stone it became supercritical (Fr = 0.78, and Fr =
1.67, 5 em upstream and 5 em downstream of the centre of the stone, respectively), mean velocities
1

increased (Uo.so = 0.575 m s· to U0 .50 = 0.925 m s·\ and near-bed velocities showed a marked

109

increase as well {Table 3-2). The near-bed high-velocity wedge was still present above the rear of
the stone, suggesting that the velocity distortion may have projected off the rear of the stone.

Table 3-2. Summary of flow conditions for all sampled stone surface locations
Stone

Date

Location on stone •

Uo.5D (m s· )

Uo.ooZm (m s·)

Rex

Fr

Tw.PST

7

Aug. 31

+100 mm

0.713

0.808

47,500

0.98

2.82

7

Aug. 31

+50mm

0.825

0.803

72,800

1.35

2.06

7

Aug. 31

0 (centre)

0.922

0.827

101 ,000

1.66

2.65

7

Aug. 31

-50mm

0.950

0.720

110,000

1.82

2.17

7

Sept. 30

0 (centre)

0.870

0.738

90,000

0.96

2.24

7

Oct. 7

0 (centre)

1.260

0.783

95,400

0.77

2.51

31

Sept. 2

+50mm

0.575

0.580

23,200

0.78

1.61

31

Sept. 2

0 (centre)

0.747

0.796

56,700

1.13

2.82

31

Sept. 2

-50 mm

0.925

1.010

103,000

1.67

3.88

25

Sept. 8

+50mm

0.646

0.938

49,800

0.87

5.77

25

Sept. 8

0 (centre)

0.418

0.982

81,000

0.60

3.52

25

Sept. 8

-50mm

0.550

0.770

89,500

0.91

2.83

23

Sept. 30

0 (centre)

0.527

0.598

46,000

0.97

1.50

10

Sept. 23

0 (centre)

0.610

0.519

51 ,300

1.28

0.761

N01

Sept. 26

0 (centre)

1.360

1.230

127,000

1.50

6.27

N02

Sept. 27

0 (centre)

0.700

0.420

36,800

0.95

0.782

N03

Aug . 7

Rear edge

0.958

1.183

88,700

1.45

6.34

Gravel riffle

Aug. 18

0.429

0.228

2130

0.41

0.217

Flume 1

July 14

Centre
1.3 m downst.c

0.465

0.350

284,000

0.60

0.536

Flume 2

July 14

1.3 m downst.c

0.382

0.290

236,000

0.41

0.339

Flume 3

July 14

1.3 m downst.c

0.242

0.191

155,000

0.32

0.127

a locations at which velocity profiles were obtained are recorded with respect to the centre of the
stone (e.g. +100 mm denotes that the profile was obtained 100 mm upstream of the stone's centre, 0
denotes the centre of the stone)
b the distance from the leading edge (at the stagnation point) of the stone was used as the length
scale to calculate local Reynolds number (Rex); in the case of the lab flume, the distance from the
mouth of the flume {1.3 m) was used
c measurements taken 1.3 m downstream of the mouth of the flume

The upper surface of stone 25 was inclined upwards in the downstream direction (see Figure
3-5). The three velocity profiles obtained over stone 25 were distinctly wedge-shaped, and revealed
high near-bed velocities (Figure 3-14 ). However, there was no consistent increase or decrease in
either near-bed (Uo.oo 2m} or mean {U 050 ) velocities as flow crossed the upper surface of the stone
(Table 3-2}, possibly due to the highly three dimensional, complex flow patterns generated by the
plunging flow . The inclined upper surface of stone 25 likely influenced both rates of acceleration and
distributions of velocity.
Stone 23 (Figure 3-6) had a relatively curved upper surface, and continuous flow constriction
over the front half of the stone produced a wedge-shaped velocity profile over the centre of the stone
110

(Figure 3-15). The shape of this profile was similar to the profiles obtained over stone 31 and the front
of stone 7. Maximum velocity was recorded 2.5 mm above the substrate, and velocities decreased
logarithmically above this point. The current velocity 2 mm above the bed was greater (Uo.oo 2 m =0.60
m s" 1 ) than the mean velocity (Uo.so = 0.53 m s·\

Flow over the centre of the stone was

approximately critical (Fr = 0.97).

Table 3-3.

Velocity profile shapes and regression parameters used in the

estimation of U* and 'tw (velocity gradient method)

Aug. 31

Location on
stone•
+100 mm

Velocity
profile shapeb
Wedge

Aug. 31

+50mm

Transit.

7

Aug. 31

0 (centre)

Transit.

Regression coefficients
n
slope(lAc
R'ad}
0.0607
4
90.90
4
98.00
0.131
4
98.03
0.0915

7

Aug. 31

-50mm

Log-nonn.

5

97.53

0.139

3.10

159.00

7

Sept. 30

0 (centre)

Transit.

4

99.61

0.0835

1.11

776.62

7

Oct. 7

0 (centre)

Log-nonn .

7

97.65

0.111

1.96

250.82

31

Sept. 2

+50mm

Transit.

5

94.53

0.0450

0.323

70.14

31

Sept. 2

0 (centre)

Wedge

3

52.07

0.0596

0.568

3.17

31

Sept. 2

-50mm

Wedge

3

96.98

0.110

1.94

65.38

25

Sept. 8

+50mm

Wedge

6

98.05

0.0956

1.46

251.87

25

Sept. 8

0 (centre)

Wedge

4

93.13

0.104

2.33

41 .66

25

Sept. 8

-50mm

Wedge

3

97.37

0.0413

0.273

75.30

23

Sept. 30

0 (centre)

Wedge

3

94.42

0.0657

0.691

34.86

Stone

date

7
7

f w-dU/dlnz

F

AN OVA
p-value

10

Sept. 23

0 (centre)

Log-nonn.

3

99.09

0.130

2.62

218.90

N01

Sept. 26

0 (centre)

Transit.

3

99.84

0.0770

0.948

1259.10

N02

Sept. 27

0 (centre)

Log-norm.

6

92.64

0.0978

1.53

64.00

N03

Aug. 7

Rear edge

Wedge

3

86.82

0.0362

0.210

14.18

Gravel
riffle
Flume 1

Aug. 18

Centre

Log-nonn.

5

96.47

0.0416

0.277

110.26

0.031
0.0067
0.0066
0.0011
0.0013
< 0.001
0.0036
0.33
0.078
< 0.001
0.023
0.073
0. 11
0.043
0.018
0.0013
0.17
0.0018

July 14

1.3 m downst.d

Log-nonn.

7

98.73

0.0591

0.559

468.14

< 0.001

Flume 2

July 14

1.3 m downst.d

Log-nonn.

7

96.53

0.0453

0.328

168.14

< 0.001

Flume 3

July 14

1.3 m downst.d

Log-nonn.

7

97.54

0.0418

0.280

198.46

< 0.001

0.589

30.95

2.72

147.32

1.34

150.60

a locations at which

velocity profiles were obtained are recorded with respect to the centre of the
stone (e.g. +100 mm denotes that the profile was obtained 100 mm upstream of the stone's centre, 0
denotes the centre of the stone)
b see text for definition
c U* is calculated as the product of the slope (13) and von Karman's constant (K = 0.4) (see Appendix 2
for a complete discussion)
d measurements taken 1.3 m downstream of the mouth of the flume

In contrast, the velocity profile obtained over the centre of stone 10 (Figures 3-1 0 and 3-15)
was approximately logarithmically shaped when compared to the profiles obtained above stones 7,
31, 23, and 25.

However, the water depth above this boulder was the lowest of the substrates

included in the study (D

=23 mm). Flow over the centre of the stone was supercritical (Fr = 1.28),
111

and mean velocity was Ua.so = 0.61 m s·1. Current speed at 2 mm above the centre of stone 10 was
0.519 m s·1, and 8 , 1.5 mm {Table 3-2).
There was little flow constriction over the upper surface of stone N01 (Figure 3-16); bed
material deposited against the front of the stone limited the curvature at the leading edge.

As

expected, the velocity profile over the centre of the stone showed little evidence of near-bed velocity
distortion.

Near-bed velocities were high (Uo.oo 2m = 1.23 m s·\

and the boundary layer was

correspondingly thin (8 , 1.5 mm). The flow over the centre of stone N01 was supercritical (Fr = 1.50),
1

and mean velocity was 1.36 m s· (Table 3-2).
Stone N02 (Figure 3-9), although similar in exposure to flow and size to stone 7, had different
geometry. The front of stone had an angular leading edge, where the front of stone 7 was rounded.
Consequently, unlike the profile obtained over the centre of stone 7, the profile obtained over the
centre of N02 was more logarithmically distributed, and showed no evidence of a near-bed high
velocity wedge. The flow over N02 was nearly critical (Fr

=0.95), and the mean velocity was Ua.so =

1

0.70 m s· . The boundary layer thickness over the centre of N02 was 15 mm, and Ua.ao 2 m= 0.42 m s·1
{Table 3-2).
The upper surface of stone N03 was inclined in the downstream direction. The inclined
surface of the stone continually constricted the flow as it crossed the upper surface of the stone. The
single velocity profile collected over the apex of N03 {the rear of the stone) showed that the velocity
distribution was highly wedge-shaped. The boundary layer thickness at this point was very thin (8 < 1
mm), current speed at 2 mm height was 1.18 m s·\ and flow over the rear of stone N03 was
supercritical (Fr = 1.45) {Table 3-2).

Bed and water surface slope
The slope of the water's surface (S 5 ) changed dramatically as it passed over each stone (see
Figures 3-3 to 3-10). High magnitude, positive rates of changes of water surface gradient in the
1

downstream direction {dS 5 / dx > 4.0% cm- ) indicated that the water surface had a strongly convex
shape, while low magnitude, positive rates of change of water surface gradient (0 < dS5 I dx < 4.0 %
1

cm- ) indicated a weakly convex free surface shape. Negative rates of change of the gradient of the
water surface indicate a concave water surface topography. Although the magnitude of the rate of

112

slope change was variable, water surfaces were generally convex over the boulders {Table 3-4). The
water surface over stones 23, 25, and 31 was typically convex as it accelerated over the stone, wh ile
the free surface over stone 7 was concave.

Table 3-4. Water surface gradient (Ss) parameters for the five deployed and
two natural stones
Stone

dS 5 / dx (% em· )

Number of measurement
p-value
points (n)
7
-2.50
64.4
13
< 0.001
31
4.64
86.8
16
< 0.001
25
4.97
69.2
18
< 0.001
23
4.07
92.3
8
< 0.001
10
2.35
93.9
11
< 0.001
N01
0.83
47.7
12
0.0053
-0.87
15.9
7
0.18
N02 (front)•
N02 (rear)•
7.50
98.1
4
< 0.001
a the water surface shape appeared to be segmented over stone N02 (i.e. the front and rear of the
stone had different water surface shapes; see Figure 3-9)
~

(%)

Shear stress measurements from velocity gradients
Shear stress estimates were made using the velocity profile data (from determinations of U*
based on the slope of dU I dlnz; see Table 3-3) for the deployed and naturally occurring boulders,
gravel riffle site, and laboratory flume . In several cases , regression analyses were conducted using a
limited number of points, as the logarithmic portion of the velocity profile was generally very thin
(ranging from 2-7 mm for the boulders). Generally, the logarithmic portion of the velocity profile was
thinner for wedge-shaped velocity profiles; as such , fewer data points were available for these
regression analyses. Estimates of shear velocity (U*) made using velocity profile data were greater in
locations where velocity profiles were log-normal (n = 5, mean = 0.044, SO = 0.017) than in locations
where velocity profiles were transitional (n = 5, mean = 0.034, SO = 0.012) or wedge-shaped (n = 8,
mean= 0.030, SO= 0.013) .

Shear stress measurements from the Preston-static tube (PST)
The Preston-static tube (PST) was used to directly measure the distribution of shear stress
on the upper surface of each of the five deployed stones (Figures 3-17 to 3-21) and two of the

113

naturally occurring stones (N01 and N02; Figures 3-22 and 3-23). Similar shear stress distributions
were observed in all five cases where the front face of the stone was exposed to the flow, providing
substantial flow resistance (stones 7, 31, 23, 10, and N02). Generally, shear stress was relatively low
over the front half of stones, but gradually increased, often to a maximum value, over the downstream
half of each stone (Figures 3-17, 3-18, 3-20, 3-21, and 3-23, respectively).

Shear stress was

generally highest towards the rear of the stones, as water accelerated over the downstream edge. In
two cases (stones 7 and 31 ), a sharp decrease in shear stress was recorded over the rear edge of
the stone (see Figures 3-17 and 3-18, respectively). In both cases, flow visualisation confirmed that
the shear stress measurements were obtained in or near the zone of flow separation .

Flow

visualisation also indicated that flow separation occurred at the rear of all stones examined in the
study; however, the curvature at the periphery of the remainder of the stones prevented deployment
of the PST near the zone of boundary layer separation . Some stones showed substantial lateral
variation in wall shear stress (e.g. stones 31, 25, 10, N01 ). In several cases, the shear stress was
lowest over the centre of the stone, and increased laterally to one or both sides.
To varying degrees, the shear stress distributions recorded on stones 25 and N01 deviated
from the patterns described above. Although low over the front of the stone, recorded values of -rw
were uniformly high over the entire upper surface of stone 25 (Figure 3-19). Stone 25 was located
beneath the 80 mm cascade; a jet of water that impinged on the left side (facing upstream) of the
upper surface of the stone may have contributed to the high shear in that area. In addition, flow
visualisation revealed that the complex pattern of flow produced by the plunging flow upstream of the
stone created a zone of separated, recirculating flow at the front of the stone.
There was only a limited increase in shear stress from the front to the back of stone N01
(Figure 3-22).

The shear stresses and velocities measured over stone N01 were the highest

measured in the study. Very little of the front edge of the stone protruded into the flow due to gravels
piled up against the leading edge.
Stone 7 exhibited unexpectedly high shear stress at the leading edge of the stone (Figure 317), possibly due to flow impingement in the region at the front of the stone where acceleration of the
near-bed flow due to constriction is initiated. The highest shear stresses measured on stones 31 and
10 were located at the lateral peripheries, rather than the midlines, of each boulder.

114

Flow

visualisation indicated that the flow separated at the front of the boulder and accelerated around the
periphery of the stone rather than flowing over the top.

Comparison of shear stress measurements made using regression (dU I dlnz) and
Preston-static tube (PST) methods
Measurements of tw-Psr (i.e. based on the Preston-static tube) were plotted against
simultaneously determined values of two{jU/dlnz (i.e. based on velocity gradient method using the slope
of dU I dlnz) for comparison (Figure 3-24). The paired determinations of shear stress (tw) included in
the analysis were obtained over the upper surface of the stream boulders (n = 17), over a gravel riffle
(n = 1), and in a laboratory flume (n = 3). When all tw-Psr and two{jU/dlnz measurements were pooled ,
there was no significant relationship between measurements of shear stress obtained using the two
methods (ANOVA, F(1 ,19)

= 0.064, p = 0.80) (Figure 3-24). If the two methods were completely

equivalent, direct measurements of shear stress using the PST would not only be related to estimates
of shear stress using the slope of dU I dlnz, but would be numerically similar as well. However, the
relationship was significant only when the data were segregated by the shape of the velocity profile
{log-normal, transitional, and wedge-shaped).

Where velocity profiles were log-normal, the

relationship between tw-PST and two{jU/dlnz was significant (ANOVA, F(1 ,6) = 7.94, p = 0.030) (Figure 325c).

Moreover, the velocity gradient (U*) and direct (PST) methods of measuring tw were

comparable in cases where the velocity profiles were log-normal (i.e. the 1:1 line of equality lies within
the confidence intervals of the regression of •w-Psr on 1wo{jU/dlnz). However, there was no significant
relationship between direct measurement using the PST and estimates using U* (extracted from the
slope of dU I dlnz) for either wedge-shaped profiles (ANOVA, F(1 ,6) = 0.23, p = 0.85) or transitional
profiles (ANOVA, F(1 ,3) = 0.070 , p = 0.81) (Figures 3-25a and 3-25b).

When velocity profiles

displayed near-bed distortion (wedge-shaped profile cases), estimates of shear stress using the slope
of dU I dlnz tended to underestimate shear stress when compared to direct measurements of wall
shear using the PST.

115

Comparison of shear velocities (U*PsT) mean velocities (Uo.so)
Shear velocity (U*PsT) ranged between 3.43 and 14.19% of mean velocity (U 0 .50 ) at all points
at which profiles were taken (see Table 3-2). At points where the measured velocity profile was lognormal , U* was 3.43 - 4.98% of U0 .50 . U* ranged between 5.44 - 6.97% of U0 .50 where the velocity
distribution was transitional, and from 6.73 - 14.19% of U0 .50 where velocity profiles were wedgeshaped. U* was 4.65-4.98% of U0 _50 when velocity profiles were measured in the lab flume, and U*
was 3.42% of Ua.so at the gravel riffle site.

Comparison of velocity profiles
In order to compare the data from different stones, the velocity profiles were normalised by
the shear velocity U*PsT and the total water depth D (i.e. U/U*, z/D; Figure 3-26). With the exception
of the three atypical (impinging flow) profiles obtained above stone 25, the wedge shaped profiles are
generally characterised by lower values of Ux/U*PsT. while log-normal profiles are characterised by
higher values.
Velocity profiles were also normalised using the velocity at mid-depth (U 0 .50 ) (Figure 3-27). In
this case, logarithmically distributed velocity profiles were located at the lower end of the range of
Ux/U 0 _50 , while wedge-shaped profiles were located at the upper end of the range, demonstrating that,
in many cases, velocities near the bed substantially exceeded velocities at mid-depth . Where profiles
were wedge-shaped, maximum velocities were found near the bed (in all cases Zmaxu < 0.20).

Local Reynolds number, Froude number, and near-bed velocity
The distance from the stagnation point at the front of the stone (located using dye
visualisation) to the point of velocity measurement (x) was used as the length scale in determining the
local Reynolds number Rex (an indicator of boundary flow conditions) for each velocity profile
sampling location (Table 3-2). Calculations of Rex were made using near-bed velocity (Uo.oo 2m) rather
than mean velocity (U 0 .50 ) in order to describe the boundary flow conditions more accurately, given
the occurrence of wedge-shaped velocity profiles (described above).

Using these criteria, local

Reynolds number (Rex) was found to be significantly related to Froude number (Fr), (R

2

= 41 .07%;

ANOVA, F(1,16) = 12.85, p = 0.0025) (Figure 3-28). However, an examination of the bed/surface
contours indicates that three of the cases included in this model may have had atypical boundary
116

layer growth . In the first case, the plunge upstream of stone 25 created cascading flow that impinged
directly upon the upper surface of the stone. In the second case, gravels piled up at the leading edge
of the stone N01 may have prevented a stagnation point from developing at the leading edge of the
stone. In the third case, skimming flow may have developed over stone 7 when the velocity profile
was recorded on October 7, due to the increased depth resulting from rain and melting snow. When
the profiles obtained in these three circumstances are excluded from the analysis, the relationship
becomes more significant and meaningful (R

2

= 81 .19%; ANOVA, F(1,11) = 47.49, p < 0.001).

Interestingly, this relationship is stronger than the relationship between near-bed velocity Uo.oozmand
mean velocity U0 .50 (not shown).
The relationship between Froude number and shear velocity determined using the PST

(U*psr) is shown in Figure 3-29. The positive linear association between the two measurements is
significant (ANOVA, F(1,19)

= 6.27, p = 0.022). The relationship between U*duJdlnz (shear velocity

determined the velocity gradient method) and Froude number was relatively stronger (ANOVA,
F(1 , 19) = 9.57, p = 0.0060). Shear velocity (U*psr) was also significantly related to mean (Uo.so) and
near-bed (Uo.oo 2 m) velocity (Figure 3-30), although the relationship was stronger for Uo.oozm (ANOVA,
F(1 ,19) = 390.37, p < 0.001) than for U0 .50 (ANOVA, F(1,19) = 14.18, p = 0.0013).

Velocity profile shape and near-bed velocity
Sampling locations on the stream bed with characteristically logarithmic velocity profiles had
significantly lower near-bed velocities (Uo.oo 2 m) (ANOVA, F(2,18) = 10.36, p = 0.0010) than locations
with either transitional (Tukey's HSD, p =0.010) or wedge-shaped (Tukey's HSD, p =0.0014) profiles
(see Tables 3-2 and 3-3). There was no significant difference in near-bed velocities between points
with transitional or wedge-shaped profiles (Tukey's HSD, p =0.91 ). Although there was no significant
difference in mean velocity between sites characterised by the three profile types (AN OVA, F(2, 18) =
1.69, p = 0.21 ), there was a significant interaction between the near-bed and mean velocity (Rao's
R(4,34) = 10.74, p < 0.001). This indicates that unlike bed locations that were characterised by lognormal and transitional velocity profiles, mean velocities were lower than near-bed velocities in
locations that had wedge-shaped velocity profiles .
A one-way analysis of covariance (mean velocity as the covariate) showed that near-bed
velocity varied significantly with profile type (ANCOVA, F(2, 17) = 28.17, p < 0.001).
117

Near-bed

velocities were higher in locations with transitional and wedge-shaped velocity profiles than in
locations with log-normal profiles (Tukey's HSD, p < 0.001 for both tests). There was no sign ifican t
difference in near-bed velocity between locations with transitional and wedge-shaped velocity profiles
(Tukey's HSD, p = 0.71 ). Even the relatively small distortion in boundary flows found in locations with
transitional velocity profiles may restrict the application of log-normal models of velocity distribution.

Discussion
Torrential stream beds are generally composed of boulders and cobbles, with gravel filling
the interstices between the larger bed materials. The boulders, however, act as the major distributed
roughness elements, and provide the majority of the flow resistance in the channel (Bergeron 1994).
The hydraulic habitats associated with streambed boulders are inhabited by a distinct community of
benthic organisms (Bouckaert and Davis 1998), each exhibiting adaptations that allow them to thrive
in high velocity and high shear stress environments (Statzner and Holm 1982; Weissenberger et al.
1991; Vogel 1994 ). The objectives of this study were to characterise the fluid dynamic characteristics
of flow over the upper surfaces of torrential stream boulders, and to relate them to the distribution of
near-bed and boundary flow parameters (e.g. 'tw, U*). The results of this study reveal that distinct
patterns of shear stress, shear velocity, near-bed and mid-depth velocity are present over the upper
surface of stream boulders, despite the complexity of the flow. Importantly, these observations differ
from predictions made using open-channel hydraulic theory {White 1999). The near-bed patterns of
flow recorded in this study also differ from the results of previous studies that have investigated
channels with different characteristics, using different methods (e.g. Jarrett 1990; Hart et al. 1996;
Bouckaert and Davis 1998).

Velocity profile shape: deviations from log-normal
This study demonstrates that velocity gradients above the upper surface of torrential stream
boulders are heterogeneous, and regularly deviate from a semi-logarithmic shape. In many cases,
water velocities increased logarithmically downward from the water surface, reaching a maximum
only a few millimetres above the stream bed (i.e. stone surface).

Below this point, velocities

decreased rapidly in a logarithmic fashion towards the boundary. In these instances, the velocity
gradients immediately above the stones were extremely steep, given the "no-slip" condition, and that
118

1

the measured velocities 0.5 mm above the stone surface reached as high as 1.1 m s· and regularly
1

exceeded 0.60 m s· • These distinct velocity distributions are referred to as "wedge-shaped" profiles ,
in order to distinguish them from the "S-shaped" profiles recorded elsewhere in high gradient natural
channels (Jarrett 1990; Bathurst 1994). While the majority of velocity profiles recorded here showed
some evidence of this type of near-bed velocity profile distortion, several profiles were approximately
log-normally distributed.

Log-normal profiles generally occurred where there was sufficient

downstream distance for boundary layer development, which suggests that bed geometry and flow
interact to create highly diverse and spatially heterogeneous near-bed flow conditions.
The wedge-shaped profiles recorded over stream boulders can be attributed to an uneven
acceleration of the water column . In many cases, the leading edge of stones presented an inclined
face to the flow, and, as noted by Hart et al. (1996), flow accelerated up the face of the stone. Where
flow was substantially constricted as it crossed the stone (i.e. relative submergences were low and
the front of the stone was inclined or curved) the water did not accelerate uniformly. Rather, flow
near the bed accelerated at a greater rate than flow near the free surface, creating a jet-like structure
in the near-bed flow (Figure 3-31 ). Velocity profiles obtained along centreline transects on stones 7
and 31 suggest that a high velocity "wedge" (a near-bed velocity profile distortion characterised by
high velocities at a point immediately above the bed , above and below which velocities decline) was
created at the leading edge of the stone where the rising bed elevation constricted the flow (e.g., see
Figures 3-3 and 3-4). As the upper surface of stone 31 was continuously curved, near-bed flows
continued to be constricted, and as a result, the high velocity wedge increased in magnitude as flow
crossed the upper surface of the stone. The curvature of stone 7 was restricted to the leading edge,
and as a result, near-bed flows decelerated over the stone with boundary layer development, and the
high-velocity wedge was dissipated by momentum transfer.

Flow near the water surface was

relatively unaffected by the constriction, and accelerated in the downstream direction due primarily to
gravity.
Where velocity profiles were wedge-shaped above stones 7 and 31, the core of the jet-like
velocity structure (i.e. the 'tip' of the wedge) was located approximately 3 to 5 mm above the surface
of the stone. Dissipation of the jet (or wedge) may have occurred when boundary layer thickness
exceeded this height (assuming that boundary layer growth contributes substantially to viscous
damping of the near-bed high velocity wedge). Based on the local Reynolds number Rex (using near119

bed velocity U0 _002m as the velocity term) and an estimated critical boundary layer thickness (8cm; in
this case the height above the substrate to which the boundary layer would have to develop to
completely envelop the near-bed, high-velocity wedge) of 5 mm, the distance downstream (x) of the
stagnation point required dissipate the wedge was approximately 164 mm above stone 7, and 162
mm above 31. The distance downstream (x) required for the development of a boundary layer of a
given thickness is given by

(3-11)

where Dent is critical boundary layer thickness and Rex is the local Reynolds number (White 1999).
The velocity profile measured downstream of this point on stone 7 had returned to log-normal, as
predicted. Conversely, the most downstream velocity profile measured above stone 31 was 163 mm
downstream of the stagnation point.

Contrary to what was expected, the high velocity wedge

dominated the near-bed portion of the velocity profiles over the entire stone. The continuously curved
upper surface of stone 31 may have contributed to continual flow constriction and, as a result,
continued near-bed flow acceleration.
In contrast, logarithmically distributed velocity profiles were recorded at relatively few
locations above boulders at the site. At each of these points, substantial flow constriction above the
stones may not have occurred, due to (1) the development of skimming flow (e.g. stone 7 during high
discharge); (2) very low relative submergences (stone 10); or (3) square leading edges (stone N02).
It is important to note that although velocity distributions were log-normal in each of these cases,
velocities were recorded only at the centre of each stone. Near-bed flow constriction, a factor that
leads to the development of wedge-shaped velocity profiles, may have occurred to a limited extent
above the leading edge of each stone. However, there was no evidence of near-bed flow distortion
by the time flow crossed the centre of the stone. The development of an approximately logarithmic
velocity profile could then result, in part, from the low depth and long distance from the leading edge
over which boundary layer growth may act to dissipate any near-bed velocity distortion. In addition,
due to the limited depth, water was forced to flow round, rather than over, stone 10. This type of
three-dimensional flow is poorly understood and difficult to model (see Bathurst 1994; Vogel 1994).
The centreline velocity profiles recorded over stone 7 provide evidence for this scenario.
120

The

distinctive near-bed velocity profile distortion (wedge) that developed over the front of the boulder was
subsequently lost as flow crossed the stone and boundary layer thickness increased.
Several authors investigating shallow flows in natural channels have also reported velocity
distributions that do not conform to a semi-logarithmic form.

The deviations from the typical

logarithmically distributed velocity profile have been variously attributed to flow separation (Bergeron
1994; Hart et al. 1996), skimming flow and high flow resistance near the bed (Bathurst 1994 ), and the
development of an internal boundary layer downstream of flow reattachment (Bergeron 1994).
However, the velocity distributions reported in these studies conform to neither a log-normal
distribution nor a wedge-shaped velocity profile. In these studies, maximum velocities were generally
recorded near the free surface, rather than near the bed . Rather than conflict with our results, the
results of Bathurst (1994), Bergeron (1994), and Hart et al. (1996) illustrate the highly heterogeneous
nature of flow fields present in the boundary layer flows of shallow streams with complex bed
geometry (i.e. rough-turbulent flows). Flow constriction over and between boulders, flow deflection by
obstacles, and boundary layer separation and reattachment will all contribute to velocity profiles that
vary substantially over short distance. In part, the highly heterogeneous near-bed flow environment
creates the wide array of "hydraulic niches" available to stream benthic organisms (see chapters 1,2,
and 4).
The S-shaped velocity profile recorded elsewhere in high gradient mountain rivers (Jarrett
1990; also see Bathurst 1994) are characterised by velocities which are lower nearer the stream bed
and greater near the water surface than a logarithmically distributed velocity distribution. In these
flow conditions, the logarithmic profile fails to develop due to drag induced by the coarse bed
material. Although both the present study and the work of Jarrett (1984; 1990; 1993) were conducted
in mountain channels with cobble and boulder bed material, vertical velocity profiles were not found to
be S-shaped here.

In fact, the wedge-shaped profiles recorded in the present study are

approximately the inverse of S-shaped profiles.

Whereas logarithmically distributed profiles

overestimate near-bed velocities and underestimate near-surface velocities in S-shaped profile flow
conditions (Jarrett 1990), they underestimate near-bed velocities and overestimate near-surface
velocities in cases where wedge-shaped profiles characterise the flow. The different findings (Sshaped vs. wedge-shaped profiles) can likely be attributed to two factors.

(1) The relative

submergence of the bed material was greater in the channels studied by Jarrett (1990); as relative
121

depths increased, skimming flow developed, and the flow constriction required to accelerate near-bed
flows did not occur.

Boundary layer development is inhibited by limited flow depth in shallow

channels (Nowell and Church 1979). (2) Previous studies, relying heavily on electromagnetic current
meters and propeller velocimeters, have not measured flows immediately above the bed with the
spatial resolution required to detect any distortions of the velocity profiles. Due to the complex bed
topography of mountain streams, irregular pressure gradients will be common (White 1999}, and
velocity gradients will be correspond ingly complex and difficult to predict.
Bathurst ( 1994) suggested that velocity gradients in mountain rivers would adopt a semilogarithmic shape only at relatively high relative submergences (D/dk > 10), which is consistent with
the depth proposed by Nowell and Church (1979) for the development of log-normal velocity
distributions (D/dk > 8). The results of th is study provide evidence that the development of semilogarithmic velocity profiles may also be dependent on bed morphology. At the gravel bed site,
velocities were log-normally distributed (D/dk = 11 .24), which conforms to the criteria of Bathurst
(1994). Log-normal velocity profiles were also recorded above stone N01 at D/dk = 0.60 and above
stone 7 at D/dk = 1.60 (when the stream was at near-flood discharge). However, gravels piled up
against the leading edge of stone N01 increased the relative submergence of the stone to
approximately D/dk =2.09. Also, because S-shaped velocity profiles develop as a result of form drag,
velocity profiles above substrates that protrude farthest into the flow under skimming flow conditions
may be relatively log-normal. Substrate size can also influence the shape of the velocity profile; lognormal profiles may develop over large boulders at relatively low values of D/dk due to the relatively
long distances available for boundary layer development. These results indicate that further study is
required to examine the complexity of rough boundary layer flows .

The influence of bed geometry on near-bed and surface flow factors
The results show that the relationship between shear velocity (U*) and mid-depth velocity

(Uo.so) is dependent on the geometry of the channel bed. In shallow, "chaotic" flows (Young 1992},
where the depth of the channel is relatively shallow with respect to the height of the substrate
elements and flows are highly three-dimensional and difficult to predict, it may be not be possible to
estimate near-bed flow parameters based on measurements of average water velocity. In flow over a
gravel riffle, U* was 3.42% of velocity at mid-depth . This agrees with the findings of lppen (1966),
122

who found that U* was 3% of free-stream velocity for water flow in a smooth-bottomed channel. In
shallow flow over boulders, U* ranged between 3.97 and 14.12% of mid-stream velocities. This was
consistent with results reported by Hart et al. (1996), who found that U* was approximately 10% of
velocities measured 10 mm above stream boulders, and Denny and Shibata (1989), who reported
that U* was between 5 and 15% of mean velocity in flows over rough bottoms. However, the results
of the present study indicate that the relationship between shear velocity and free-stream velocity was
also dependent on the shape of the velocity profile.

Where near-bed acceleration distorted the

velocity profile over boulders in shallow flows, U* was proportionally greater (6. 74 - 14.19% of U0 .50 )
than in instances where velocity profiles were log-normal (3.97- 4.90%). This demonstrates that bed
geometry substantially influences the local variability of near-bed hydraulic conditions in shallow
streams. Where complex bed geometry creates an unpredictable relationship between mid-depth
velocity and near-bed and wall flow parameters, measurement must replace estimation until a general
theory is developed.
Due to the heterogeneous bed geometry and flow conditions at the site, it was not
unexpected that velocity measured 2 mm above the bed would be a better predictor of shear velocity
U* than mean velocity (measured at 0.5D). Given the relationship between shear velocity and wall
shear stress (•w = pU* 2 ) , velocity 2 mm above the boundary should have a correspondingly significant
relationship with wall shear, as this depth would be assumed to be in, or near, the log layer. At
several of the measurement locations, velocity profiles deviated so strongly from the logarithmic
shape that near-bed velocities (Uo.oozm) exceeded velocities at mid-depth (U 0 .50 ) . In circumstances
where boundary layer profiles vary from the typical log-normal shape, approximations of wall shear
stress based on mean velocity may fail to adequately describe the relationship between local flow and
wall shear stress . Physical instances where this type of boundary layer and velocity profile distortion
may occur include flows with wall jets, free jets, or strongly positive or adverse pressure gradients
(White, 1999).
Models that predict average velocity and discharge based on the average slope of the free
surface have been used extensively to interpret the flow environment in streams and rivers.
However, these results show that the water surface slope is highly variable over very short distances.
As flow crossed the upper surface of stream boulders, the water surface slope changed as much as
1

5% cm· . This indicates that while water surface geometry may be used to estimate reach-averaged
123

velocities (see White 1999), it is of little use in measuring velocities over shorter distances (e.g. 1 em)
in shallow, high gradient streams.

Unfortunately, it is this spatial scale that is relevant to benthic

macroinvertebrates. It is of interest, however, to note that the rate of change of the water surface
gradient is remarkably consistent over the boulders. Positive rates of change of the water surface
gradient suggest a convex surface slope shape, and thus imply supercritical approaching flow (White
1999). This conclusion may, unfortunately, represent an oversimplification of flows over boulders in
natural channels. In shallow, "chaotic", flows over coarse substrates, the shape of the free surface is
the product of the complex relationship between depth (and change in depth as flow crosses
bedforms), bed geometry, and the specific energy of the flow. Further experimentation is required to
clarify the relationship among these factors .

Wall shear stress in torrential streams
Wall shear stress is commonly estimated using a determination of the shear velocity (U*)
based on velocity profile data. Accurate estimates of wall shear stress are dependent on accurate
determinations of U*, which are, in turn, a function of the slope of the velocity gradient. As such,
errors in assessment of the thickness of the log-layer or 'overlap' layer may lead to substantial errors
in estimates of U*. A visual or statistical assessment of the linearity of the near-boundary portion of
the plot of dU I dlnz is usually used to determine the thickness of the log layer. However, in flow
environments characterised by high velocities and high wall shear stress, the log layer can be very
thin, and several measurements of velocity adjacent to the bed may be required to accurately
determine the slope of the velocity gradient.

Unfortunately, many instruments commonly used to

measure water velocities in the field (e .g. propeller velocimeters) are incapable of measuring flows at
the high spatial resolutions required . Moreover, the application of this method assumes that the
shape of the near-bed velocity gradient is predictably logarithmic. The law of the wall (see White
1999) will hold for regions immediately adjacent to the wall in all flow conditions where the bed is noneroding. However, in flows where the velocity profile is substantially distorted due to the presence of
either adverse or positive pressure gradients, the accuracy of the velocity gradient method of
estimating shear stress is not known . Our results indicate that in instances where the velocity profiles

124

were non-logarithmic {i.e. wedge-shaped or transitional}, estimates of tw based on the slope of dU I
dlnz tended to greatly underestimate wall shear stress by an average of 60.78% (SE = 9.42%).
Where relative submergences were sufficiently high, stream flows produced a distinctive
pattern of wall shear stress (tw) over the exposed surface of the boulders. Often, the highest shear
stresses were recorded on the lateral surfaces of the boulders (see Figures 3-18, 3-19, and 3-21).
Shear stress was generally low over the front of the stones, and increased towards the rear of the
stone as flow accelerated .

In addition, flow visualisation of basic flow patterns showed that

separation occurred over the rear edge of each of the stones. This was expected, as at the range of
local Reynolds numbers calculated for the deployed stones at the site (Re = 45,200 to 108,000),
boundary layers should separate at the lee side of each stone, and vortices should develop and be
shed in the turbulent wake of each stone (Vogel 1994). Abrupt decreases in shear stress accompany
flow separation (White 1999), and this pattern was observed in two of the seven stones over which
shear stress distributions were recorded . Presumably, in cases where a sudden decrease in shear
stress was not recorded, measurements of shear stress were not taken sufficiently far downstream, in
the zone of recirculating, detached flow. These distinctive shear stress distributions are potentially
important to the ecology and behaviour of benthic invertebrates, especially those taxa that inhabit the
upper surfaces of stream boulders in torrential streams {i.e. torrenticolous fauna ; Vogel 1994). For
example, grazing invertebrates that exhibit adaptations to maintain contact with the stream bed and
feed in high velocity habitats may be able to utilise high-shear stress areas of the substrate as a
refuge from predators that lack similar adaptations (e.g. Hart and Merz 1998).

Measurement of near-bed and wall flow conditions
There have been several methods used to measure near-bed flow conditions in streams .
Among these, the FST (Fiiesswasserstammtisch} hemisphere method has been used extensively to
estimate flow conditions close to substrate in lotic environments (e.g. Statzner and Muller 1989;
Dietrich and Schmedtje 1995). FST hemispheres provide estimates of near-bed "flow characteristics"
or "hydraulic stress" (Statzner and Muller 1989), parameters that integrate several near-bed flow
parameters including velocity, lift, drag, and shear stress. As such, FST hemispheres provide little
information on the relative importance of these various flow characteristics .

125

In practice, FST

hemispheres integrate flow conditions over relatively large horizontal and vertical distances (-50 mm
and 25 mm, respectively). Consequently, the use of FST hemisphere methods may be inappropriate
in investigations of small-scale phenomena, especially the fine-scale flow preferences of benthic
invertebrates in shallow streams. This is especially true where velocity gradients deviate from lognormal.

In these cases, the FST hemisphere method will likely misrepresent flow conditions

immediately adjacent to the stream bed. Another recent development involves the use of acoustic
Doppler velocimeters, which have been used to measure the Reynolds stress (i.e. free shear stress
above the bed), rather than the wall shear stress (Bouckaert and Davis 1998).

While these

measurements may be of interest to issues such as mass transfer in the water column , Reynolds
stress may be of limited importance in describing the immediate forces experienced by benthic
insects, especially where the movement of water is highly complex and unpredictable (i.e. roughturbulent boundary layer conditions).
The Preston-static tube (PST) used in this study can be used to directly measure wall shear
stress as well as measure current velocity immediately adjacent to the bed when deployed as a
Preston-static tube (Ackerman et al. 1994; Ackerman and Hoover 2001 ). As such, it is capable of
providing a description of the near-bed flow conditions experienced by benthic insects in high-velocity
flows. The vertical distance over which the PST integrates velocity is very short (0.88 mm, the model
used in this study) compared to propeller and bucket-wheel velocimeters (often> 10 mm). This high
spatial resolution allows for the characterisation of flows in the region immediately above the bed
(providing several measurements within 10 mm of the substrate), where other types of apparatus are
limited. For example, the electromagnetic current meter used by Bergeron (1994) was unable to
measure velocities closer than 19 mm from the bed . Similarly, the acoustic Doppler velocimeter,
contrary to recent published reports (e.g. Bouckaert and Davis 1998), is unable to accurately
measure velocities within centimetres of the bed (Finelli et al. 1999), and is thus unable to
characterise the near-bed flow environment. The PST represents a simple, easily deployed device
capable of measuring velocities at fine scales immediately above the bed , the region of the most
relevance and interest to benthic biologists . Future PST designs could employ pressure transducers
with faster response times and data logging to provide higher order moments of fluid dynamic
measurements. Constant temperature (hot-film) anemometry can be used to measure near-bed
velocities in streams at high spatial and temporal resolution, but is time consuming and notoriously
126

difficult to deploy in natural channels (Li 1994; Hart et al. 1996). Regardless , Hart et al. (1996) were
able to successfully obtain near-bed velocity measurements in shallow streams using these
instruments.
The results indicate that the PST provides an accurate measurement of wall shear stress
when compared to the traditional velocity gradient method (i.e. estimating shear stress from the slope
of the regression of dU I dlnz). In the rough boundary layer flow conditions present in shallow,
bouldery streams, velocities are seldom logarithmically distributed, and log-layers are often quite thin
and restricted to a region adjacent to the bed where most current meters are unable to measure
velocity. However, where boundary layers were very thin (8 < 3 mm), assessment of the thickness of
the log-layer was problematic using the PST. Where near-bed velocity distortions impinge on the
upper portion of the log-layer, estimates of shear stress using the velocity gradient method will
consistently underestimate

•w· In rough boundary layer flows similar to the ones described here, it

may not be possible to assess log-layer flow conditions using the coarse spatial resolution provided
by other velocity measurement devices.
In ideal circumstances (uniform flow with a fully developed boundary layer over a
topographically simple (flat) bed), wall shear stress measurements made using the Preston-Static
tube corresponded with those obtained by estimating U* from the slope of the plot of U vs. In z
(Figure 3-25; also see Ackerman and Hoover 2001 ). A well-developed log-layer would provide the
relatively large number of velocity measurements required for an accurate estimate of the wall shear
stress using the velocity gradient or shear velocity (U*) method (see Bergeron and Abrahams 1992;
Biron et al. 1998). However, in flows over torrential stream boulders the log-layer is seldom thick
enough to obtain an accurate estimate of U* using the velocity gradient method.

The number of

points used in the regression of U on In z were, in several cases, fewer than the suggested number
required to accurately predict the slope of a line (Zar 1984). This is, however, a physical constraint
imposed by the nature of near-bed flows over boulders in shallow streams.

Near-bed velocity

distortions often restricted the log-linear portion of the velocity profile to a few millimetres above the
bed, limiting the number of velocity measurements that could be obtained in the log-layer. However,
in instances where the velocity profile was not logarithmically distributed (i.e. wedge-shaped),
estimates of shear stress were consistently lower in magnitude than measurements of shear stress
made using the PST. This suggests that the log-layer may be thinner than indicated by the log127

transformed velocity profile. If this is the case, the PST will provide a more accurate shear stress
value than the velocity gradient method, assuming that the log-layer is thinner than the diameter of
the PST (0.88 mm). As flows are rarely logarithmically distributed in shallow torrential streams (see
Hart et al. 1996; Bergeron 1994), near-bed flow parameters should be directly measured rather than
estimated, if possible.

Conclusions
The unexpectedly high near-bed velocities recorded over the surface of stream boulders in
this study suggest that it may be necessary to re-evaluate the ecological role played by hydrodynamic
forces in shallow, high gradient, stream environments (i.e. rough-turbulent flows). The importance of
surface and bulk flow phenomena on the distribution and behaviour of benthic taxa has been
explored by several authors (e.g. Wetmore et al. 1990; Delgado and Carbonell 1997). However, it
has generally been assumed that the velocities to which invertebrates are exposed (i.e . near-bed
velocities; approximated in this case as Ua.oo 2m) could be accurately estimated using models based
upon a logarithmic distribution of velocity.

However, the results of the present study show that

velocity profiles are often distorted by near-bed acceleration over the front and upper surfaces of
large stones in shallow flows , and rarely conform to the familiar log-normal shape. As such, the
velocities that invertebrates are exposed to may, in many cases, be greater than previously thought.
Consequently, the practice of estimating near-bed flow parameters from measurements of mean
velocity may lead to the underestimation of the lift and drag forces that torrential macroinvertebrates
experience.
The results of the present study also demonstrate that near-bed flow conditions in
shallow, torrential streams are heterogeneous.

In general, near-bed velocities and wall shear

stresses appear to be greatest toward the rear (upstream of the separation point) and sides of
boulders exposed to the flow. Torrential stream macroinvertebrates that show a preference for these
areas of the stream bed might benefit from protection from predators, enhanced feeding
opportunities, or increased rates of gas exchange (e.g. Golubkhov et al. 1992; Hart and Merz 1998).
Benthic taxa that avoid these areas of the stream bed may be unable to cope with the shear, lift, and
drag forces associated with this microhabitat, or lack the adaptations necessary to maintain contact
with the substrate in high velocities.

The data presented here indicate that at present only
128

measurement can ensure an accurate characterisation of the near-bed environment experienced by
torrential stream invertebrates.

Until models that describe the relationship between the bed

geometry, channel characteristics, and flow are developed, the relationship between near-bed
hydraulics and the behaviour, morphology, and ecology of benthic taxa in high velocity habitats will
remain largely unknown for most taxa.

129

Acknowledgements

I would like to thank the members of the Aquatic Sciences Lab at UNBC, especially Noboru
Yonemitsu and Mike Nishizaki, for their technical assistance and encouragement.

I would like to

extend a special thank you to my principal field assistants; Dan Gable, Terri Fortune, and Shelley
Hoover. Also, a big thank-you goes out to those who helped me understand the basic concepts of
open-channel flow . This work was supported by funding from NSERC and FRBC to J.D. Ackerman .

130

References
Ackerman, J.D ., and T.M. Hoover. 2001 . Measurement of local bed shear stress in streams using a
Preston-static tube. Limnol. Oceanogr. 46: 2080-2087.
Ackerman, J.D., L. Wong, C. R. Ethier, D.G. Allen , and J.K. Spelt. 1994. Preston-static tubes for the
measurement of wall shear stress . J. Fluids Eng. 116: 645-649.
Bathurst, J.C. 1994. At-a-site mountain river flow resistance variation . Proc. Hydraul. Engrg . (1994),
ASCE . 1: 682-686 .
Bergeron, N.E. 1994. An analysis of flow velocity profiles, stream bed roughness, and resistance to
flow in natural gravel bed streams. Proc. Hydraul. Engrg . (1994), ASCE . 1: 692-696 .
Bergeron , N.E. , and A.D . Abrahams. 1992. Estimating shear velocity and roughness length from
velocity profiles. Water Resour. Res . 28 : 2155-2158.
Biron, P.M., S.N. Lane, A.G . Roy, K.F. Bradbook, and K.S. Richards. 1998. Sensitivity of bed shear
stress estimated from vertical velocity profiles: the problem of sampling resolution . Earth
Surf. Proc. Land. 23: 133-139.
Blizard, C.R. , and E.E. Wohl. 1998. Relationships between hydraulic variables and bedload
transport in a subalpine channel, Colorado Rocky Mountains, U.S.A. Geomorphology 22 :
359-371 .
Bouckaert, F.W., and J. Davis. 1998. Microflow regimes and the distribution of macroinvertebrates
around stream boulders . Freshwater Bioi. 40: 77-86.
Buffin-Belanger, T., and A.G . Roy. 1998. Effects of a pebble cluster on the turbulent structure of a
depth-limited flow in a gravel-bed river. Geomorphology 25: 249-267.
Carling , P.A.. 1992. The nature of the fluid boundary layer and the selection of parameters for
benthic ecology. Freshwater Bioi. 28: 273-284.
Carling, P.A., J.J. Williams, A. Kelsey, M.S. Glaister, and H.G. Orr. 1998. Coarse bedload transport
in a mountain river. Earth Surf. Proc. Land . 23: 141-157.
Davis, J. A., and L. A. Barmuta. 1989. An ecologically useful classification of mean and near-bed
flows in streams and rivers . Freshwater Bioi. 21 : 271-282 .
Delgado, J.A., and R. Carbonell. 1997. Case features of caddisfly larvae (Sericostoma se/ys1) as
related to water velocity and potential to drift. J. Freshwater Ecol. 12: 193-197.
Denny, M.W. 1994. Extreme drag forces and the survival of wind- and water-swept organisms. J.
Exp. Bioi. 194: 97-115.
Denny, M.W. and M.F. Shibata. 1989. Consequences of surf-zone turbulence for settlement and
external fertilization . Amer. Nat. 134: 859-889.
Dittrich , A., and U. Schmedtje. 1995. Indicating shear stress with FST-hemispheres - effects of
stream-bottom topography and water depth . Freshwater Bioi. 34: 107-121 .
Finelli, C.M., D.O . Hart, and D.M. Fonseca . 1999. Evaluating the spatial resolution of an acoustic
Doppler velocimeter and the consequences for measuring near-bed flows . Limnol. Oceanogr.
44:1793-1801.
Golubkhov, S.M ., T.M. Tiunova, and S.L. Kocharina . 1992. Dependence of the respiration rate of
aquatic insects upon the oxygen concentration in running and still water. Aquatic Insects 14:
137-144.
131

Hart, D.O., B.D . Clark, and A. Jasentuliyana. 1996. Fine-scale field measurements of benthic flow
environments inhabited by stream invertebrates. Limnol. Oceanogr. 41 : 297-308.
Hart, D.O., and C.M. Finelli. 1999. Physical-biological coupling in streams : the pervasive effects of
flow on benthic organisms. Ann . Rev. Ecol. Syst. 30: 363-395.
Hart, D.O., and RA. Merz. 1998. Predator-prey interactions in a benthic stream community: a field
test of flow-mediated refuges. Oecologia 114: 263-273.
lppen, AT.
1966.
Sedimentation in estuaries . pp . 648-672 . In Estuary and Coastline
Hydrodynamics. Ed . A. T. lppen. McGraw-Hill, New York, NY.
Jarrett, RD. 1984. Hydraulics of high gradient streams. J. Hydraulic Eng. 110: 1519-1539.
Jarrett, RD. 1990. Hydrologic and hydraulic research in mountain rivers. Water Res . Bull. 26 : 419429.
Jarrett, RD. 1991 . Wading measurements of vertical velocity profiles. Geomorphology 4: 243-247 .
Lacoursiere, J.O. 1992. A laboratory study of fluid flow and microhabitat selection by larvae of
Simuliium vittatum (Diptera: Simuliidae). Can . J. Zool. 70: 582-596.
Li, M.Z. 1994. Direct skin friction measurements and stress partitioning over movable sand ripples .
Water Resour. Res. 30: 791-799.
Moog, D.B., and G.H. Jirka. 1999. Stream reaeration in non-uniform flow: macroroughness
enhancement. J. Hydraulic Eng. 125: 11-16.
Nikora, V.I., D.G. Goring, and B.J .F. Biggs. 1998. On gravel-bed roughness characterization . Water
Resour. Res. 34: 517-527.
Nowell, ARM., and M. Church. 1979. Turbulent flow in a depth-limited boundary layer. J. Geophys.
Res. 84: 4816-4824.
Peckarsky, B.L., S.C. Horn, and B. Statzner. 1990. Stonefly predation along a hydraulic gradient: a
test of the harsh-benign hypothesis. Freshwater Bioi. 24 : 181-191.
Pommen, G.D.W., and D.A. Craig. 1995. Flow patterns around gills of pupal net-winged midges
(Diptera: Blephariceridae): possible implications for respiration . Can . J. Zool. 73: 373-382 .
Preston , J.H. 1954. The determination of turbulent skin friction by means of Pitot tubes. J. Royal
Aeronaut. Soc. 58: 109-121 .
Robert, A., A.G. Roy, and B. De Serres . 1996. Turbulence at a roughness transition in a depth
limited flow over a gravel bed. Geomorphology 16: 175-187.
Silvester, N.R, and M.A. Sleigh. 1985. The forces on microorganisms at surfaces in flowing water.
Freshwater Bioi. 15: 433-448 .
Statzner, B., and T.F. Holm. 1982. Morphological adaptations of benthic invertebrates to stream flow
- an old question answered by means of a new technique (laser doppler anemometry).
Oecologia 53: 290-292.
Statzner, B., and R Muller. 1989. Standard hemispheres as indicators of flow characteristics in lotic
benthos research. Freshwater Bioi. 21: 445-459.
Sukhodolov, A., M. Thiele, and H. Bungartz. 1998. Turbulence structure in a river reach with sand
bed. Water Resour. Res. 34 : 1317-1334.
132

Vogel, S. 1994. Life in Moving Fluids, 2nd Edition . Princeton University Press, Princeton, NJ .
Weissenberger, J., H.-Ch. Spatz, A. Emanns, and J. Schwoerbel. 1991. Measurement of the lift and
drag forces in the mN range experienced by benthic arthropods at flow velocities below 1.2 m
1
s· . Freshwater Bioi. 25:21-31 .
Wetmore, S.H ., R.J. Mackay, and R.W. Newbury. 1990. Characterization of the hydraulic habitat of
Brachycentrus occidentalis, a filter-feeding caddisfly. J. N. Am . Benthol. Soc. 9: 157-169.
1

White, F.M . 1999. Fluid Mechanics- 4 h Edition . WCB McGraw-Hill , Boston .
Wiley, M.J., and S.L. Kohler. 1980. Positioning changes of mayfly nymphs due to behavioral
regulation of oxygen consumption . Can. J. Zool. 58: 618-622 .
Young, W.J. 1992. Clarification of the criteria to identify near-bed flow regimes . Freshwater Bioi. 28 :
383-391.
Young, W.J . 1993. Field techniques for the classification of near-bed flow regimes . Freshwater Bioi.
29 : 377-383.
Zar, J.H. 1984. Biostatistical Analysis, 2nd Edition . Prentice-Hall, Englewood Cliffs, NJ .

133

Chapter 3 - Figures

British Columbia

the Torpy River
watershed
Prince
George

Figure 3-1. Map of British Columbia, showing the location of the Torpy River watershed.

A.

B.

g

Figure 3-2. Diagrammatic representation of the grid system and deployment of the experimentally
deployed stones. A. Side view. B. Plan view. Labels are as follows: d =datum marking the centre of
the stone, g = grid point, id = stone identification label.

134

'E 30

"'b

.....X 25

-

20

c

15

""0

-~

10

(.)

5

N

(1)
(.)

<tl

ro

t(1)
>

sz

uc::=>

- ------

_,

0
0

5

10

15 20 25

30 35 40 45

50

55 60 65 70

75

2

downstream distance X (X1 0" m)

Figure 3-3. Stream bed and water surface topography of experimentally deployed stone 7 (centreline
transect) measured on Sept. 5, 1998. The white bar delineates the extent of stone 7. The four
locations at which velocity profiles were obtained are shown. The location marked "3" denotes the
centre point (datum) of the stone.

'E 15

'":'

-

10

c
<tl
]i

5

0
.....
X
N

(1)
(.)

""0

ro

uc:::::>

(.)

t(1)
>

0
0

5

10

15

20

25

30

35

40

downstream distance X (X1 0"2m)

Figure 3-4. Stream bed and water surface topography of experimentally deployed stone 31
(centreline transect) measured on Sept. 2, 1998. The white bar delineates the extent of stone 31.
The three locations at which velocity profiles were obtained are shown . The location marked "2"
denotes the centre point (datum) of the stone.

135

E'

25

"'b
..... 20

-

15

Cll

10

X

N

a.>
c:
(.)

.!!?

"C

ro

5

a.>

0

(.)

t:
>

0

5

10

15

20

25

30

35

40

45

50

55

60

downstream distance X (x1 0-2 m)

Figure 3-5. Stream bed and water surface topography of experimentally deployed stone 25
(centreline transect) measured on Sept. 8, 1998. The white bar delineates the extent of stone 25.
The three locations at which velocity profiles were obtained are shown. The location marked "2"
denotes the centre point (datum) of the stone.

E' 20

~

0
.....

~
N

a.>
c:
(.)

Cll

.!!?
"C

~

t:

a.>

>

15

~

10
5

0
0

5

10

15

20

25

30

35

40

45

50

2

downstream distance X (X1 0- m)

Figure 3-6. Stream bed and water surface topography of experimentally deployed stone 23
(centreline transect) measured on Sept. 30, 1998. The white bar delineates the extent of stone 23.
The location at which the velocity profile was obtained is shown . The location marked "1" denotes the
centre point (datum) of the stone .

136

E 25

"'b

\?

...... 20

~
N

Q)

~

15

0

c:

ro 10

.!a
"C

ro0

t:
Q)
>

5
0
0

5

10

15

20

25

30

35

40

45

50

55

60

2

downstream distance X (X1 0- m)

Figure 3-7. Stream bed and water surface topography of experimentally deployed stone 10
(centreline transect) measured on Sept. 30, 1998. The white bar delineates the extent of stone 10.
The location at wh ich the velocity profile was obtained is shown. The location marked "1" denotes the
centre point (datum) of the stone.

"'

30
E
b 25
......
~
N

Q)

20

0

15

( /)

10

c:

ro

:0

ro0

t:
Q)
>

uc::::::>

5
0
0

5

10

15

20

25

30

35

40

45

50

55

60

65

downstream distance X (X1 0-2 m)

Figure 3-8. Stream bed and water surface topography of naturally positioned stone N01 (centreline
transect) measured on Sept. 26, 1998. The white bar delineates the extent of stone N01. The
location at which the velocity profile was obtained is shown . The location marked "1" denotes the
approximate centre of the stone.

137

E

30

~

20

"'0 25
....
N

Q)
(.)

r:::

~

15

Cll

.!a 10
"0

-ro

5

>

0

(.)

tQ)

0

5

10

15

20

25

30

35

40

45

50

55

60

65

2

downstream distance X (X1 0- m)

Figure 3-9. Stream bed and water surface topography of naturally positioned stone N02 (centreline
transect) measured on Sept. 27, 1998. The white bar delineates the extent of stone N02. The
location at which the velocity profile was obtained is shown. The location marked "1" denotes the
approximate centre of the stone.

E
....

"'0

~
N

uc::::>

10

~

r:::

Cll

~

"0

5

ro
(.)

t

~

0
-5

0

5

10

15

20

25

downstream distance X (X1 0-2 m)

Figure 3-10. Stream bed and water surface topography of naturally positioned stone N03 (centreline
transect) measured on Aug . 7, 1998. The white bar delineates the extent of stone N03. The location
at which the velocity profile was obtained is shown.

138

E

-

15

~

0

..-

><

N

~

10

Q)

(.)

c

ro

.!a

5

"0

ro
:e
(.)

Q)

>

0
-10

-5

0

5

10

15

20

25

2

downstream distance X (X1 0- m)

Figure 3-11. Stream bed and water surface topography of the gravel riffle site (GR) measured on
Aug. 17, 1998. The white bar delineates the diameter of the stone above which the profile was
measured. The location marked "1" is the point at which the velocity profile was obtained.

139

---

0.00
0.55

0.60

0.65

~~~~~
0.70

0.75

0.80

0.85

----- 0.90

0.95

1.00

1

velocity U (m s- )

Figure 3-12. Velocity profiles obtained along the centreline of stone 7. The horizontal line above
each velocity profile represents the water depth at that location . Mid-depth and maximum velocities
increase as flow crosses stone 7. Wedge-shaped, transitional, and log-normal velocity profiles are
labelled "w-s", "tr", and "1-n", respectively. Velocity profiles measured over the centre of the stone, 5
em upstream of centre, 10 em upstream of centre, and 5 em downstream of centre are labelled "0",
"1", "2", and "-1 ", respectively . Measurement points correspond to the marked grid locations . The
velocity profile at the front of the stone is wedge shaped ; the near-bed flow distortion becomes less
evident as flow crosses the stone .

140

0.06

--

+1 (tr)

E 0.05

N

0 (w-s)

~
ca 0.04

"'C

c::

:::1

0
..0

E
0
....
Q)

(.)

c::

ca
.....

•!:Q
"'C

0.03

-1 (tr)

0.02
0.01
0.00
0.4

0.5

0.6

0.7

0.8

0.9

1.0

1.1

velocity U (m s·1 )

Figure 3-13. Velocity profiles obtained along the centreline of stone 31. The horizontal line above
each velocity profile represents the water depth at that location. Mid-depth and maximum velocities
increase as flow crosses stone 31. Wedge-shaped and transitional velocity profiles are labelled "w-s"
and "tr'', respectively. Measurement points correspond to the marked grid locations. Velocity profiles
measured over the centre of the stone, 5 em upstream of centre, and 5 em downstream of centre are
labelled "0", "1", and "-1", respectively.

141

--

0.06
+1 (w-s)

E 0.05

0 (w-s)

N

~

n3

"'C

0.04

c

::l

0
..0

E

0
.....
Q.)

0

c

0.03
0.02

n3

-~

"'C

0.01
0.00
0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

1.1

1

velocity U (m s" )

Figure 3-14. Velocity profiles obtained along the centreline of stone 25. The horizontal line above
each velocity profile represents the water depth at that location. Velocity profiles measured over the
centre of the stone, 5 em upstream of centre, and 5 em downstream of centre are labelled "0", "1",
and "-1", respectively. Measurement points correspond to the marked grid locations. All velocity
profiles above stone 25 were distinctly wedge-shaped, and labelled "w-s".

142

-E

N

0.030

s23, 0 (w-s)

0.025
s10, 0 (1-n)

~

(13

"'C

c::

0.020

::I

0

..0

E 0.015
0

.!:::
Q)
(.)

c::

.....

0.010

(13

.~

"'C

0.005
0.000
0.30

0.35

0.40

0.45

0.50

0.55

0.60

0.65

velocity U (m s·1)

Figure 3-15. Velocity profiles obtained above the centres ("0") of stone 23 (labelled "s23") and stone
10 (labelled "s1 0"). The horizontal line above each profile represents the water depth at that location.
Wedge-shaped and log-normal velocity profiles are labelled "w-s" and "1-n", respectively.
Measurement points correspond to the marked grid locations.

143

0.12

--

GR, 0 (1-n)

E 0.10

N

~

ctl

"C

N01, 0 (tr

0.08

c::
:::s

0
.0

E
0
....
Q)

(.)

c::

0.06
N02, 0 (1-n)
N03, 0 (w-s)

0.04

ctl

.!Q

"C

0.02
0.00
0.2

0.4

0.6

0.8

1.0

1.2

1.4

velocity U (m s- 1 )

Figure 3-16. Velocity profile obtained above the centres ("0") of naturally positioned stones N01,
N02, and N03, and the gravel riffle site ("GR"). The horizontal line above each profile represents the
water depth at that location. Wedge-shaped, transitional, and log-normal velocity profiles are labelled
"w-s", "tr", and "1-n", respectively.

144

co 4

e:.
(/)
(/)

!!!
iii

iii
Q)

.c
(/)

~ = separation zone

3
2

1

iii
3: 0
15
mean U

0

'E 10

~

0

E
>.
fl
c

jg

5

(/)

0

E
cu
!!!
iii
ch
(/)

-5

0

-10

'0

0

0 = 1 .0Pa
-15
-15

-10

-5

0

5

Streamwise distance X (X1 0-2m)

10

15

0

2

3

4

wall shear stress (Pa)

Figure 3-17. The distribution of wall shear stress (-rw) over the upper surface of stone 7. Wall shear
stress was measured using a Preston-static tube (PST) in two transects; a centreline transect in the
direction of flow (upper left), and a second, cross-stream transect (lower right) perpendicular to the
first, crossing the first transect at the centre of the stone. A composite bubble plot (lower left) shows
the relative magnitude of shear stress across the surface of stone 7. The upper right hand diagram
shows a top and side view of stone 7. The area superimposed on the top view of the stone
corresponds to the area represented in the bubble plot. Error bars are standard error (SE).

145

Cil 5

e:. 4

-o
~

~ = seperation zone

<JJ
<JJ

!!!
iii 3

....
nl

2

Q)

.s:::
<JJ

(ij
~

0
20

mean U
15

E
'b
..- 10

0

~

>~

5

.!!!
<JJ

0

~

-5

c:

:0
E
iii

.;,<JJ -10

e0

-15

0 = 1 .0Pa

-

---------~--------~---------

-10

-5

----------

5

0
2

streamwise distance X (x1 0- m)

10

0

2

3

4

5

wall shear stress (Pa)

Figure 3-18. The distribution of wall shear stress (tw) over the upper surface of stone 31 . Wall shear
stress was measured using a Preston-static tube (PST) in two transects; a centreline transect in the
direction of flow (upper left}, and a second, cross-stream transect (lower right) perpendicular to the
first, crossing the first transect at the centre of the stone. A composite bubble plot (lower left) shows
the relative magnitude of shear stress across the surface of stone 31 . The upper right hand diagram
shows a top and side view of stone 31. The area superimposed on the top view of the stone
corresponds to the area represented in the bubble plot. Error bars are standard error (SE).

146

Cil 5

e:.. 4

--o

~ = seperation zone

II)
II)

~
iii 3

:u 2
Q)

~

.s:::.

II)

iii
3: 0
20
mean U

E'

"!

0

E
>.

~
1!!
II)
:0
E

15
10
5

c:

nl

~

0

0
-5

iii

.;, -10

e0
II)

-15

0=1.0Pa

-20
-10

-5

5

0
2

streamwise distance X (X1 0" m)

10

0

2

3

4

5

wall shear stress (Pa)

Figure 3-19. The distribution of wall shear stress ('rw) over the upper surface of stone 25. Wall shear
stress was measured using a Preston-static tube (PST) in two transects; a centreline transect in the
direction of flow (upper left), and a second, cross-stream transect (lower right) perpendicular to the
first, crossing the first transect at the centre of the stone. A composite bubble plot (lower left) shows
the relative magnitude of shear stress across the surface of stone 25. The upper right hand diagram
shows a top and side view of stone 25 . The area superimposed on the top view of the stone
corresponds to the area represented in the bubble plot. Error bars are standard error (SE).

147

<U 5

e:. 4

-(0

~ = seperation zone

U)
U)

~

3

Ill

2

c;;
....
Q)

.r::

~

U)

Iii

3: 0
15

e

N

mean U
10

b

-10
0 = 1 .0Pa
-15
-10

-5

0

5

10

streamwise distance x (x10' 2m)

15

0

2

3

4

5

wall shear stress (Pa)

Figure 3-20. The distribution of wall shear stress (tw) over the upper surface of stone 23 . Wall shear
stress was measured using a Preston-static tube (PST) in two transects; a centreline transect in the
direction of flow (upper left), and a second, cross-stream transect (lower right) perpendicular to the
first, crossing the first transect at the centre of the stone. A composite bubble plot (lower left) shows
the relative magnitude of shear stress across the surface of stone 23. The upper right hand diagram
shows a top and side view of stone 23 . The area superimposed on the top view of the stone
corresponds to the area represented in the bubble plot. Error bars are standard error (SE).

148

Iii 5

e:. 4
"'
"' 3
~

~ = seperation zone

"'....co 2
Q)

.r:.

"'
"iii
~

0
30
mean

25

u,..

"E 20

"!

0

15

E 10
>-

fl
c:

5

]l

0

co

""0

0

0

E -5
co
!!!
1i) -10

.;, -15

"'eu -20
0 = 1 .0Pa

-25
-30
-15

-10

-5

0

5

streamwise distance x (x10-2m)

10

15

0

2

3

4

5

wall shear stress (Pa)

Figure 3-21. The distribution of wall shear stress (tw) over the upper surface of stone 10. Wall shear
stress was measured using a Preston-static tube (PST) in two transects; a centreline transect in the
direction of flow (upper left), and a second, cross-stream transect (lower right) perpendicular to the
first, crossing the first transect at the centre of the stone. A composite bubble plot (lower left) shows
the relative magnitude of shear stress across the surface of stone 10. The upper right hand diagram
shows a top and side view of stone 10. The area superimposed on the top view of the stone
corresponds to the area represented in the bubble plot. Error bars are standard error (SE).

149

ro- 10

e:. 8
"'

"'~
.,....

6

iii

Q)

4

"'

2

.s::

Iii
~

.... o

~ =seperation zone

~

0
15
mean U

'E 10

0

N

0

E

5

>-

2l
c

.,

0

~

"0

.,E
Q)

.:=

-5

"'ch
"'e
-10
0

-

-15

-----~------

-----~------

-10

0

-5

------

------

10

5
2

streamwise distance X (X1 0" m)

15

0

2

4

6

8

10

wall shear stress (Pa)

Figure 3-22. The distribution of wall shear stress ('tw) over the upper surface of naturally positioned
stone N01 . Wall shear stress was measured using a Preston-static tube (PST) in two transects; a
centreline transect in the direction of flow (upper left), and a second, cross-stream transect (lower
right) perpendicular to the first, crossing the first transect at the centre of the stone. A composite
bubble plot (lower left) shows the relative magnitude of shear stress across the surface of stone N01 .
The upper right hand diagram shows a top and side view of stone N01 . The area superimposed on
the top view of the stone corresponds to the area represented in the bubble plot. Error bars are
standard error (SE).

150

ro- 4

--en

~ =seperation zone

~

"' 3

"'~

u;

2

lu
Q)

~

.J::;

"' 1

Iii
~

0
15
mean U

'E

10

"'.....
0

5

~

>-

g 0

u;

Ill

:c

E

Ill

-5

~

u;

c), -10

"'e0

-15

0 = 1 .0Pa
-

-------

-10

-5

0

--------

0

------~------~--------

10

5

streamwise distance x (x10-2 m)

15

0

2

3

4

wall shear stress (Pa)

Figure 3-23. The distribution of wall shear stress ('tw) over the upper surface of naturally positioned
stone N02. Wall shear stress was measured using a Preston-static tube (PST) in two transects; a
centreline transect in the direction of flow (upper left), and a second, cross-stream transect (lower
right) perpendicular to the first, crossing the first transect at the centre of the stone. A composite
bubble plot (lower left) shows the relative magnitude of shear stress across the surface of stone N02.
The upper right hand diagram shows a top and side view of stone N02. The area superimposed on
the top view of the stone corresponds to the area represented in the bubble plot. Error bars are
standard error (SE).

151

c

Q)

7

•

E
Q)
....
6
:::l

en
rn

0

•

Q)

E 5

f-

(/)

rn

• •

a.. 3
en
en
Q)

....
en
....
rn

Q)

..c

en

2
0

1

·tf ·

-ro 0
~

0

..

.. ·

.. ·. •

.··

0

•
0

I

0

•
o·

0

wedge-shaped profiles
transitional profiles
log-normal profiles (field)
log-normal profiles (lab)

• •

a.. 4

-

•

1

..
.•

.. ..·

...

.•

.•

..

..

..

..

.. . •

•

0

•

•
2

3

4

wall shear stress (Pa) - du I dlnz estimate

Figure 3-24. Relationship between wall shear stress measured by the PST (tw-Psr) and wall shear
stress estimated from shear velocity based on the slope of of dU I dlnz (tw-<lU/dlnz) . The 1:1 regression
line ("line of equality") is shown . Locations with log-normal velocity profiles are shown as circles
(solid circles = field measurements, hollow circles = lab flume measurements), transitional profiles
shown as grey squares, and wedge-shaped profiles shown as solid diamonds. Error bars are
standard error (SE).

152

7

•

6

•

5
4

c:

Q)

E
Q)
....

A

•

• •

3
2

....

·········•·····

::J
(/)

ro 7

Q)

E 6

D

1-

8

(/)

--

c.. 5
ro 4
3
(/)

c..

(/)

Q)
....
....
ro
( /)

Q)

~
(/)

D

2

....

D

....

·····

D

D

1
7

ro 6

3: 5

c

4

3
2
1

0

0

2

3

4

wall shear stress (Pa) - du I dlnz estimate

Figure 3-25. Relationship between wall shear stress measured by the PST (•w-Psr) and wall shear
stress estimated from shear velocity based on the slope of dU I dlnz (-rw-dU/dlnz). Data are plotted
separately, based on the velocity profile shape at the point at which shear stress was measured.
Velocity profiles were wedge-shaped in (A), transitional in (B), and log-normal in (C). Locations where
log-normal velocity profiles were recorded in the field are shown as solid circles , hollow circles
indicate log-normal velocity profiles measured in the lab flume . The 1:1 regression line ("line of
equality"; shown as a dotted line) is shown in each case. The linear regression line is also shown in
each case (solid line). See text for significance tests of each plot. Error bars are standard error (SE).

153

c

N

1.0 .,.-------

- - - - - - - - - - - - - - - - ---,

>C

~

0.8

_g

0.6

~
c

0.4

ro
"0
c

::J

-e
E

ro
......
en
:0
Q)
>
~

0.2

-

log-nonnal
transitional
- - wedge-shaped

~

0

5

10

15

20

25

30

ux I U* PST
Figure 3-26. Comparison of the velocity profiles obtained over naturally positioned boulders (n = 3),
experimentally deployed boulders (n = 14), and a gravel riffle (n = 1). Velocities are normalised by
shear velocity (U*) as measured by the PST. The vertical axis (distance from the boundary) is
normalised by the total water depth (D). Wedge-shaped profiles are shown as a dashed line,
transitional profiles shown as a solid grey line, and log-normal profiles shown as a solid black line.

154

c

1.0 . . . , - - - - - - - - - - - - - - - - - - - - - - - - - ,

-

0.9

~

0.8

N><
ctl

-g
::::J

0

0.7

..c

0.6

0

0.5

(.)

Q)

0.4

.....

0.3

E

~

c:

ctl

.~

"C
Q)

>

:0:::
ctl

~

- - log-normal
transitional
- - wedge-shaped

0.2
0.1
0.0

--

0.2 0.4

~

0.6

~-

0.8

---

1.0

1.2

- ---

1.4

1.6

---

--- ---

1.8 2.0

_.J

---

----- -~

2.2 2.4 2.6

relative velocity Ux I U0. 50

Figure 3-27. Comparison of the velocity profiles obtained over naturally positioned boulders (n = 3),
experimentally deployed boulders (n = 14 ), and a gravel riffle (n = 1). Velocities are normalised by
velocities at mid-depth (U 0 .50 ) . The vertical axis (distance from the boundary) is normalised by the
total water depth (D). Wedge-shaped profiles are shown as a dashed line, transitional profiles shown
as a solid grey line, and log-normal profiles shown as a solid black line .

155

2.0

...

1.5

.....

Q)

..0
:::J

c: 1.0
Q)

"0

•

0.5

6.

stone 25
lab flume

~

0

0

:::J

0
.....
1..1..

0

0
0

• •

LL

E

stones
• exposed
non-exposed stones

., •

0

0.0 +------,------.----,-------,----.,-----1
0
50x1 03
1OOx1 03 150x1 03 200x1 03 250x1 03 300x1 03

local Reynolds number Rex

Figure 3-28. Relationship between Froude number (Fr) and local Reynolds number (Rex) at velocity
sampling locations over the upper surface of the stones (including all experimentally deployed and
naturally positioned stones). A significant relationship between the two parameters exists when all
2
data are considered together {R = 41 .07% ; ANOVA, F(1,16) = 12.85, p = 0.0025). However, when
the stones which have impeded boundary layer growth {i.e. flow impinging on the stone surface from
above - stone 25) and situations where skimming flow has possibly developed ("non-exposed
stones"; stone N01 and stone 7 during near-flood discharge) are excluded from the analysis, the
relationship is more significant (R2 = 81 .19%; ANOVA, F(1,11) = 47.49, p < 0.001). The regression
line shown was calculated using all field data (the three lab flume points (hollow triangles) were
excluded). "Exposed stones" included stones 7 (non-flood discharge), 31, 25, 23, 10, N02, and N03.

156

2.0

•

...u. 1.5

•

.....

Q)

.0

E
::I

c

1.0

•

Q)

"C

::I

0
.....
LL

0.5
0

0.0
0.00

•

0

0.02

0.04

•

•
0.06

0.08

0.10

1

shear velocity U* Psr (m s- )

Figure 3-29. Relationship between Froude number (Fr) and shear velocity (U *). The relationship is
significant (R2 = 25.21%; ANOVA, F(1 ,19) = 6.27, p = 0.022). The linear regression line is shown
(solid line). Symbols and conventions as in figure 3-24 .

157

0.10 ~

---------------

o uo.so
- - - U0 _50 regr.

'en
N

•

u0.002m
- - U0002 mregr.

0.08

E

•

0

0

0

c 0.06

..-

0

«

/

'o

0

~

«) 0

~

' (3

0

Q3
>
.....

/

.

.....-'

•
/

/

/

~

0 / /

/

/

/

/

/

/

0

0

0.04

ro

Q)

.J::.

en 0.02

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

velocity U (m s- 1 )

Figure 3-30. Relationship between shear velocity (U*; measured using the PST} and water velocity,
2
at mid-depth {U 0 .50 ) (R = 36.62%; ANOVA, F(1,19) = 14.18, p = 0.0013) and near the bed (Uo.oo 2 m}
2
(R = 87.08%; ANOVA, F(1,19) = 390.37, p < 0.001).

158

Figure 3-31. Diagrammatic representation of the flow field above isolated boulders in shallow
streams (D/dk < 2). In this simplified model , the approaching flow can be characterised as having two
zones. The bulk flow zone, where velocities (Ubulk) are relatively homogeneously distributed in the
vertical direction due to momentum transfer in the turbulent flow. The second , bed influenced zone
(BIZ), is characterised by reduced velocities (boundary layer effects) and turbulence generation
(growing I shedd ing edd ies, and boundary layer separation). Flow at the front of the boulder is
characterised by a horseshoe vortex (HV) and stagnation point (often unstable and oscillating). As
flow crosses the front of the stone, near-bed flows are constricted ; the influence of the constriction is
vertically limited. As a result, the flow near the bed accelerates at a greater rate than flows near the
surface. However, as flow crosses the relatively smooth upper surface of the stone, boundary layer
growth dissipates the near-bed, high velocity "wedge". Flows subsequently separate off the rear of
the stone, giving rise to recirculating flow zone (RFZ). The depth at which maximum velocities occur
(Zmaxu) increase as flow crosses the boulder.

159

Chapter 4

Microdistribution of Epeorus longimanus and periphyton on
torrential stream boulders

Summary
1. Field measurements and experiments were conducted to examine the microdistribution of the
torrential mayfly Epeorus /ongimanus (Heptageniidae) with respect to periphyton biomass, shear
stress, and substrate characteristics (i.e. stone roughness, topography).
2. Different factors influence larval distribution during daytime and night-time periods . During the
day, the few larvae that remain on the upper stone surface tend to be found in high shear stress
areas on the stone surface, generally towards the rear of the upper surface of the stones. During
the night-time, larvae were found on the upper surface of the stone surface, and avoided areas
where boundary layer flows had detached (i.e. the downstream edge of the stones).
3. Periphyton densities were significantly related to stone surface roughness and topography (i.e.
more algae on higher, rougher areas of the substrate).

The relationship between the

microdistributions of periphyton and E. longimanus larvae were nearly significant, suggesting that
high densities of this grazing mayfly may tend to reduce local periphyton standing stock. In some
cases, algal density and larval abundance appear to be inversely related .
4. Shear stress generally increases from the front to the rear of submerged , protuberant stones (i.e.
stones whose upper surfaces are beneath the water surface, but protrude upward into the main
flow to a greater extent than the surrounding bed material). At the rear of the stones , shear
stress abruptly decreases at or near where the flow separated .
5. E. longimanus larvae migrate diurnally. Larvae generally retreat to the underside of stones during
the day, and emerge during the night to graze on the upper, exposed surfaces of the stones at
night.

However, the rates of larval emergence and withdrawal differ, suggesting that the

migration behaviour during dusk emergence may be functionally different from that of dawn
withdrawal .

160

6. A stone reversal experiment suggests that the characteristic microdistribution of larval E.
longimanus is a response to near-bed flow factors.

However, larvae may be using this

environmental cue to identify high-resource feeding areas .
7.

The adaptive significance of the results, especially as they relate to trophic and predator-prey
interactions, are discussed.

Introduction
The examination of biotic processes at different spatial scales has become a major
consideration in the study of benthic systems (e.g. Muotka and Pentinnen 1994; Voelz and Ward
4

1996). Traditionally, the importance of large-scale (i.e. 10 - 10 m) processes has been a major
focus in stream ecology, and the roles of these processes in determining macroinvertebrate diversity
(Statzner and Higler 1986; Growns and Davis 1994), invertebrate drift (Lancaster et al. 1996), primary
productivity (Wellnitz et al. 1996), and stream hydrology and hydraulics (Jarrett 1990) are well
understood. However, microscale (i.e. 1o· m - 1 m) phenomena, such as local variations in near2

bed hydraulics, appear to be of greater ecological importance within riffle segments (Dudgeon 1982;
Collier et al. 1995) and on individual stones (Muotka and Pentinnen 1994; Hart et al. 1996; Voelz and
Ward 1996).
Recently, it has been suggested that the microdistribution of stream invertebrate are the
result of species-specific responses to environmental factors such as substrate size (Jowett et al.
1991 ), substrate roughness (Casey and Clifford 1989), habitat complexity (Way et al. 1995), food
resource availability (Peckarsky 1996), and hydrodynamic parameters (Statzner 1981; Collier et al.
1995). Biotic interactions , including predator-prey interactions (Mcintosh and Peckarsky 1996), have
also been found to play roles in determining microhabitat use by benthic invertebrates. Most studies,
however, have ignored the potentially important role that interactions between environmental and
biotic factors might play in determining the microdistribution of stream invertebrates (but see Muotka
and Pentinnen 1995; Hart and Merz 1998).
Hydrodynamic variables are generally thought to be the most important factors affecting the
ecology of benthic organisms in the stream environment (Hart et al. 1996; Carling 1992; Davis and
Barmuta 1989).

Near-bed velocities have been shown to influence the morphology, physiology,

behaviour, and microdistribution of stream organisms (e.g. Wiley and Kohler 1980; Statzner and Holm
161

1982; Hart et al. 1996; Delgado and Carbonell 1997; Bouckaert and Davis 1998). However, the exact
nature of the relationsh ip between the forces generated by moving water and the biology of the
benthic organisms is poorly understood, for two reasons . First, and primarily, stream ecologists have
used methodologies and field techniques that are quite different in spatial scale from that of benthic
invertebrates. For example, Muotka and Pentinnen ( 1994) found that aggregative responses of two
stream invertebrate predators (/soper/a grammatica and Rhyacophila obliterata) to blackfly larvae
prey in microhabitats of overlapping velocity preferences could not be detected unless they examined
the distribution of the organisms at scales much smaller than those traditionally used in field
experiments . Second, although it is widely recognised that benthic communities are sensitive to the
forces generated by the moving water of streams, the precise nature of the hydrodynamic parameters
to which stream invertebrates respond is poorly understood . The microdistribution and behaviour of
benthic stream invertebrates have been variously attributed to near bed velocity (Hart et al. 1996),
mean velocity (Collier 1994), turbulence intensity and Reynolds stress (Bouckaert and Davis 1998),
lift and drag (Weissenberger et al. 1991 ), Froude number (Wetmore et al. 1990), Reynolds number
(Statzner and Higler 1986), and the thickness of the viscous sublayer (Statzner 1981 ).

This

proliferation of causal factors has arisen because, in many cases, just one or two hydrodynamic
parameters are measured and related to patterns of faunal abundance. Furthermore, many of the
hydrodynamic parameters are closely related, and often derived from the same basic measurements
of water velocity.
Several recent studies of natural flows in stream channels have examined hydrodynamic
parameters (i.e. velocity, turbulence intensity, turbulence structure) within the benthic boundary layer
(e.g. Hart et al. 1996; Robert et al. 1996; Buffin-Belanger and Roy 1998). Of these studies, only Hart

et al. (1996) have successfully measured and related near-bed flow parameters (velocity, tubulence
intensity) to the distribution of benthic invertebrates. However, the relative rarity of studies of this kind
is due to the fact that the spatial resolution of conventional instruments used to measure water
velocity is quite coarse. Only recently have techniques been developed that allow for the examination
of water flow at the fine spatial scales (i.e. mm - em) experienced by benthic invertebrates. For
example, the electromagnetic current meter (EMCM) used by Robert et al. (1996) had a spatial
resolution of 1.3 em, while the hot-film anemometer used by Hart et al. ( 1996) had a spatial resolution
of 0.7 mm. Unfortunately, the hot-film anemometer used by Hart et al. was fragile and notably difficult
162

to deploy and calibrate in the field . Alternatively, fine-scale velocity measurements can be made with
a suitably small Pitot-static tube (White 1999). An appropriately proportioned Pitot-static tube can
also be deployed as a Preston-static tube (PST), and used to directly measure wall shear stress (see
Ackerman et al. 1994). This device has been tested and applied in the field (Ackerman and Hoover
2001 ).

Wall shear stress (-rw). which is typically measured indirectly (e.g. Collier et al. 1995) or

ignored (e.g. Jowett et al. 1991), is one of the most important forces affecting benthic invertebrates
exposed to flow on the stream bed (Davis and Barmuta 1989; Vogel 1994 ).
Stream hydrodynamics can also affect the structure of the benthic community. Peckarsky et
al. (1990) demonstrated that predation effects of the stonefly predator Dinocras cepha/otes on the
mayfly Baetis rhodani were greatest in hydraulic habitats favourable (benign) to the stonefly.

By

applying the "harsh-benign" hypothesis to stream ecosystems, Peckarsky et al. (1990) illustrated that
predator impacts were lower in abiotic regimes that were harsher to predators. Similarly, Tikkanen et
al. (1997) reported that simuliid larvae inhabited exposed microhabitats on the upper surface of
stones, where they were rarely encountered by the stonefly predator Diura bicaudata; different
microhabitat preferences between predators and prey lead to lower predator/prey encounter rates ,
and consequently, lower rates of predation . Hart and Merz (1998) further suggested that the inability
of predators to access high-velocity microhabitats might create sites which, being "harsh" to
predators, act as refuges from predation. Use of flow-mediated refuges would reduce not only the
direct, lethal effects of predation (Kerans et al. 1995), but also the indirect effects predators have on
benthic insect prey populations, such as reduced feeding rates (Peckarsky et al. 1993; Scrimgeour et
al. 1994) and increased drift (Mcintosh and Peckarsky 1996). Moreover, prey species may be able to
assess predation threat by associating the level of risk with some aspect of the environment (such as
a characteristic range of water velocities) shared by predator and prey (Dill 1987; Peckarsky 1996).
Flow can also influence the microhabitat preferences of many benthic invertebrate taxa by
mediating the availability of resources .

For example, Golubkov et al. (1992) found that the

rheophilous stream insects Stenopsyche marmorata and Skwala pusilla were sensitive to low levels
of dissolved oxygen, and consequently were found only in high velocity microhabitats in order to
maximise rates of gas exchange across their tracheal gills. This concurs with the earlier results of
Wiley and Kohler (1980), who found that mayflies moved to increasingly current-exposed surfaces as
dissolved oxygen concentrations decreased, presumably to increase gas transfer rates across
163

respiratory exchange surfaces. In addition, rates of nutrient delivery and gas exchange are higher for
benthic algae in high velocity microhabitats due to thinner and increasingly turbulent boundary layers
(Nowell and Jumars 1984).
The microdistribution and community structure of periphyton in streams is influenced by
grazing by herbivorous invertebrate taxa (e.g. Lamberti and Resh 1985; Feminella et al. 1989;
McCormick and Stevenson 1989}. Karouna and Fuller (1992) found that grazing by the mayflies
Paraleptophlebia and Ephemerella subvaria reduced algal densities and significantly altered the

community structure. In addition, invertebrate grazers may use periphyton abundance to assess the
quality

of a

microhabitat

(Kerans

1996}.

However,

because

the

microdistributions

of

macroinvertebrate grazers, periphyton, and near-bed flow parameters are interrelated, the relative
influence of algae and flow factors on the local abundances of herbivorous stream invertebrates
remains largely unknown.
Nymphs belonging to the heptageniid mayfly genus Epeorus are abundant in fast flowing
streams across western North America {Edmunds and Allan 1964). Epeorus nymphs are categorised
as clingers, and are found in erosional stream habitats (Edmunds and Waltz 1996}. Epeorus is a
representative torrential insect genus, displaying several adaptations to high velocity habitats
including a flattened body and a ventral , "sucker-like" arrangement of gills (see Weissenberger et al.
1991 ; Vogel 1994).

While there is some disagreement regarding the manner in which these

adaptations function, it is accepted that they enable Epeorus to maintain contact with the substrate in
high velocities. Given the association of Epeorus larvae with high velocity habitats, it represents an
ideal model organism to examine the relationship between near-bed hydrodynamics and the benthic
invertebrates that inhabit high-velocity microhabitats.
Given the linkages between hydrodynamics, predator-prey interactions, and periphyton
communities, it seems reasonable to attempt to relate the microdistribution of the benthic biota of
high-velocity streams to the abiotic components of the near-bed environment. While the relationships
that exist between environmental parameters and the microdistribution of sessile, suspension-feeding
organisms (i.e. caddisfly and blackfly larvae) are well documented, the microdistribution of mobile,
grazing benthic macroinvertebrates such as Epeorus longimanus in relation to hydrodynamic forces
are poorly understood. This purpose of this study is to examine the microdistributions of the mayfly
E. longimanus and primary producers (periphyton) over the upper surface of torrential stream

164

boulders , and relate their distributions to a variety of environmental parameters, including near-bed
hydrodynamics (e.g. shear stress) and stone surface characteristics (e.g. topography, roughness , and
within-stone position). Diurnal variation in microdistribution of this grazing mayfly was also examined
in order to determine if the same factors influence the distribution of E. longimanus larvae during dayand night-time periods.

Methods and materials
Study site and experimental design
The microdistribution of benthic invertebrates, algae, and near-bed flow parameters were
examined during September and October 1998, in a 20 m reach of a third-order tributary within the
Torpy River watershed (see chapter 3 for a description of the study site). Thirty small, light coloured ,
rounded stones, ranging in diameter from 22- 33 em, were marked with a 5 x 5 em reference grid of
points and deployed in the active stream channel at the study site in early August 1998 (see chapter

3 for a detailed description of the methodology used). Where possible, existing stones were removed
from the stream bed and replaced with experimental stones of similar shape and volume to ensure
that stream bed geometry within the reach was not altered significantly.

To minimise hydraulic

interactions between experimentally deployed substrates, the stones were placed in the channel at
approximately 1m intervals in an offset pattern, such that no stone was immediately upstream of
another (see Young 1992).
After deployment, the stones were left for 10 days to accumulate biofilm, periphyton, and
benthic invertebrates characteristic of the adjacent streambed. During this time the densities of larval
E. longimanus (a representative high velocity, torrential habitat mayfly species) were monitored on

each of the experimental stones.

After the acclimatisation period had ended, the distribution of

periphyton, macroinvertebrates, and shear stress (in addition to several other flow characteristics see chapter 3) were measured over the upper surface of the five stones that had the highest densities

of E. longimanus.
When possible, all parameters were measured during periods of similar stream discharge in
order to ensure hydrodynamic similarity between sampling dates.

165

Insect distribution
The number and distribution of larval E. longimanus within the 5 x 5 em grid marked on the
upper surface of each of the five stones were recorded several times over a four week period (see
Table 4-1).

Densities of E. longimanus larvae were determined on a per quadrat basis, and
2

converted to areal density (i.e. E. longimanus m· =number of insects within a 5 x 5 em quadrat I 0.05
m x 0.05 m). Daytime distributions were recorded on 8 dates between September 2 and October 4,
1998, while night-time distributions were recorded during the nights of September 16 and October 3,
1998. Observations of larval mayfly distributions were made using a small plexiglass sheet (20 x 10 x
0.7 em) held against the water surface and tilted in the upstream direction such that it allowed an
unimpeded view of the stream bed . Use of this device did not substantially influence the behaviour of
the larvae.

Prior to the experiment, other viewing devices (e.g. plexiglass observation boxes,

periscope viewers, etc.) were tested , but it was observed that larger, more visible apparatus disturbed
the mayflies. Night-time observations of insect distributions were made using a headlamp (Petzl
Zoom (4 .5 volts); Petzl Distribution Sport, Crolles, France) for illumination . As the red portion of the
visible spectrum has been reported as being less visible to stream insects (Mcintosh and Peckarsky
1996), a red filter over the lens of the headlamp was initially used in an attempt to minimise
disturbance to the larval mayflies. However, in this case, it was observed that movement of the light
source rather than colour was the source of the disturbance. Subsequently, the use of the filter was
discontinued, and movement of the headlamp minimised .

Flow characterisation
Using a Preston-static tube (see below), wall shear stress (tw) distributions were measured
across the upper surface of the experimental stones in two orthogonal transects crossing the upper
surface of each stone (Figure 4-1 c). Shear stress was measured every 2.5 em (i.e. at every grid
point, and at intermediate positions between grid points) in a streamwise transect, and at 5 em
intervals (i.e. at every grid point) in a cross-stream transect. The two transects intersected in the
centre of the stone (position 0,0 in the marked grid). When insect and wall shear stress distributions
were recorded on the same date, the insect distributions were recorded first to minimise disturbance
of the larval mayflies.
166

Wall shear stress was measured using a Preston-Static tube (PST) (see Ackerman et al.
1994; Ackerman and Hoover 2001 ). The PST used consists of a 90° bevelled tip syringe needle fixed
to a side-bored syringe needle (tube diameter= 0.88mm) . A Validyne model CD379-1-2 membranetype differential pressure transducer (Validyne Engineering; Northridge, CA), connected to the PST
via 2m of flexible pressure lines (inside diameter= 3 mm), was used to measure differential pressure
(the difference between static pressure Ps and dynamic pressure p 0 ), which can be converted into
shear stress (see chapter 3).

An average differential pressure was obtained by averaging 15

measurements obtained over approximately 30 seconds . The pressure transducer and pressure lines
were periodically flushed with water in order to minimise the possibility of the presence of trapped air,
which adversely affects differential pressure measurements.

The accuracy of wall shear stress

measurements using this method is dependent on proper orientation of the PST; the PST must be
resting on the stone surface and oriented parallel to local flow such that a stagnation point forms at
the dynamic tap (Ackerman and Hoover 2001 ). This orientation of the PST was achieved using
sodium fluorescein dye and an underwater magnifying periscope viewer (Edmund Scientific, New
Jersey, USA). Typically, when measuring shear stress at a point, the PST would be lowered until the
tip was in contact with the surface of the stone. A filament of dye was then released immediately
upstream of the PST using a hypodermic syringe fitted with a long, thin needle. Using the underwater
viewer, the orientation of the dye filament was compared to the orientation of the PST. If the two
were not parallel (i.e. the dynamic tap was not perpendicular to the flow), the PST was repositioned in
the correct orientation using the micromanipulator and positioning elbow (see chapter 3).
The distribution of boundary layer flow conditions (attached vs. separated boundary layer
flows) over each stone was noted using flow visualisation . Using a hypodermic syringe fitted with a
long, thin needle, fluorescein dye was injected into the flow at several points immediately upstream of
each of the experimental stones. The extent of areas of recirculating dye (i.e. backflow associated
with detached boundary layer flows) above each stone were recorded .

Periphyton distribution
At the end of the field study (October 7, 1998), all five stones were removed from the stream
in order to measure the microdistribution of periphyton.

167

The stones were removed at night (to

minimise exposure to ultraviolet light) and packed in ice in an insulated cooler for transport to the
aquatic sciences lab at the University of Northern British Columbia, where periphyton samples were
removed in circular quadrats using a rotating brush-and-suction sampling device (diameter = 2.72
em). The algal sampling device consisted of a PVC cylinder (closed at the upper end and encircled at
the lower end with a high-density foam rubber skirt) which housed a circular, nylon-bristled brush.
The brush, in turn , was connected to a rotary tool (Dremel MultiPro rotary tool model 395; Winconsin,
USA) via a brass shaft. Two lines opened into the side of the cylinder. The first was a water line,
which allowed a jet of cold, distilled water to be introduced when the brush was activated, creating an
algal slurry. The second was a vacuum line, which drew off the slurry.
Algal samples were removed in three transects in the streamwise direction. The centre of
each quadrat corresponded to a marked point (intersection) in the 5 x 5 em reference grid (Figure 41b). Only the upper surface of the stone was sampled; the effectiveness of the sampling device used
in this study was limited, as are other devices, when used to obtain algal samples from the curved
surface at the edges of rounded stones. However, this method was shown to be as effective as direct
extraction in a comparison of methods (Hoover et al. in prep.). For chlorophyll a analyses, each algal
sample was vacuum filtered onto a glass fibre filter (GF/C; Whatman, UK), immediately frozen at 40°C, and extracted in 10 ml of 90% acetone (buffered with MgC0 3 ; as per Geider and Osborne
1992) for 24h at

soc in the dark.

Chlorophyll

a concentrations were determined using a

spectrophotometer (Perkin Elmer UVNis Lambda 2S Spectrometer; Perkin Elmer, Germany) using
the methods of Geider and Osborne (1992).

Stone surface characteristics (surface roughness and topography)
A measure of stone roughness was obtained by placing a sheet of white printer/copier paper
2

(75 g m" ) over each 5 x 5 em quadrat. The paper was held firmly against the stone surface, and a
pencil (1600 HB) was rubbed firmly and evenly over the surface of the paper producing a 'graphite
rubbing' of the stone surface. The texture and irregularities of the stone surface were defined on the
resultant 'rubbing'. All indentations in the stone surface whose long axis exceeded 5 mm in length
were then counted on each 5 x 5 em 'rubbing', producing an estimate of the surface roughness
(degree of pitting) within each quadrat.

168

The surface topography of each stone was determined by measuring the distance down from
a level rod to each point on the grid. For example, the topographical measurement of the highest
point of the stone surface was recorded as "0 em", a grid point 1.5 em lower was recorded as "1.5
em", etc.

Diurnal variation in behaviour
Observations of diurnal changes in larval distribution {i.e. changes in numbers of larval
mayflies on the upper surface of the five experimentally deployed stones) were made on two dates.
The first set of observations was made from 13:00 h on Sept. 26 to 04 :00 h on Sept. 27, 1998. The
exact observation times varied slightly; depending on the numbers of mayflies, it took approximately
five minutes to record mayfly numbers and positions and travel to the next stone. Observations were
made at 2 hour intervals during the dusk hours (18:00 h - 22:00 h).

In addition, a single

measurement was taken at 09 :00 h Sept. 27 (i.e. the following morning).
On the second observational date, the density and distribution of larval E. longimanus over
the upper surface of the five experimental stones were recorded every two hours over a 22-hour
period, from 14:00 h on October 3 to 12:00 h on October 4, 1998 (with the exception of the mid-night
period; no record was made at 02:00 h). In all cases, all larvae visible over the upper surface of each
stone were recorded. Night-time observations were made with the aid of a headlamp.
Light measurements were taken every 30 minutes using a quantum sensor and light meter
(model LI-192SA and model Ll-250, respectively; LI-COR Inc., Nebraska, USA), except during
periods of rapid light change (dawn and dusk, approximately 08 :45 h- 09:30 hand 18:00 h- 19:45
h, respectively), when they were recorded every 15 min. Light measurements were made on the
streambank at the study site.

169

Table 4-1. Comparison of measured parameters at all sampling dates
Stone

(diameter)

Date (1998)

Depth above datum
(centre) (mm)

Shear
stress

Parameters measured
I E. /ongimanus dist'n

night
I Day
Yes
54
September 2
Yes
27
September 5
Yes
SeptemberS
2S
d = 29.0 em
Yes
September 13
55
3
Yes
Yes
September 16
36
Yesb
Yes
90
September 26
Yes
Yes
September 27
95
S4
Yes
September 30
Yes
Yes
October 3
120
Yes
Yes
115
October4
3
0
Yes
Yes
September 2
50
31
Yesb
31
September 5
Yes
d = 17.5 em
SeptemberS
42
Yes
September 13
60
Yesc
September 16
40
Yes
September 17
41
Yesb
Yes
September 26
96
Yesb
Yes
September 27
99
Yes
Yes
October 3
11S
Yesb
Yes
October4
114
Yes
September 2
53
25
3
Yesb
September 5
4S
Yes
Yes 3
Yes
SeptemberS
47
d = 21 .0
Yesc
September 13
45
Yesc
September 16
4S
Yes
September 17
4S
Yes
Yes
September 26
55
Yes
Yes
September 27
56
Yes
Yes
October 3
Yes
October4
Yes
S2
Yes 0
23
September 2
25
Yesb
September 5
15
Yesb
d = 1S.5 em
SeptemberS
19
Yesc
September 13
32
Yesc
September 16
26
Yesb
September 26
62
Yes
Yes
66
Yes
September 27
Yes
September 30
30
Yes
Yes
October 3
S2
Yes
Yes
October4
0
Yes
September 2
22
10
Yesb
September 5
9
Yesc
10
d = 27.0
SeptemberS
Yesc
September 13
25
Yesb
September 16
15
September 23
26
Yes
Yesb
September 26
60
Yes
Yes
Yes
September 27
63
Yes
October 3
Yes
October4
76
Yes
Yes
a shear stress measured only at centre of stone (for purposes of companson); ent1re distribution
recorded on all other dates
b stone surface examined, no E. longimanus present
c E. longimanus distribution recorded after 18:00 h (not included in analyses)
7

-

-

-

-

-

-

-

ss

-

so

-

-

so

170

-

-

-

-

Stone rotation experiment
The proximal relationship between hydrodynamic forces (e.g. shear stress), E. /ongimanus ,
and periphyton was examined by a substrate reversal experiment. The night-time microdistribution of
E. longimanus larvae was recorded on the upper surface of three naturally occurring stones by

recording the average density of larval E. /ongimanus in 5 em bins in the downstream direction. The
stones selected were comparable in both size and E. longimanus density to the experimental stones .
After the original distribution was recorded , each of the three stones were then rotated 180° (i.e. the
original downstream edge became the upstream edge). After 40 minutes, the microdistribution of the
mayfly larvae was recorded, and the stones were rotated 180° to their original orientation . The
microdistribution of the mayflies was again recorded after 40 minutes. A one-way ANOVA was used
to test if the mayflies had moved to new locations (i.e. were proximately responding to the new
distribution of near-bed flow parameters) or had remained in their original microdistribution (i.e .
proximately responding to substrate and food resource (periphyton) factors) .

Analyses
All mayfly microdistributions recorded between 08:00 hand 18:00 h (all dates) were averaged
to produce a mean daytime microdistribution, while records between 20:00 h and 06:00 h were
averaged to produce a mean night-time microdistribution . Dusk and dawn observations (06:00 h 08 :00 h and 18:00 h - 20:00 h, respectively) were not used in these analyses, in order to minimise
the influence of the rapidly changing light level and corresponding changes in larval behaviour on
analyses. In order to relate larval microdistributions to environmental factors measured over a period
of several weeks, mean daytime and night-time larval densities were determined for each 5 x 5 em
quadrat (see Table 4-1 for dates).

To account for among-stone differences in biotic and abiotic

parameters, stone diameter (a discrete, continuous variable) was included in all ANOVA models as a
covariate, and in all standard multiple regression models as a covariable.
Data were examined for normality and homogeneity of variance (Kruskai-Wallace W test and
data visualisation including boxplots and frequency histograms). Daytime and night-time abundances
05

of E. /ongimanus were not normally distributed. However, ln(x+1) and x · transformations of these
data did not meaningfully increase the normality of the data. As a result, in order to maximise the
171

interpretability of the results, all analyses were performed using the untransformed mayfly
distribution/abundance data.
As shear stress was measured at fewer points on the surface of the experimental stones than
all other variables, two multiple regression analyses were performed with per quadrat mayfly density
(daytime and night-time densities of E. /ongimanus) as the response variable.

As such, for the

ANOVAs, a Bonferonni adjustment of a was used (Tabachnick and Fidel! 1996), and the level of
significance was reduced from a = 0.05 to a = 0.025.
All analyses were conducted using Statistica 5.1 (1996; Statsoft, Tulsa, Oklahoma).

Results
It was possible to observe the behaviour and position of larval insects on the upper surface of
stream stones, and the generic identity of individual insects could be determined . With the exception
of extremely rare observations of larval blackflies (Simuliidae), and larval mayflies of the genera
Baetis and Rhithrogena, the habitat represented by the upper surface of the stones was inhabited

exclusively by larval Epeorus longimanus (Heptageniidae).

Abiotic factors influencing the microdistribution of E. longimanus
The microdistribution of E. longimanus larvae was dependent on both within- and amongstone factors . When E. longimanus larvae were present on the upper surface of the stones in high
numbers (generally at night), they were at highest densities upstream of the flow separation zone (at
the downstream edge), near the stones' centre . There was no significant differences in per quadrat
mayfly densities among stones during the night-time (ANOVA, F(4,130)

=0.40, p =0.81). There was,

however, a significant difference in mayfly densities among stones during the day (AN OVA, F(4, 130)

= 5.08, p < 0.001) (Figure 4-2A).
An ANCOVA was performed on daytime and night-time densities of E. /ongimanus, with
boundary layer flow conditions as the predictor variable and stone diameter included as a covariate.
During the night, there were significantly greater densities of mayflies in areas of the stones with
attached boundary layer flows compared to areas with separated flows (F(1, 132) = 8.83, p = 0.0035)
(Figure 4-28). In the daytime, although not significant (F(1,132) = 3.41, p = 0.067), the trend was
172

similar.

In addition , regression analysis shows that night-time and daytime distributions of E.
2

Jongimanus were significantly and positively related (R = 21 .12%; ANOVA F(1,133) = 35.62, p <

0.001) (Figure 4-3). This trend demonstrates that although daytime densities were much lower,
daytime microdistributions of E. Jongimanus were similar to those observed at night.
During the night, per quadrat densities of E. longimanus varied significantly with streamwise
(x) distance along the upper surface of the stones (ANCOVA, F{6, 127) = 4.32, p < 0.001) (Figure 44). When the mayfly distributions of all stones were pooled , there were more mayfly larvae in the
centre of the stones than at either the upstream or downstream edges (Figure 4-4; Table 4-2).
However, there was no corresponding trend in daytime distributions of E. longimanus larvae
(ANCOVA, F(6,127) = 1.63, p = 0.15).

Table 4-2.

Post-hoc comparison of means {Tukey's HSD test} following the

significant ANCOVA on night-time distribution of Epeorus longimanus and
streamwise {x} position
Position (x)
+1Scm
+1Scm
+10 em
ns
+Scm
ns
0 cma
ns
-S em
ns
-10 em
ns
-1S em
ns
* significant at a=0 .05
ns = p > 0.075
a centre of stone.

+10 em

+Scm

0 cma

-S em

-10 em

ns
0.039*
ns
ns
ns

ns
0.037*
ns
ns

ns
0.067
0.053

0.063
0.050

ns

-1S em

Two sets of standard multiple regressions were performed between larval mayfly distribution
(as the response variable} and a series of environmental factors (as predictor variables). Due to
unequal sample sizes in the data subsets, the multiple regressions were performed separately. In the
first multiple regression analysis, mayfly distribution {daytime and night-time microdistributions,
analysed as subsets) was included as the response variable and shear stress as the predictor
variable. In the second multiple regression analysis, mayfly distribution (again, daytime and nighttime microdistributions, analysed as subsets) was included as the response variable, and streamwise
(x) position within the stone, stone surface roughness, and stone topography were included as the
173

predictor variables .

In all analyses, stone diameter was included in both models as an ordinal ,

discrete variable to account for among-stone differences
There were marked differences in the factors influencing the microdistributions of E.
/ongimanus larvae between day and night periods . During the daytime, there were significantly more

larval E. longimanus in areas of the substrate with higher shear stress (Figure 4-5), and on stones
with larger diameters (Table 4-3; see also Figures 4-6 to 4-1 0). During the night-time, however, there
was no sign ificant relationship between larval microdistribution and either shear stress or stone
diameter.
Larval microdistribution during the day was also positively related to streamwise position and
stone diameter, and negatively related to topography (i.e. there were more larvae higher and further
back on exposed stone surfaces; Table 4-4).

During the night, larval microdistribution was only

related to topography (i.e. more larvae higher on exposed stone surfaces; Table 4-4). However, the
relationship between mayfly microdistribution and stone roughness approached significance (p =
0.087), suggesting that there may be more mayflies on rough stone surfaces during the night.

Table 4-3.

Multiple regression analyses of the relationship between the

microdistributions of periphyton and Epeorus /ongimanus (daytime and nighttime distributions) and wall shear stress ('tw)
{df) = F

shear stress (z-w)
Stone diametef
p-va/ue
fJ
p-va/ue
jJ
a) daytime E.longimanus
36.15% (2,36) = 10.19 0.021
16.66
< 0.001
22.96
b) night-time E. longimanus
2.08%
(2,36) = 0.38
0.61
25.47
0.42
31.44
c) periphytonb
5.11%
(2,36) = 0.67
0.53
-0.16
0.27
-0 .19
a stone diameter entered into the model as an ordinal discrete variable to partition out variance due to
among-stone differences
b presented as mass of chlorophyll a per cm 2
Model

~

Biotic factors influencing microdistribution of E. longimanus
The relationship between benthic biota, bed characteristics, and stream hydraulics was
variable among stones . On stone 7, the centreline distributions of E. longimanus and periphyton had
an inverse relationship, as did shear stress and E. longimanus density (Figure 4-6). However, this
trend was not consistent among all stones.

While the shear stress on the surface of stone 31

increased from the front of the stone to the back, where it sharply declined, there is no clear

174

association with either algal or mayfly distributions (Figure 4-7).

Similarly, on stone 23, there was a

distinct increase in shear stress from front to back of the stone, where flow eventually separated
(Figure 4-8). While this was mirrored by a front-to-back increase in mayfly densities, algal biomass
tended to be greatest over the front and centre of the stone. Aga in, on stone 10, there was a similar
front-to-back increase in shear stress (Figure 4-9).

However, this was poorly related to the

distribution of algae and mayflies, which tended to be inversely related (E. /ongimanus densities
highest, algal biomass lowest over the centre of the stone). The shear stress over stone 25 was
relatively uniform over the surface of the stone (Figure 4-1 0), with the exception of the leading edge
{due to wake interference from upstream roughness elements).

While there was little variation in

algal densities over the surface of stone 25, densities of E. /ongimanus increased from the front to the
back.

Table 4-4.

Multiple regression analyses of the relationship between the

microdistribution of Epeorus /ongimanus (daytime and night-time) and
streambed characteristics
Model
A) daytime E. longimanus
Explanatory variables
Streamwise position (x)
Roughness
Topography
3
Diameter

R'

22.95%

(df) = F
(4,130)

=9.68

p-value
< 0.001

jJ

0.014
0.82
< 0.001
< 0.001

6.40
-0.15
-7.26
13.51

=

B) night-time E. longimanus 25.75%
(4, 130) 11.28 <0.001
Explanatory variables
Streamwise position (x)
0.38
12.72
0.087
6.21
Roughness
Topography
< 0.001
-66 .57
3
Diameter
0.50
13.86
a stone diameter entered into the model as an ordinal discrete variable to partition out variance due to
among-stone differences

Factors influencing the microdistribution of periphyton
There were significant differences in periphyton growth among the five stones (ANOVA
F(4,61) = 10.49, p < 0.001) (see Figures 4-6 to 4-1 0). Two standard multiple regressions were
performed between periphyton growth (measured as mass of chlorophyll a cm.2 ) as the response
variable and streamwise distance, stone roughness , topography (height on stone), and local E.
175

/ongimanus density (the first and second regression models including daytime and night-time mayfly

microdistribution densities, respectively) as predictor variables. Stone diameter was included in both
models as an ordinal, discrete variable to account for among-stone differences. In both cases, only
three of the variables contributed significantly to prediction of algal microdistribution density (Table 45); roughness, topography, and stone diameter (the latter variable reflecting significant among-stone
differences). Periphyton biomass tended to be greater higher on the stone surface, and greater algal
densities were found in areas of the stone surface with greater surface roughness. In addition, the
relationship between algal density and daytime and night-time microdistributions of E. longimanus
approached significance (p = 0.060 and p = 0.055, respectively), suggesting that algal densities were
reduced in areas of the stone surface where densities of E. longimanus were higher. Algal biomass
tended to be highest over the centre of the stones, and lowest on the upstream and downstream
edges (Figure 4-11 ); this difference, however, was not significant (ANCOVA, F(4,57) = 1.05, p =
0.39). Periphyton microdistribution was not significantly related to shear stress (Table 4-3; see also
Figure 4-6 to 4-10).
An ANCOVA was performed on algal densities, with boundary layer flow conditions
(separated vs. attached boundary layer flows) as the predictor variable and stone diameter as the
covariate. While there tends to be more periphyton growth in those areas of the stone surface with
attached boundary layer flows (Figure 4-12), the difference is not significant (ANOVA; F(1 ,63) = 2.65,
p =0.11).

Diurnal variation in positioning
E. longimanus displayed a distinct diel periodicity in their positioning (Figures 4-13 & 4-14).
There were significantly more larvae on the upper surface of the stones during the night-time hours
than there were during the day (ANOVA, F(1 ,268) = 58.35, p < 0.001 ).

It is important to note,

however, that daytime withdrawal of larvae from the upper surface of the stones was not complete.
Daytime densities of E. longimanus on the upper surface of the stones were approximately 10% of
night-time densities. While there were significant differences in daytime densities among stones, the
daytime distribution of E. longimanus was significantly correlated with both shear stress and
topography (Tables 4-3 & 4-4). This suggests that mayflies that remain on the upper surface of the

176

stones during the day were found predominantly in high shear stress areas near the apex of the stone
surface.

Table 4-5.

Multiple regression analyses of the relationship between the

microdistribution of periphyton to streambed characteristics and grazing
mayfly densities
Model
Ft
A) Periphyton densitya
29.76%
~~~~~~~~----~
Variables
Streamwise position (x)
Roughness
E. longimanus {day)
Topography
Diameterb

{df) = F
(5,60) = 5.08

p-va/ue
< 0.001

0.20

< 0.001

0.060

< 0.001

0.016

fJ
0.11
0.093
-0.0049
-0.30
-0 .26

29.92%
< 0.001
B) Periphyton densitya
(5,60) = 5.12
Variables
0.38
0.072
Streamwise position (x)
< 0.001
0.10
Roughness
-0.00080
0.055
E. longimanus (night}
-0.35
< 0.001
Topography
Diameterb
-0 .36
< 0.001
a mass of chlorophyll a per cmL
b stone diameter entered into the model as an ordinal discrete variable to partition out variance due to
among-stone differences

~~~~~~~~----~

The withdrawal of the mayfly larvae from the upper surface of the stones at dawn was abrupt
compared to the variable timing of the emergence at dusk (Figure 4-15). There was an abrupt change
in larval density on the upper surface of the stones at dawn (-07 :00 h). The withdrawal of individuals
from the upper surface coincided with the onset of detectable light levels in the morning. However,
larval emergence was more gradual. Larvae began to emerge during late afternoon, when light levels
were still relatively high, and continued to move to the exposed upper surface of the stones until well
after dusk. The asymmetry between the highly variable rates of emergence and the relatively rapid
withdrawal suggests that additional factors other than light level may have influenced the migration
rates of E. longimanus from the underside of stream boulders to the upper surface. Net migration of
larvae was limited during the night; there was little change in larval density between 23:00 and 05:00
h.

177

Stone reversal
The handling of the stones during the stone reversal experiment did not noticeably disturb the

E. longimanus larvae. Larvae did not immediately or rapidly migrate from their original positions .
Larval densities before and after stone handling were similar, and few, if any, larvae drifted from the
stone surface.
Figure 4-16 shows the results of the stone reversal experiment. In all three cases, there were
generally higher densities of E. longimanus on the downstream portion of the stone (Figure 4-16a).
When the stones were reversed, the mayflies retreated downstream to new positions on what were
originally the leading edges of the stones (Figure 4-16b ). When the stones were returned to their
original orientations, the mayflies reversed their movements, and returned to their original
distributions (Figure 4-16c).

This demonstrates that the proximal factor to which E. longimanus

responded was local, near-bed hydrodynamic factors rather than distribution of algal food resources .
There was a significant timex distribution interaction (ANOVA, F(12, 42) = 3.86, p < 0.001). The
rotation of the stones resulted in a significantly different distribution of mayflies.

Post-hoc linear

regressions show that the initial (before initial rotation) and final (after the second rotation when
stones were returned to their original orientation) distributions of E. longimanus on the upper surface
2

of the three stones were significantly related (R adi = 33.0%; ANOVA, F(1, 121) = 11.98, p = 0.0023).
The initial and rotated (after the stone had been rotated 180°) distributions of E. longimanus were not
related (R2adi = 0.058%, ANOVA, F(1, 121) = 1.01, p = 0.33), nor were the rotated and final
2

distributions (R adi= 0.98%; ANOVA, F(1,121) = 0.21, p = 0.65).

Discussion
The objective of this study was to examine the microdistribution of a highly mobile species of
grazing heptageniid mayfly (Epeorus longimanus) with respect to near-bed flow parameters,
streambed characteristics, and the microdistribution of algal food resources across the surface of
torrential stream boulders. The results of the present study suggest that the proximate relationship
between primary producers, grazers, and environmental variables may be more complex than
previously thought. As hypothesised, the microdistribution of E. longimanus was significantly related
to several environmental parameters, each measured at organismal spatial scales . However, the

178

results revealed that the functional relationship between the microdistributions of this torrenticolous
grazer, periphytic algae, and near-bed and substrate characteristics differed between day- and nighttime periods, suggesting important behavioural inputs into the relationship .

The importance of scale
The densities of E. /ongimanus larvae and periphyton were significantly different among
stones of similar size located adjacent to one another in the same reach . This demonstrates the
importance of the spatial scale at which the behaviour of benthic biota is examined. The planar area
of each of the stones used in this study was similar to the area traditionally sampled in benthic
ecology studies. Surber or box-type samplers, which sample relatively large areas of the stream bed
(- 500 cm 2 ), are often used to sample both the density and diversity of benthic taxa (e.g. Hynes 1971 ;
Statzner 1981; Dudgeon 1982).

If employed in this study, however, such methods would have

sampled across several microhabitat types (e.g. sides vs . upper surfaces of streambed stones) and
the microdistributions of both E. longimanus and the environmental parameters that influence the
positioning of individual larvae would not have been detected.

In order to determine which

parameters influence the behaviour and ecology of stream invertebrates, the immediate environment
of the invertebrates must be characterised at the spatial scales of the organisms themselves, rather
than the scale of the observer.
Recent studies have measured either the microdistribution of benthic taxa or environmental
factors at small spatial scales, but rarely are both sets of parameters measured at fine, organismdefined scales (but see Hart et al. 1996).

For example, Voelz and Ward (1996) examined the

microdistribution of the caddisflies Arctopsyche grandis and Brachycentrus occidenta/is using 3 x 3
em quadrats, but measured current velocity only at the front, top, and bottom (leading edge) of each
stone. In order to examine the spatial relationship between biotic and abiotic parameters at the scale
of a torrential stream invertebrate, the present study quantified the microdistribution of larval E.
2

longimanus over the surface of stream stones using 5 x 5 em (25 cm ) quadrats, periphyton using
2

5.84 cm circular quadrats, and shear stress at even finer spatial scales (measurements either 2.5 or
5.0 em apart).

179

Factors influencing microdistribution
The relationship between the microdistribution of E. longimanus larvae and biotic,
hydrodynamic, and substrate factors differed between daytime and night-time periods . Night-time
distributions of E. longimanus were significantly related to topography and flow separation factors .
Larvae avoided areas of recirculating, separated flow, and grazed on algal growth high on the surface
of streambed stones. During the night, the larvae were not strongly influenced by other measured
substrate or hydrodynamic variables. However, in addition to topography, daytime distributions of E.
/ongimanus on upper stone surfaces were related to shear stress and streamwise (x) distance along

the stone surface. This demonstrates a diurnal shift in the functional responses of E. longimanus to
environmental parameters. The relatively few (-10% of night-time densities) larvae that remained on
the upper surface of streambed stones occupied high shear areas towards the rear of the upper stone
surface.
These results are consistent with the suggestion that the factors that influence the
microdistribution of benthic invertebrates are different between daytime and night-time periods . The
presence of visually oriented fish predators, for instance, is generally thought to cause several stream
insect species to retreat to the underside of streambed stones during the day (e.g. Cowan and
Peckarsky 1994). This ceases to be an important determinant of insect micropositioning during the
night when fish predators can no longer see their prey. Similarly, Donahue and Schindler (1998)
found that simuliid larvae migrated to shaded microhabitats during periods of intense solar radiation,
demonstrating that exposure to UV also influences invertebrate microposition.

In this fashion , UV

acts as an environmental factor that modifies behaviour only during daylight hours. The results of the
present study indicate that there may be diurnal variation in the responses of benthic invertebrate
taxa to environmental factors that do not vary diurnally.

Ecological hydrodynamics
During the night, there were significantly fewer larval E. longimanus found in areas of the
stone surface where flow separation occurred (i.e. areas of recirculating flow at the rear and , to a
lesser extent, the front of the stones) . Larval E. longimanus tended to be found at highest densities
over the centres and just downstream of the centres of the experimental stones, suggesting that they
180

preferred regions of stone surfaces where unidirectional, generally linear flow occurred . Peckarsky
and Cowan (1995) showed that E. /ongimanus has distinct preferences for upper stone surfaces over
upstream and downstream faces , while Rader and Ward (1990) found that mayflies in a Colorado
river rarely used the front and rear sides of substrates , almost exclusively occupying the upper, lower,
and lateral surfaces of substrates. Larval E. /ongimanus may preferentially select upper stone surface
microhabitats where flow is unidirectional because they rely on several highly specialised
adaptations, including a sucker-like arrangement of gills and a flattened, streamlined body shape, to
maintain contact with the substrate in high velocity, high shear microhabitats. Weissenberger et al.
(1991) note that in microhabitats with steep velocity gradients, organisms are at greater risk of being
dislodged by lift than drag. Thus, the morphological adaptations displayed by larval E. longimanus
may largely act to counter the typically high lift forces encountered by organisms on the upper
surfaces of stones in torrential streams.

The adaptations of this torrential mayfly enable it to

effectively travel on the exposed surfaces of stones located in high velocity areas of the stream
channel.

E. /ongimanus larvae observed were able to move forward , sideways, and backward

without difficulty in microhabitats where current velocities measured 2 mm above the substrate
exceeded 1.0 m s·1 .
The efficacy of these adaptations may be reduced, or even negated, in areas of variable flow
direction and high turbulence intensities. As such, the upstream and downstream faces of stones
may represent areas where contact with the substrate becomes more difficult for E. longimanus
larvae. Vortices shed from the lee side of protuberant boulders alter the structure of the flow,
producing high levels of turbulence above the downstream side of boulders (Robert et al. 1996). The
streambed at the study site was composed of stones that protruded upwards into the water column
(e.g. rough-turbulent flow) . Flow in channels of this type is characterised by convected eddies and

flow structures whose motion is predominantly towards the bed ("sweeps"), which further acts to
create unstable, turbulent flow conditions at the leading edge of exposed stream boulders.

In

addition, solenoidal {"horseshoe") vortices can form around the upstream side of stones protruding up
into the flow (Vogel 1994; Bouckaert and Davis 1998), creating directionally variable, oscillating flows .
Other torrenticolous mayfly species, including Rhithrogena, Ephemerella doddsi (see Minshall 1984)
and Deleatidium (type myzobranchia) (Collier 1994), which have morphological adaptations similar to
those of E. longimanus, would be expected to have similar microhabitat requirements .
181

Lacoursiere (1992) also found that simuliid larvae preferred regions of artificial substrates
where flow was moving in a unidirectional fashion, and generally avoided areas exposed to high
turbulence and variable flow direction. However, although similar, the microdistributions of simuliid
and mayfly larvae are the result of different behavioural processes. While the grazing mayfly larvae
migrated to locations that minimised the probability of accidental entrainment and maximised access
to algal food resources, suspension-feeding blackfly larvae chose flow microhabitats that maximised
delivery rates of food particles to their labral fans while minimising drag on the posterior portion of
their abdomens (Lacousiere 1992; Lacoursiere and Craig 1993).

Flow mediation of predator-prey interactions
Larval E. longimanus made extensive use of the microhabitat represented by the exposed,
relatively high-shear upper surfaces of torrential stream stones during the night and, to a lesser
extent, during the day. Although predatory invertebrates (e.g. Rhyacophila, Drunella, and Megarcys)
are relatively numerous in microhabitats preferred by Epeorus (see Chapter 2), and were collected at
the study site {T. Hoover, unpublished data), these predators were never observed on the upper
surface of the experimental stones during either day- or night-time periods, during the entire two
month study period. This suggests that the upper surfaces of torrential stream boulders, wh ich are
typically exposed to high velocities and high shear stress, act as flow-mediated refuges (Hart and
Merz 1998) from invertebrate predation for E. longimanus larvae. This conclusion conflicts with the
findings of Peckarsky and Cowan ( 1995), who reported that the use of upper stone surfaces by E.
longimanus did not represent a refuge from stonefly predation, as they observed that stonefly

predators (Kogotus) were able to access all areas of streambed stone surfaces. However, the "high
flow microhabitats" in the experimental stream chambers used by Peckarsky and Cowan {1995) had
1

average current velocities of only 15.6 em s- , while measured mean velocities over the substrates in
the present study were much higher (range = 24.2- 136 ems-\ mean = 71 .8 em s-1) . The difference
between the values recorded in the field in this study and the values reported by Peckarsky and
Cowan (1995) suggests the existence of a threshold velocity above which some stonefly predator
taxa may not be able to forage for prey.
Invertebrate predators, which generally lack the highly specialised morphological adaptations
of torrenticolous grazers to maintain contact with the substrate in high velocity habitats (see
182

Weissenberger et al 1991; Statzner and Holm 1989; also see Minshall 1984; Vogel 1994), may either
avoid high velocity areas (Muotka and Pentinnen 1994), or be eroded from the substrate when they
attempt to access prey. Hart et al. (1996) note that while many prey species are sessile or semisessile and have morphological adaptations that anchor them securely to the stream bed, many
predators depend on a foraging mode to search for prey, a trait which selects for high speed and
manoeuvrability rather than tenacity. Additionally, predators (especially the caddisfly Rhyacophila
and the perlodid stonefly Megarcys) might be unable to exploit prey in high-velocity areas of the
substrate due to the fact that they are generally larger that their mayfly prey, and larger organisms
experience proportionally higher drag (Vogel 1994; Statzner 1981 ).

The highly turbulent, high

Reynolds number flows on exposed surfaces of torrential stream boulders may also prevent
predators from recognising prey. The hydrodynamic cues (species-specific wave patterns generated
by prey during escape manoeuvres) used by the predatory stonefly Kogotus modestus to discriminate
prey species from non-prey species (Peckarsky and Wilcox 1989) might be disrupted in the highly
turbulent flows above the bed of torrential stream boulders (see Hart et al. 1996).
These results support the conclusions of Hart and Merz (1998) and Peckarsky et al. (1990)
who found that direct predator impacts {i.e. lethal encounters) to prey populations were decreased in
habitats which were hydrodynamically unfavourable to the predators.

During the night, when E.

/ongimanus were grazing on the upper surface of stream boulders, the degree of microhabitat overlap

between predators (which have presumably remained in interstitial spaces or on the underside of the
boulders) and prey was at a minimum. As such, the rate of predator/prey encounters should also be
lowest during the night. In this system, however, the flow-mediated refuge from predation coincided
spatially with high resource patches (periphyton densities were also greatest on the upper surface of
stones). The negative effects of flow on predators and the spatial co-occurrence of high flow and
high resource microhabitats may have combined to produce a positive indirect effect on larvae,
similar to that observed in simuliid larvae populations by Hart and Merz (1998). As a result, other
types of indirect impacts to mayfly prey populations attributed to the presence of predators may not
apply to E. /ongimanus larvae in torrential habitats. For instance, Peckarsky et al. (1993) found that
predator avoidance behaviour by Baetis bicaudatus resulted in lower feeding rates, and, as a
consequence, lower growth rates and fecundities. Similarly, Peckarsky (1996) stated that Epeorus
longimanus had reduced resource acquisition rates as a cost of predator avoidance. However, due to

183

the spatial overlap of areas of algal resource and refuge from predation, the feeding rates of E.
/ongimanus in the torrenticolous stream habitats described here were probably not reduced in this

fashion. In addition, increases in predator-induced drift, such as that reported in the mayfly Baetis by
the stonefly Doroneuria baumanni (Lancaster 1990), may not occur in E. longimanus populations due
to the low probability of predator-prey encounters during night-time feeding periods .

Peckarsky

(1996) found the presence of the stonefly predator Kogotus induced very little night-time drift in E.
longimanus.

Wooster and Sih (1995), in their review of drift and activity responses of benthic

invertebrates to the presence of predators, found that stream invertebrate drift increased when
invertebrate predators were present; however, no studies of the responses of torrenticolous taxa were
included in their analyses .
The daytime positioning of the larval E. longimanus that remained on the upper surface of the
stones may represent a trade-off between exposure to predators and exposure to UV radiation.
Several stream invertebrate taxa (e.g. simuliid larvae; Donahue and Schindler 1998) are known to be
sensitive to UV radiation. When light levels increased to detectable levels in the morning, the majority
of larval E. longimanus migrated to the underside of the stones . Only approximately 10% of the
larvae observed on the upper surface of the stones during the night remained there during the day
(Figures 4-13 and 4-14). The number of E. longimanus larvae that remained on the upper surface of
streambed stones was related to the level of incident radiation; more larvae remained exposed on
cloudy, overcast days than sunny days (personal observation). As a result, while high shear areas
may act as a flow-mediated refuge from predation, levels of UV radiation may, in turn, mediate the
use of these refugia .

Diel periodicity
The diel periodicity in positioning that was observed in E. longimanus is typical of the
behaviour of mayflies seen in streams with visually oriented predators (Mcintosh and Peckarsky
1996; Rader and Ward 1990). However, among-stone differences in the rates of dusk emergence
and dawn withdrawal from the upper surfaces of the stones suggests that the behaviour exhibited by

E. longimanus may be more than merely negatively phototactic. The timing of the emergence of E.
longimanus larvae was highly variable among stones, and larvae emerged over several hours (see

Figure 4-14). However, the withdrawal of larvae from the upper surface of all stones appeared to be
184

synchronous in that larvae withdrew to the underside of the stones within a relatively short period (< 2
h). There are two possible explanations for the apparent asymmetry in the migratory behaviour of
larval E. /ongimanus. First, the response to the proximal stimulus (light level) may vary between dusk
and dawn .

If larval E. longimanus are cued to emerge from the underside of the stones when

ambient light levels decrease below a certain threshold value, the precise tim ing of the emergence
should vary locally due to shading by adjacent stones. In sites with narrow, deep interstices between
stones, larval emergence should begin somewhat earlier. By the same rationale , the timing of larval
withdrawal should appear relatively co-ordinated . The upper surfaces of the protuberant stones,
being higher than the surrounding streambed, should be essentially unshaded by adjacent stones.
Larvae on the upper surfaces should have been exposed to similar light levels, and thus, when
ambient light exceeded the threshold value, the withdrawal behaviour should have been cued at the
same time for all individuals. Second, the dusk stimulus (light dropping below the threshold value)
cued E. longimanus larvae to migrate to the upper surface and begin grazing; as a result, they may
have immediately begun to forage as they migrated upward, reducing their travel speed . When light
levels exceeded the threshold level at dawn, larvae were cued to cease feeding and migrate to the
underside of the stone, unimpeded by incidental feeding . In this manner, speed of travel during the
dawn withdrawal of diurnally active invertebrate grazers inhabiting coarse substrates should exceed
travel speeds during dusk emergence.

Shear stress
In all cases where the stone protruded up into the flow (i.e. all stones except stone 25, which
was located beneath a small plunge/cascade), shear stress (-rw) increased gradually from the front to
the rear of the stone. Shear stresses recorded at the front of the boulder were generally low (-rw < 1.0
Pa), while maximum shear stress values were generally recorded near the rear of the stone,
immediately upstream of the separation point. Shear stress decreased abruptly at the rear of each
stone, as shear stress is zero at points of flow separation (see White 1999).
The relationship between the biology of benthic invertebrates and shear stress is generally
poorly understood, despite several prominent reviews on the biological importance of near-bed
hydrodynamics (e.g. Nowell and Jumars 1984; Davis and Barmuta 1989; Carling 1992) that

185

emphasised the potential importance of shear stress. Importantly, the distribution and magnitude of
shear stress has yet to be measured at small (em) scales over the rocky surface of naturally occurring
stream beds . This is due primarily to the difficulty in measuring 'tw in the field . Although shear stress
can be determined from the near-bed, log-linear portion of the boundary layer velocity profile, such
measurements are difficult and time-consuming to obtain in high velocity, shallow flows (see Hart et
al. 1996), and are of questionable validity in instances where boundary layer profiles are distorted by
pressure gradients (see Chapter 3).
Statzner (1981) identified a relationship between several species of benthic invertebrates and
"hydraulic stress", a factor that describes flow conditions at the surface of the substrate and is related
to the theoretical boundary layer thickness .

Recognising the potentially important role that bed

stresses might play in benthic ecology, Statzner and Muller (1989) developed a set of standardised
hemispheres (Fieisswasserstammtisch (FST) hemispheres) whose movement was correlated with
boundary shear stresses and near-bed flow conditions. However, FST hemispheres were found to
have several substantial limitations. Frutiger and Schib (1993) found that FST hemisphere movement
was more correlated to mean velocity than near-bed velocity, and that the standardised deployment
plate substantially interfered with local near-bottom microhydraulics.

In addition, Dittrich and

Schmedtje (1995) found that the relationship between FST hemisphere movement and wall shear
stress depended on bed topography.

Due to its small diameter and method of deployment, the

Preston-static tube (PST) used in this study to measure the distribution of shear stress at small
spatial scales does not experience these complications.

Compared to other devices (i.e. hot-film

anemometers; see Hart et al. 1996), the PST is relatively easy and quick to position , and is capable
of directly measuring shear stress (Ackerman et al. 1994; Ackerman and Hoover 2001 ). As a resu lt,
the results presented here represent the only study to date that directly relates the microdistributions
of benthic invertebrates and wall shear stress.
If shear stress was the primary factor influencing the microdistribution of E. /ongimanus, the
mayflies should have showed either a positive (rheophilic) or negative (rheophobic) response to
increasing shear stress during the night-time period when they were actively foraging on the upper
surface of the stones . As this was not the case, the distribution and movement of E. longimanus was
apparently not limited by the high shear stress recorded on the upper surface of the torrential stream
stones. In fact, the results show that E. longimanus larvae that remained on upper stone surfaces
186

during the day were found in high shear areas, possibly because such areas might act to exclude
visually oriented vertebrate predators such as fish (e.g. Bull Trout Salvelinus confluentus) or birds
(e.g. the American Dipper Cine/us americanus).

Salmonid predators, such as those found in the

study reach (Salvelinus confluentus; G. Cho, UNBC graduate student, unpublished data), generally
exhibit a lie-in-wait mode of predation (e.g. Hughes 1998), and generally do not actively forage in high
velocity microhabitats for prey.

Stone reversal
The results of the stone reversal experiment apparently contradict the conclusion that the
night-time microdistribution of E. /ongimanus is unrelated to hydrodynamic factors . In the original
orientation of the stones, E. /ongimanus larvae were aggregated towards the downstream side of the
upper surface of all three substrates. When the stones were rotated 180°, the larvae quickly migrated
downstream on the stone surface, adopting the same relative distribution {i.e. with respect to local
flow patterns). When the stones were again rotated 180° (returning them to their original orientation),
the larvae again migrated downstream on the stone surface, returning to their original distribution.
This demonstrates that the larvae responded proximately to hydrodynamic cues, not algal densities or
substrate characteristics. Although this result appears anomalous, the larval E. longimanus may
have responded to some aspect of the near-bed flow environment (possibly shear stress -rw. near-bed
velocity U0 .002 m. or turbulence intensity) to assess another, more directly important, ecological factor.
For example, larvae may associate high-flow areas with abundant periphyton, and make use of this
association to forage more efficiently. Alternatively, E. /ongimanus larvae may forage in high shear
stress areas to minimise predation (Hart and Merz 1998) or maximise respiration (Wiley and Kohler
1980).

Periphyton microdistribution
Periphyton density on the upper surface of the experimental stones was significantly and
positively correlated with roughness and topography {i.e. greater mass of chlorophyll a cm·2 in higher
and rougher areas of the stones). These results concur with the findings of Sanson et al. (1995) who
reported a positive association between surface roughness and chlorophyll a. Peckarsky and Cowan
187

(1995) suggested that periphytic diatoms might be less prone to scouring from rough surfaces, but
failed to propose a mechanism by which this might be possible. Rough surfaces provide a more
suitable site for colonisation of diatoms (see Sanson et al. 1995). Furthermore, flows over rough
surfaces become turbulent at lower velocities, which can, in turn, produce thicker boundary layers
and viscous sublayers. Periphyton densities are known to be higher on the upper surfaces of stones
(e.g. Vaughn 1986), a characteristic distribution produced by higher light levels (less shading and

attenuation) and greater rates of nutrient delivery.
The results suggest that E. longimanus may have had a negative impact on periphyton
density.

Although marginally insignificant, periphyton density was negatively correlated with both

daytime (p

=0.060) and night-time densities (p =0.055) of E. /ongimanus. Grazing invertebrates are

known to have impacts on local algal densities.

For example, Quinn et al. (1996) found that

periphyton density declined with increasing near-bed velocities, and attributed this result to an
increase in collector-browser invertebrates.

Similarly, Feminella et al. (1989) found that grazing

invertebrates reduced periphyton standing crops to a constant, low level, regardless of growth
conditions. Vaughn (1986) found that the trichopteran grazer Helicopsyche borealis aggregated on
the exposed upper surfaces of stones, the microhabitat with the greatest periphyton abundance, while
the results of Casey and Clifford (1989) suggest that mayfly grazers tend to aggregate on areas of the
substrate with the highest periphyton densities.
No studies to date, however, have determined if E. longimanus, or any species of torrential
grazer, locally reduce periphyton standing stock. Using the results of Cowan and Peckarsky (1994),
a conservative feeding rate estimate of 6.0 x 1o- g of pigment per animal per night can be calculated
7

for Baetis bicaudatus. Assuming that E. longimanus and B. bicaudatus graze at approximately the
same rate, the observed densities of E. /ongimanus should have a substantial impact on local
standing crop of periphyton , grazing as much as 12% of the periphyton from the upper surface of the
stones per night {based on larval densities of 400 individuals m- , and periphyton densities of 2.0 x 1o2

2
3 g pigment m- ).

This suggests, however, that E. longimanus impacts on periphyton abundance and

distribution should be more substantial than our results indicate. The limited impact of E. /ongimanus
on periphyton density is possibly due in part to mouthpart morphology; Karouna and Fuller (1992)
found that the "brusher'' mouthparts of Epeorus have little impact on diatom densities on a variety of
substrate types.
188

Conclusions
The near-bed flow conditions measured in this study were heterogeneous both among and
within stones.

The local variability in measured hydrodynamic factors (i.e. wall shear stress,

boundary layer separation) directly influenced the behaviour and distribution of the invertebrate taxa
that inhabit shallow, high-velocity streams. Spatial variation in ecologically important hydrodynamic
factors such as shear stress can influence dispersal, predator I prey interactions, and feeding.
However, the relationship between periphyton, invertebrate grazers, and near-bed flow conditions is
complex; while E. longimanus larvae responded proximally to hydrodynamic factors, the larvae in turn
may have, through herbivory, reduced periphyton in areas of the substrate where they aggregated .
However, the growth and distribution of benthic algae is also known to be directly influenced by nearbed flows (e.g. Quinn et al. 1996). Due to the complexity of the relationship between hydrodynamics,
periphyton, and benthic macroinvertebrates, studies at organism-defined spatial scales are needed
before a more complete understanding of the linkages between hydrodynamics and benthic ecology
can be gained.
In addition, the majority of our knowledge of the ecology of mobile benthic invertebrates is
based on observation of these organisms on the upper surfaces of streambed stones.

However,

most stream invertebrate taxa, including those which diurnally migrate to the upper surface of stones
to graze on algae, spend much of their lives in interstitial spaces within the streambed (e.g. Rader
and Ward 1990). Few studies have attempted to examine the activity of either predator or prey
species during the day in the interstitial spaces beneath stones. Clearly, we need to know more
about the behaviour of benthic invertebrate taxa at a wider variety of spatial scales before we can
understand the relationships that exist between flow and the distribution of benthic organisms.

189

Acknowledgements
I am indebted to many colleagues for their generously offered help. Many thanks to all the members
of the Aquatic Sciences Lab at UNBC for their technical and field assistance. Special thanks go to
my field assistants, especially Shelley Hoover and Danial Gable.
NSERC and FRBC grants to Josef Ackerman .

190

This study was supported by

References
Ackerman , J.D., L. Wong , C. R. Ethier, D.G. Allen , and J.K. Spelt. 1994. Preston-static tubes for the
measurement of wall shear stress . J. Fluids Eng . 116: 645-649.
Ackerman, J.D., and T.M. Hoover. In press. Measurement of local bed shear stress in streams using
a Preston-static tube. Limnol. Oceanogr.
Bouckaert, F.W., and J. Davis. 1998. Microflow regimes and the distribution of macroinvertebrates
around stream boulders. Freshwater Bioi. 40: 77-86.
Buffin-Belanger, T., and A.G. Roy. 1998. Effects of a pebble cluster on the turbulent structure of a
depth-limited flow in a gravel-bed river. Geomorphology 25: 249-267.
Carling, P.A. . 1992. The nature of the fluid boundary layer and the selection of parameters for
benthic ecology. Freshwater Bioi. 28 : 273-284.
Casey, R.J ., and H.F Clifford. 1989. Colonization of natural substrata of different roughness and
colour by Ephemeroptera nymphs using retrieval and direct observation techniques.
Hydrobiol. 173:185-1 92.
Collier, K. 1994. Influence of nymphal size, sex and morphotype on microdistribution of De/eatidium
(Ephemeroptera: Leptophlebiidae) in a New Zealand river. Freshwater Bioi. 31 : 35-42 .
Collier, K.J., G.F. Croker, C.W. Hickey, J.M. Quinn, and B.S. Smith. 1995. Effects of hydraulic
conditions and larval size on the microdistribution of Hydrobiosidae {Trichoptera) in two New
Zealand Rivers . New Zealand J. Mar. Freshwater Res. 29 : 439-451 .
Cowan, C.A., and B.L. Peckarsky. 1994. Diel feeding and positioning periodicity of a grazing mayfly
in a trout stream and a fishless stream . Can . J. Fish . Aquat. Sci. 51 : 450-459.
Davis, J. A., and L. A. Barmuta. 1989. An ecologically useful classification of mean and near-bed
flows in streams and rivers. Freshwater Bioi. 21 : 271-282 .
Delgado, J.A., and R. Carbonell. 1997. Case features of caddisfly larvae (Sericostoma se/yst) as
related to water velocity and potential to drift. J. Freshwater Ecol. 12: 193-197.
Dill , L.M . 1987. Animal decision-making and its ecological consequences : the future of aquatic
ecology and behaviour. Can. J. Zool. 65: 803-811.
Dittrich, A., and U. Schmedtje. 1995. Indicating shear stress with FST-hemispheres - effects of
stream-bottom topography and water depth. Freshwater Bioi. 34: 107-121 .
Donahue, W.F., and D.W. Schindler. 1998. Diel emigration and colonization responses of blackfly
larvae (Diptera : Simuliidae) to ultraviolet radiation . Freshwater Bioi. 40:357-365.
Dudgeon, D. 1982. Aspects of the microdistribution of insect macrobenthos in a forest stream in
Hong Kong. Arch . Hydrobiol. , Suppl. 64: 221-239 .
Edmunds, G.F., and R.K. Allan . 1964. The Rocky Mountain species of Epeorus {Iron) Eaton
(Ephemeroptera: Heptageniidae). J. Kansas Ent. Soc. 37: 275-288 .
Edmunds, G.F. Jr., and RD . Waltz. 1996. Ephemeroptera. in An Introduction to the Aquatic Insects
of North America - Third Edition . Eds. Merritt, Richard W. , and Kenneth W. Cummins.
Kendall/Hunt Publishing Company, lA.

191

Feminella, J.W., M.E. Power, and V.H. Resh. 1989. Periphyton resonses to invertebrate grazing and
riparian canopy in three northern California coastal streams. Freshwater Bioi. 22 : 445-457.
Frutiger, A. , and J.-L. Schib. 1993.
Freshwater Bioi. 30:463-474 .

Limitations of FST hemispheres in lotic benthos research.

Golubkhov, S.M ., T.M. Tiunova, and S.L. Kocharina. 1992. Dependence of the respiration rate of
aquatic insects upon the oxygen concentration in running and still water. Aquatic Insects 14:
137-144.
Geider, R.J., and B.A. Osborne. 1992. Algal Photosynthesis. Chapman and Hall, New York.
Growns, 1.0., and J. Davis. 1994. Longitudinal processes in near-bed flows and macroinvertebrate
communities in a Western Australian stream. J. N. Am. Benthol. Soc. 13: 417-438.
Hart, D.D., B.D. Clark, and A. Jasentuliyana. 1996. Fine-scale measurement of benthic flow
environments inhabited by stream invertebrates. Limnol. Oceanogr. 41: 297-308.
Hart, D.D., and R.A. Merz. 1998. Predator-prey interactions in a benthic stream community: a field
test of flow mediated refuges . Oecologia 114: 263-273.
Hoover, T.M., J.L. McConnachie, and J.D. Ackerman . In prep. Measuring periphyton biomass on
natural and artificial substrates of differing roughness.
Hughes, N.F. 1998. A model of habitat selection by drift-feeding stream salmonids at different
scales. Ecology 79: 281-294.
Hynes, H.B.N. 1971 . Benthos of flowing water (methods of collection). In A Manual on Methods for
the Assessment of Secondary Productivity in Fresh Waters. Eds. W.T. Edmonson and G.G.
Winberg . IPB Handbook 17: 48-59.
Jowett, I.G., J. Richardson, B.J.F. Biggs, C.W. Hickey, and J.M. Quinn . 1991 . Microhabitat
preferences of benthic invertebrates and the development of generalised Deleatidium spp .
habitat suitability curves, applied to four New Zealand rivers. N. Zea. J. Mar. Freshwater
Res . 25: 187-199.
Karouna, N.K., and R.L. Fuller. 1992. Influence of four grazers on periphyton communities
associated with clay tiles and leaves . Hydrobiol. 245: 53-64 .
Kerans, B.L. 1996. The influence of periphyton and rock texture on the diel drift of a Hydropsychid
caddisfly. J. Freshwater Ecol. 11: 163-169.
Kerans, B.L. , B.L. Peckarsky, and C. Anderson . 1995. Estimates of mayfly mortality: is stonefly
predation a significant source? Oikos 74: 315-323 .
Lamberti, G.A., and V.H. Resh. 1985. Comparability of introduced tiles and natural substrates for
sampling lotic bacteria, algae, and macroinvertebrates. Freshwater Bioi. 15: 21-30.
Lacoursiere, J.O. 1992. A laboratory study of fluid flow and microhabitat selection by larvae of
Simulium vittatum (Diptera: Simuliidae). Can. J. Zool. 70 : 582-596 .
Lacoursiere, J.O., and D.A. Craig. 1993. Fluid transmission and filtration efficiency of the labral fans
of black fly larvae (Diptera : Simuliidae): hydrodynamic, morphological, and behavioural
aspects. Can . J. Zool. 71 : 148-162.
Lancaster, J. 1990. Predation and drift of lotic macroinvertebrates during colonization . Oecologia 85:
48-56

192

Lancaster, J., A.G . Hildrew, and C. Gjerlov. 1996. Invertebrate drift and longitudinal transport
processes in streams. Can . J. Fish . Aquat. Sci. 53: 572-582.
McCormick, P. V., and R. J. Stevenson . 1989. Effects of snail grazing on benthic algal community
structure in different nutrient environments . J. N. Am. Benthol. Soc. 8: 162-172.
Mcintosh, A.R., and B.L. Peckarsky. 1996. Differential behavioural responses of mayflies from
streams with and without fish to trout odour. Freshwater Bioi. 35: 141-148.
Minshall, G.W. 1984. Aquatic insect-substratum relationships. in The Ecology of Aquatic Insects.
Eds. V.H. Resh and D.M. Rosenberg. Praeger Publishers, New York, NY.
Muotka, T., and A. Penttinen. 1994. Detecting small-scale spatial patterns in lotic predator-prey
relationships: statistical methods and a case study. Can . J. Fish. Aquat. Sci. 51 : 2210-2218 .
Nowell, A.R.M ., and P.A. Jumars. 1984. Flow environments of aquatic benthos. Ann . Rev. Ecol.
Syst. 15: 303-328.
Peckarsky, B.L. 1996. Alternative predator avoidance syndromes of stream-dwelling mayfly larvae.
Ecology 77: 1888-1905.
Peckarsky, B.L., and C.A. Cowan . 1995. Microhabitat and activity periodicity of predatory stoneflies
and their mayfly prey in a western Colorado stream . Oikos 74(3): 513-521 .
Peckarsky, B.L., C.A. Cowan, M.A. Penton, and C. Anderson. 1993. Sublethal consequences of
stream-dwelling predatory stoneflies on mayfly growth and fecundity . Ecology 74: 18361846.
Peckarsky, B.L., S.C. Horn, and B. Statzner. 1990. Stonefly predation along a hydraulic gradient: a
field test of the harsh-benign hypothesis. Freshwater Bioi. 24: 181-191.
Peckarsky, B.L., and R.S. Wilcox. 1989. Stonefly nymphs use hydrodynamic cues to discriminate
between prey. Oecologia 79: 265-270.
Quinn, J.M., C.W. Hickey, and W. Linklater. 1996. Hydraulic influences on periphyton and benthic
invertebrates: simulating the effects of bed roughness. Freshwater Bioi. 35: 301-309.
Rader, R.B., and J.V. Ward. 1990. Diel migration of microhabitat distribution of a benthic stream
assemblage. Can. J. Fish . Aquat. Sci. 47: 711-718.
Robert, A., A.G. Roy, and B. De Serres. 1996. Turbulence at a roughness transition in a depth
limited flow over a gravel bed. Geomorphology 16: 175-187.
Sanson, G.D., R. Stolk, and B.J . Downes. 1995.
A new method for characterizing surface
roughness and available space in biological systems. Funct. Ecol. 5: 127-135.
Scrimgeour. G.J ., J.M. Culp, and F.J. Wrona. 1994. Feeding while avoiding predators: evidence for
a size-specific trade-off by a lotic mayfly. J. N. Am . Benthol. Soc. 13: 368-378 .
Statzner, B. 1981. The relationship between "hydraulic stress" and microdistribution of benthic
macroinvertebrates in a lowland running water system, the Schierenseebrooks (North
Germany). Arch . Hydrobiol. 91: 192-218.
Statzner, B., and B. Higler. 1986. Stream hydraulics as a major determinant of benthic invertebrate
zonation patterns. Freshwater Bioi. 16:127-139.
Statzner, B., and T.F . Holm. 1982. Morphological adaptations of benthic invertebrates to stream flow
- an old question answered by means of a new technique (laser doppler anemometry).
Oecologia 53: 290-292 .
193

Statzner, B., and T.F. Holm. 1989. Morphological adaptation of shape to flow: microcurrents around
lotic macroinvertebrates with known Reynolds numbers at quasi-natural flow conditions.
Oecologia 78: 145-157.
Statzner, B., and R. Muller. 1989. Standard hemispheres as indicators of flow characteristics in lotic
benthos research. Freshwater Bioi. 21: 445-459.
Tabachnick, B.G., and L.S. Fidel!.
Publishing Co., New York.

1996. Using Multivariate Statistics, 3rd Edition . HarperCollins

Tikkanen, P., T. Muotka, A. Huhta, and A. Juntunen. 1997. The roles of active predator choice and
prey vulnerability in determining the diet of predatory stonefly {Piecoptera) nymphs. J. Anim .
Ecol. 66: 36-48.
Voelz, N.J., and J.V. Ward . 1996. Microdistributions of filter-feed ing caddisflies {Insecta: Trichoptera)
in a regulated Rocky Mountain river. Can. J. Zool. 74: 654-666.
Vaughn, C.C. 1986. The role of periphyton abundance and quality in the microdistribution of a
stream grazer, Helicopsyche borealis (Trichoptera: Helicopsychidae). Freshwater Bioi. 16:

485-493 .

Vogel, S. 1994. Life in Moving Fluids, 2"d edition . Princeton University Press, Princeton .
Way, C.R., A.J . Burky, C.R. Bingham, and A.C. Miller. 1995. Substrate roughness, velocity refuges,
and macroinvertebrate abundance on artificial substrates in the lower Mississippi River. J. N.
Am . Benthol. Soc. 14: 510-518.
Weissenberger, J., H.-Ch . Spatz, A. Emanns, and J. Schwoerbel. 1991. Measurement of the lift and
drag forces in the mN range experienced by benthic arthropods at flow velocities below 1.2 m
s-1 . Freshwater Bioi. 25: 21-31.
Wellnitz, T.A. , R.B. Rader, and J.V. Ward . 1996. Importance of light and nutrients in structuring an
algal community in a Rocky Mountain stream. J. Freshwater Ecol. 11 : 399-413.
Wetmore, S.H ., R.J . MacKay, and R. Newbury. 1990. Characterization of the hydraulic habitat of
Brachycentrus occidentalis, a filter-feeding caddisfly. J. N. Am . Benthol. Soc. 9: 157-169.
1

White, F.M. 1999. Fluid Mechanics, 4 h edition . WCB I McGraw-Hill, Boston .
Wiley, M.J., and S.L. Kohler. 1980. Positioning changes of mayfly nymphs due to behavioral
regulation of oxygen consumption . Can . J. Zool. 58: 618-622 .
Wooster, D., and A. Sih. 1995. A review of the drift and activity responses of stream prey to predator
presence. Oikos 73: 3-8.
Young, W.J . 1992. Clarification of the criteria to identify near-bed flow regimes. Freshwater Bioi. 28:

383-391

194

Chapter 4 - Figures

c

8

A
0

0

0
I
I
I

0

0

0

0

0

0

I>

~

0

0

Or

0

0

0

u•

I

0

0

@)@)@)@)@)
@)@@@)@
@@)@@)@)

0

0

0

0

0

•
•

............
•
0

0

0

0

0

0

•

0

0

0

0

0

0

BF

10 em

Figure 4-1. Diagrammatic representation of the upper surface of a stone, showing the distribution of
each of the types of factors sampled . A) The distribution of pre-marked points over the stone surface
(delineating a 5 x 5 em grid). The distribution of all parameters were recorded with respect to this
grid; triangular marker in the middle of the stone denotes the centre of the stone ('datum'). The
distribution of boundary layer flow conditions were also recorded (BF denotes the extent of the
backflow region associated with boundary layer detachment at the rear of the stone). B) The
distribution of algal samples (three parallel transects in the streamwise direction). The location of
each algal sample is denoted by a larger circle superimposed over a grid 'point'. C) The distribution
of shear stress measurements obtained on each stone (two orthogonal transects, crossing at the
centre of the stone). The location of each shear stress measurement is denoted by a solid symbol.
Scale bar is shown.

195

700

-

Dday

600

•night

N

--

'E 500
c

Cl)

400

·-......§-~ 300
(.)

Q)

"0

LLi

200
100
0
7

31

25

23

10

stone
Figure 4-2A. E. /ongimanus larvae density among stones, and between day- and night-time periods
(error bars = SD). Data are larval densities pooled over the entire study period. While differences in
per quadrat mayfly densities among stones during the night-time are not significant (ANOVA,
F(4,130) = 0.40, p = 0.81), differences in mayfly densities among stones during the day are (ANOVA,
F(4, 130) = 5.08, p < 0.001 ).

196

300

-

-night
C=:::J day

250

~

E
c 200
Cl)

:::3

:::.
:.;:;

~
(,)

150

~ 100
u.i
50
0

separated

attached

boundary layer flow conditions

Figure 4-28. E. /ongimanus larvae density in areas of the substrate with different boundary layer
flow conditions (error bars = SE). Larval densities quantified at small scales (within 5 x 5 em
quadrats). Data are mean larval densities, calculated from per quadrat densities measured over the
entire study period and all stones .

197

250

-

200

>.

150

--

~

E

•
•

c:

ctl

•

•

•

•
•

"'C
Cl)

::::s

:::.

:<::::

20

~

lU

100

50

0

200

400

600

800

1000

1200

1400

1600

E. deceptivus (night) (n m-2 )

Figure 4-3. Relationship between daytime and nighttime microdistributions of E. /ongimanus larvae.
Areas of the stone surface that have relatively high densities of larvae at night tend to have high
daytime densities. Data are mean larval densities, calculated from per quadrat densities measured
2
over the entire study period and all stones . (Regression line shown; R = 21 .12%.)

198

400

-

!')'

E
c
........

-night
c:::=J day

350
300
250

s 200
Cl)

~

Q.

(!)
(.)

~
LLi

150
100
50
0
20

15

10

5

-5

0

-10

-15

-20

2

streamwise distance X (X1 Q- m)

Figure 4-4. The distribution of E. /ongimanus larvae {day and night) along the centre line of the
upper surface of the experimental stones (error bars SE). Data are larval distributions pooled over
the entire study period. Daytime means were not significantly different. Nighttime means that were
not significantly (at a = 0.05) or nearly significantly (p < 0.075) different are underlined (see Table 4-2
for exact p-values). Flow is from left to right; stone centre is denoted by streamwise distance = 0.

=

199

Figure 4-5. Relationship between E. /ongimanus density (day and night) and wall shear stress (-rw).
2
(Regression lines shown; upper line daytime larval densities, R
36.15%, lower line nighttime
2
larval densities, R = 2.08%.)

=

=

200

=

-

~

E
c:

~------------~------------~

1200

~

----- night

-o- day

1000
800
600
400
200

E
(.)

2.5

C>
:::::1

2.0

~

>.

..c:
c.

e0

----

1.0
0.5

-c.
-

3.5
3.0

en
en

2.5

....

1.5
1.0

.....en~
ro

Q)

..c:
en

-~~~~~~--~

1.5

..c:
(.)

ro

~--~----~----~----

2.0

0.5

0.0 - -------- ---~-- ----20

15

10

0

5

------

-5

--~--

-----

-10

-15

-20

2

streamwise distance X (X1 0" m)

Figure 4-6. Distribution of larval E. longimanus density (error bars = SE), periphyton density (error
bars= SE), and shear stress ('tw) (error bars= SD) along the centreline transect of stone 7. Densities
of E. longimanus larvae and periphyton appear to have an inverse relationship. Shear stress ('tw)
decreases abruptly at the rear of the stone due to flow separation . Periphyton samples were
collected from the field on October 7, 1998; E. longimanus densities and shear stress measurements
were made on various dates throughout the study period (see Tables 4-1 and 3-2, respectively).

201

.........

800
------ night

~

E

-s
c:
Cl)

g.
.g
~

(.)

LJ.i

-o-- day

600
400
200

---

0

.........

~

E

(.)

-

0.8

C)

::J

co

0.7

>-

..c:

c.

e0

0.6

..c:
(.)

.........

co

a..

en
en
~

0.5
3.0
2.5
2.0

en
co
Q)

1.5

en

0.5

.....

..c:

1.0
0.0
10

5

0

-5

-10
2

streamwise distance X (X1 0- m)
Figure 4-7. Distribution of larval E. /ongimanus density (error bars = SE}, periphyton density (error
bars SE), and shear stress ('tw) (error bars SO) along the centreline transect of stone 31. Shear
stress ('tw) decreases abruptly at the rear of the stone due to flow separation. Periphyton samples
were collected from the field on October 7, 1998; E. longimanus densities and shear stress
measurements were made on various dates throughout the study period (see Tables 4-1 and 3-2,
respectively).

=

=

202

-

C)!

E

-s
c

( /)

1200

600
400

~

200

-E
-

0

u

LJ.i

C)!

(.)

5

C)

4

co

3

>.
..c:.

2

..Q

1

::J

0..
0.....

..c:.
(.)

co

a_

0
4

C/)
C/)

3

C/)

2

Q)

.....

.....
co
Q)

--o- day

800

g.

:0:::.

------ night

1000

..c:.
C/)

0
10

5

0

-5

-10

streamwise distance X (X1 Q"2 m)
Figure 4-8. Distribution of larval E. longimanus density (error bars = SE), periphyton density (error
bars = SE), and shear stress (-rw) (error bars = SD) along the centreline transect of stone 23. Shear
stress (-rw} increases as flow accelerates as it crosses the stone; no measurements were obtained in
the region of flow separation at the rear of the stone. Periphyton samples were collected from the
field on October 7, 1998; E. longimanus densities and shear stress measurements were made on
various dates throughout the study period (see Tables 4-1 and 3-2, respectively).

203

-

~

E

700

c:

500

CJ)

400

·-.......
:::1

:>

Q

(l)

u
(l)

"0

u.i

-

- e - night

600

--o- day

300
200

-----o----

100
0

~

E
(.)

-

C)

0.4

:::::J

C'O

>.

..c:
c..
0L..

.Q

..c:
(.)

C'O

a..

0.3
0.2
0.1

3

(/)
(/)

Q)

L..
......
(/)

2

L..

C'O

Q)

..c:

(/)

1
0
10

5

0

-5

-10

2

Streamwise distance X (x1 Q- m)

Figure 4-9. Distribution of larval E. /ongimanus density (error bars = SE), periphyton density (error
bars
SE), and shear stress ('tw) (error bars
SO) along the centreline transect of stone 10.
Densities of E. longimanus larvae and periphyton appear to have an inverse relationship. Periphyton
samples were collected from the field on October 7, 1998; E. longimanus densities and shear stress
measurements were made on various dates throughout the study period (see Tables 4-1 and 3-2,
respectively).

=

=

204

-

":'

E
c:

-s
Cl)

:;::

~

0

~

LLi

-

":'

E
0

C>
::I

co

>.
~

a.

0
....

..Q
~

0

co
c..

C/)
C/)

~

....
C/)

co
Q)

~
C/)

900
800
700
600
500
400
300
200
100
0

----night
--o- day

1.2
1.0
0.8
0.6

3
2
1
0
10

5

0

-5

-10

2

streamwise distance x (x10- m)
Figure 4-10. Distribution of larval E. longimanus density (error bars = SE), periphyton density (error
bars = SE), and shear stress {tw) (error bars = SD) along the centreline transect of stone 25. Flow
from upstream plunged directly on the upper surface of stone 25; as a result, shear stress is high and
relatively uniform across the upper surface. Periphyton samples were collected from the field on
October 7, 1998; E. longimanus densities and shear stress measurements were made on various
dates throughout the study period (see Tables 4-1 and 3-2, respectively).

205

-

~

E
()

1.5

C)

::::l
..._...

ctl

>.

1.0

£

a.

....0
0

£

()

0.5

20

15

10

5

0

-5

-10

-15

-20

streamwise distance X (X1 0-2m)
Figure 4-11. Periphyton (chlorophyll a) densities along the centre line of the upper surface of five
experimentally deployed stones (error bars = SE). Flow is from left to right; stone centre denoted by
streamwise distance= 0. Sample sizes at each streamwise distance vary due to inter-stone variation
in diameter.

206

1.50

-

1.25

~

E
(.)

0>

1.00

:::s

co 0.75

>.

..c:

0.

0
....
0.50
0

..c:
(.)

0.25
0.00
separated

attached

boundary layer flow conditions
Figure 4-12. Periphyton density in areas of the bed (i.e. experimentally deployed stones) with
different near-bed flow conditions (error bars = SE). Sample sizes differ due to the relatively small
area of each stone over which boundary layer separation had occurred .

207

-

~

E

c
........,

400

400

350

350

300

300

250

Cl)

·....-:::. 200
:::J

0..
Q)
(.)

-

~

150

~
u.i 100
50

0
12:00

250

l

200

E

";"
(/)

0

E

:J
........,

150 0:::

<(

100

a..

50

20:00

04:00

12:00

0

time (h)

Figure 4-13. Densities of E. /ongimanus larvae on the upper surface of the five experimentally
deployed stones (error bars = SD) during September 26 - 27, 1998. Light (photosynthetically active
rad iation; 400-700 nm) fluxes are also shown (heavy solid line).

208

-

~

E

c
.........
Cl)

350

350

300

300

l

250
200

·-......~ 150
:::s

:::.

(.)

~
LJ..i

250
200

1

150

E

:J
.........

0::

<(

100 a..

·.

50
0
12:00

E

";"
(/)

0

·.

100

-

~

50

20:00

04:00

12:00

0

time (h)

Figure 4-14. Densities of E. longimanus larvae on the upper surface of the five experimentally
deployed stones (error bars = SO) during October 3 - 4, 1998. Light (photosynthetically active
radiation ; 400-700 nm) fluxes are also shown (heavy solid line).

209

J::

~

E
c

>.....
·u;
c

Q)

"C

80
60
40
20

Cl)

0

:0::.
Q

-20

"0

-40

c

-60

~

Q)
(.)
Q)

u.i

Q)

Cl

c

C13

..c
(.)

-80
-100
12:00

16:00

20:00

00:00

04:00

08:00

12:00

time (h)

Figure 4-15. Change in E. /ongimanus density (error bars= SE) recorded on the upper surface of the
experimentally deployed stones during October 3 - 4, 1998. Periods of light and dark represented by
bar at bottom of figure.

210

u¢

A.

¢u

500
400

B.

u¢

500
400

20

15

10

5

0

-5

-10
2

streamwise distance X (x1 0- m)

c.

211

-15

-20

Figure 4-16 (previous page). The effect of stone reversal on the positioning of E. longimanus
larvae. (A) In their original orientations of the three stones (0°), larvae were at highest densities over
the rear of the three stones. (B) After the stones were rotated 180°, the larvae moved downstream to
new positions on the stone surface, although the distribution of larvae with respect to flow was similar
(i.e. at highest densities over the rear of the stone). (C) When the stones were rotated back to their
original orientations (0°), the larvae returned to their original distributions. E. /ongimanus distributions
recorded approximately every 40 minutes (A: 18:58 h, 8: 19:48 h, C: 20:12 h). Open arrows indicate
direction of flow.

212

Chapter 5

Torrential stream microhabitats: flow, form, and function

Summary
1. Hydrodynamics are generally thought to be the single most important factor influencing the
biology of benthic organisms.

However, the near-bed flow microenvironments inhabited by

torrential stream fauna are poorly understood, and currently accepted models of near-bed flows
may, in some cases, be misleading.
2.

While near-bed flows in open channels with geometrically simple beds are predictable, flows in
shallow streams with complex geometry and large relative roughness (e.g. rough-turbulent flows)
are not. Pressure gradients, flow compression, and boundary layer separation ensure that the
distribution of velocity in the near-bed region may rarely, if ever, conform to the standard,
logarithmic model of boundary layer velocity distribution.

3. Where local protruding obstacles constrict flow, the near-bed portion of the boundary is distorted,
creating a "wedge-shaped" velocity profile.

Invertebrates inhabiting these regions of the

streambed will experience greater lift, drag, and shear forces than predicted by normal boundary
layer theory.
4.

Epeorus longimanus (Heptageniidae), a torrential mayfly species, exhibits behavioural and

morphological adaptations that allow it live in high velocity, high shear environments.

The

flattened shape and ventral, sucker-like gill arrangement of E. /ongimanus act to minimise the risk
of accidental detachment from the upper surface of torrential stream boulders. The high velocity,
lift, drag, and shear stress of this environment may also, in turn, act to decrease predation (by
excluding both invertebrate and vertebrate predators) and increase food availability (by
increasing rates of nutrient delivery to periphyton).
5. The ecological implications of local heterogeneity in near-bed velocity (Uo.oo 2m) and wall shear
stress (tw) are discussed.

213

Physical - biological coupling in the stream environment
It is generally thought that hydrodynamic parameters are the most important environmental
factors affecting the ecology of organisms in lotic habitats (e.g. Davis and Barmuta 1989; Carling
1992; Hart and Finelli 1999). Several studies have demonstrated that the forces of moving water
influence the morphology, behaviour, distribution, trophic relationships, and intra- and interspecific
interactions of benthic stream organisms (e.g. Chance and Craig 1986; Weissenberger et al. 1991 ;
Collier 1994; Hart et al. 1996). However, despite the fact that the ecological importance of open
channel hydraulics is widely accepted , the current understanding of stream flows at organism-defined
spatial scales remains limited, especially in torrential stream environments.
Knowledge of the relationships that exist between flow microenvironments and stream biota
is based largely on fine-scale measurements obtained in laboratory flumes (e.g. Statzner and Holm
1982; 1989; Weissenberger et al. 1991) and relatively coarse measurements of flow in naturally
occurring stream channels (e.g. Wetmore et al. 1990; Statzner and Higler 1986; Collier et al. 1995;
but see Hart et al. 1996). Laboratory-based studies use flow channels (flumes) to control basic flow
variables (i.e. mean velocity, depth, turbulence intensity, etc.) and to minimise the influence of
secondary flows and channel-specific flow patterns. As a result, flow in these channels is usually
characterised by well-defined , fully developed, logarithmically distributed boundary layers with
min imal turbulence, and bulk flow patterns that are approximately two-dimensional (see Statzner and
Holm 1989; Weissenberger et al. 1991 ). Problems arise, however, when the results of these studies
are used to extrapolate the near-bed flows of natural channels, and velocity distributions are
assumed to approximate the standard logarithmic shape (see Carling 1992}.

However, this

assumption is not valid in the rapidly varied, highly three-dimensional flows of torrential streams.
Measurement of near-bed flows at organismal scales is needed to elucidate the relationship between
benthic organisms and the hydrodynamic factors that define torrential stream environments .

The relationship between mean and near-bed flows
Vertical distributions of velocity in channels are generally thought to conform to a standard
logarithmic shape (Carling 1992; White 1999). As such, the relationship between mean and near-bed
velocities is reasonably predictable in geometrically simple (flat-bedded and walled) channels of
214

known depth. However, the flow separation, local accelerations, and pressure gradients associated
with topographic irregularities and complex bed geometries lead to complex, often unpredictable,
distributions of velocity in natural streams .
Mean velocities are correlated with velocities measured 2 em above the bed (Figure 5-1A)
across a variety of stream habitat types (i.e. riffles, pools, etc.); velocities at this height above the bed
are approximately half of mean velocity. However, when mean velocity is compared to the velocity
measured immediately adjacent to the bed (Uo.oo 2 m. measured 2 mm above the bed), the two
determinations of velocity are approximately equal (Figure 5-1 B).

This suggests that where

roughness elements protrude substantially into the flow of shallow channels, near-bed flows are
accelerated to near-mean (or even exceeding mean) velocities. This demonstrates that the boundary
layers (defined as the height above the boundary where velocities reach 90% of mean velocity) over
the upper surface of protuberant stones are, in many cases, thinner than 2 mm. While benthic
invertebrates in many stream habitats (i.e. gravel riffles) may be protected from mean velocities by
thick boundary layers (Hart et al. 1996), invertebrates on the upper surface of torrential boulders may
not be afforded such protection.

Near-bed flows in torrential streams
The results of the present study support the development of a model to describe the
distribution of velocity within boundary layer flows above isolated roughness elements in shallow
streams (where D < 3d).

Flow approaching protuberant boulders in shallow torrential streams is

generally turbulent due to the downstream convection of vortices detached from the lee side of
upstream boulders (Hart et al. 1996). The vertical distribution of near-bed velocities approaching
boulders is dependent on upstream bed geometry (for illustrative purposes will be assumed to be
approximately log-normal; Figure 5-2). The vertical distribution of mean velocities higher in the water
column would be expected to be relatively homogenous, as a result of efficient momentum transfer in
the turbulent, high Reynolds number flow.

As flow reaches the front of the boulder, a solenoidal

("horseshoe") vortex may form, provided that flow interference from upstream roughness elements is
not too great (Young 1992; 1993).

If the stone provides a 'bluff-body' obstacle to the flow, a

stagnation point will form at the leading edge of the stone as well.
215

Over the upper surface of the leading edge of the boulder, patterns of water movement will
be three-dimensional as the flow splits and flows over and around the stone. Flow that crosses the
upper surface of the stone will be constricted by the rising, inclined leading edge of the stone.
However, the flow will not accelerate uniformly in the vertical direction; acceleration is evident
primarily in fluid adjacent to the bottom, producing a distinct "wedge-shaped" profile (Figure 5-2, also
see chapter 3). Two factors may contribute to the development of this unusual velocity distribution .
First, near-bed flows are accelerated due to flow constriction rather than gravity. This conclusion
(that flow near the bottom accelerates rapidly while flow near the surface remains relatively
unaffected) is supported by the jet-like properties (namely a high-velocity core region; see White
1999) of the near-bed "wedge" portion of the velocity profile (see chapter 3).
weathering of substrates

Second, due to

in high-velocity streams, naturally occurring boulders are often

characteristically rounded . The gradually curving leading edge of the front of the stones may be an
essential element for the development of wedge-shaped velocity distributions. Nowell and Church
(1979) did not find any evidence of a near-bed velocity profile distortion of this type when they
recorded velocity profiles over equally sized, but rectangularly shaped, objects. Gradual constriction
of the flow (and the development of the near-bed "wedge") may not occur when an inclined face is not
presented to the flow. If wedge-shaped velocity profiles are the result of near-bed flow constriction ,
the distortion should be greatest at points where flow constriction is greatest. The effects of the 'noslip' condition cannot be ignored, however; boundary layer growth (and retardation of near-bed
velocity) occurs downstream of the stagnation point due to the effects of viscosity.
As flow crosses the centre of protuberant stones in shallow flows, the velocity profile
distortion that was created by constriction over the leading edge of the stone begins to dissipate,
presumably due to momentum transfer in the free shear layer above the high-velocity wedge and
frictional effects and shear stress in the boundary layer below the velocity distortion. When the
thickness of the boundary layer exceeds the height above the substrate where the magnitude of the
near-bed distortion is greatest {the 'tip' of the 'wedge'), the distortion disappears completely. If the
streamwise diameter {length) of the stone is shorter than the distance over which the 'wedge' is
dissipated, the near-bed distortion is advected off the rear of the stone
Although the development of wedge-shaped profiles in natural channels has not been
discussed previously in the literature, examination of the results of earlier studies reveals that this

216

type of near-bed velocity profile distortion has been measured previously. Given that velocity at the
streambed-fluid interface must be zero, four of the five stones above which Hart et al. (1996)
measured flow fields have velocity profiles that, by the criteria used in chapter 3, would be
categorised as wedge-shaped. Similarly, of the 15 serial profiles measured by Bergeron ( 1994) along
a 3.6 m shallow stream reach, one of the two profiles (his profile 4) obtained over the leading edge of
protuberant stones has a near-bed high-velocity wedge. It is difficult to determine if additional profiles
measured by Bergeron (1994) also have near-bed velocity distortions, as he was unable to measure
velocities within 1.9 em of the bed . In the present study, the velocity profile distortions producing the
wedge were generally located within 1 em of the bed . It is likewise difficult to assess the shapes of
the velocity profiles measured by Hart et al. (1996) because they neither provide the portions of the
profile more than 1 em from the bed nor a measure of mean velocity.

Ecohydraulics of Epeorus longimanus
In shallow flows with relatively large, protuberant stones, the marked spatial heterogeneity of
near-bed hydrodynamic parameters (notably near-bed velocity and wall shear stress) will have
substantial impacts on the distribution and biology of benthic organisms . The highest densities of E.
longimanus larvae were often found on the upper surface of torrential stream boulders (especially

during the night}, where near-bed velocities and shear stresses were generally high. Where near-bed
velocity profile distortions ('wedges') increased near-bed velocity gradients, the lift and drag
experienced by mayfly larvae found on the exposed upper surface of boulders in torrential streams
will be substantially higher than in regions of the bed with comparable mean velocities but thicker,
normally distributed, boundary layers.
Extreme hydraulic habitats do not appear to limit the movement or behaviour of this mayfly.
E. longimanus larvae were observed to move easily about on areas of the substrate where near-bed
1

velocities (Uo.oo 2 m} exceeded 1.0 m s- . Morphological and behavioural adaptations effectively counter
the forces of lift, drag, and shear stress experienced by larvae in their preferred nighttime habitat.
The ventral gills of larvae are arranged in a 'sucker-like' arrangement, their flattened, streamlined
bodies have a relatively low drag coefficient (Weissenberger et al. 1991 ), and the reduction in number
of anal cerci (from three to two) can also be regarded as a further adaptation to reduce drag.

217

Furthermore, the efficacy of the morphological adaptations of this species may be augmented if
coupled with behavioural modifications of body position that further reduce hydraulic stresses (as for
Simulium vittatum ; Chance and Craig 1986).

Although the physical means by which these

adaptations allow E. longimanus to maintain contact with the substrate are poorly understood, it is
apparent that they allow this mayfly to thrive in high velocity, high shear environments.
Although invertebrates in high velocity habitats must incur energetic costs in order to maintain
position, the results of the present study suggests that near-bed flow regimes of this type also confer
substantial ecological benefits to those taxa with the requisite adaptations. A conceptual model of the
relationship between hydrodynamic factors, predators, resource availability, and E. longimanus is
shown in Figure 5-3. Invertebrate predators, which are generally adapted for locomotory efficiency
rather than tenacity (Hart and Merz 1998), are unable to forage for prey (in this case, E. longimanus)
in the high velocities to which the upper surfaces of torrential stream boulders are exposed . In this
fashion, high velocity, high shear patches of the substrate may act as flow-mediated refuges from
predation (Hart and Merz 1998; Peckarsky et al. 1990). Due to the direct, negative influence of flow
on predators, flow may have an indirect, positive effect on local E. longimanus populations. The
steep velocity gradients and high shear stresses of this environment also increase the rates of
nutrient and dissolved gas delivery to the substrate surface, enhancing the growth of periphyton
(especially diatomaceous species), which in turn act as a food resource of larval E. longimanus. This
provides another pathway by which flow has indirect, positive effects on this mayfly.

Through

herbivory, larvae may have a negative impact on algal densities. However, because the near-bed
flow regime also tends to exclude other grazing invertebrate taxa, interspecific competition for
available algal food resources may be low, and total impacts may actually be quite limited. Finally,
flow may also have a beneficial, direct influence on E. longimanus larvae. High near bed velocities
and shear stresses may act to increase the respiratory efficiency of this rheophilous mayfly (see
Golubkhov et al. 1992).

Stream hydraulics I invertebrate interactions
The results presented here are possibly the first report of shear stress distribution patterns
across the surface of naturally occurring stream bedforms {i.e. generally, front-to-back asymmetry,
218

with shear stresses increasing from the front to the rear of the stone). As such, it may be necessary
to re-evaluate the conclusions drawn by previous studies examining the microdistribution of stream
For example, Wetmore et al. (1990), studying the microdistribution of the suspension-

insects .

feeding caddisfly Brachycentrus occidentalis, concluded that Froude number was the best predictor
of larval position . However, as shown in chapter 3, shear stress and Froude number both generally
increased from the front of torrential stream boulders to the rear. Near-bed flow parameters may be a
valid , untested alternative explanation for the microdistribution of suspension-feeding benthic insects.
In order to estimate near-bed flows from mean velocity measurements, Wetmore et al. (1990)
assumed, as have many recent studies, that the water column accelerates uniformly. The results
presented here show that this assumption, if used to estimate near-bed flow conditions in shallow
streams with complex bed geometries, could lead to substantial errors.

Similarly, Delgado and

Carbonell (1997) found that the case morphology of the caddisfly Sericostoma selysi varied with
respect to water surface velocity, and subsequently attributed the differences as a response to
counter the forces of drag. However, the results of this study suggest that the near-bed velocities to
which benthic insects are exposed may, in fact, be only weakly related to mid-stream or near-surface
velocities .

As a result, the extent to and mechanisms by which shear, lift, drag, and near-bed

velocities modify the behaviour and morphology of benthic insects remain largely unstudied .
Moreover, the relative importance of the various hydraulic forces to which stream benthic insects are
exposed remains poorly understood . The near-bed flows recorded in this study are variable over
short distances, a factor that will influence the ecology of organisms inhabiting the exposed surfaces
of boulders in high gradient streams.

While some of the observed flow patterns conformed to

commonly applied hydraulic theory (e.g. flow accelerating and becoming supercritical as it crosses
boulders), some did not (e.g. wedge-shaped velocity profiles).
With a few notable exceptions, few studies have explored the near-bed flow environments of
streams at the scales at which benthic invertebrates experience them . Even fewer studies have
examined the range and distribution of hydraulic parameters across the streambed at these
organism-defined scales.

Nowell and Jumars (1984) suggested that in shallow streams with

topographically complex beds (where D < 3k), the three-dimensional patterns of flow generated would
be of greater importance to the ecology of benthic invertebrates than average boundary skin friction .
The results of the present study show that boundary flow conditions are, indeed, heterogeneous and
219

highly dependent on local bed geometry. As flow accelerates over the upper surface of stream-bed
stones, shear stress at the boundary increases; however, the relationship is modified substantially by
the shape of the substrate over which the water is flowing . Both flow constriction (which accelerates
near-bed flow at a greater rate than at mid-depth) and thin boundary layers (a result of high near-bed
velocities or short distances available for boundary layer development) define the near-bed flow
environment.
These findings, however, provide only a limited understanding of the relative importance of
the various near-bed flow parameters to the ecology of the stream invertebrates that inhabit the upper
surfaces of streambed stones. In comparison, the relative importance of the many hydraulic forces to
which marine benthic invertebrates are exposed has been determined for several intertidal species.
For example, Denny et al. (1985) demonstrated that acceleration reaction may perhaps be the most
important force limiting the distribution of relatively large intertidal taxa, while Denny (1987) showed
that lift may be the most important factor determining the distribution of intertidal mussels. However,
coastal marine systems are very different from stream systems in that the flows, while of comparable
velocity, are intermittent and bi-directional.

In comparison, flows in stream habitats are relatively

constant and unidirectional, and as such, the relative ecological importance of the various near-bed
forces will be different. The forces of lift, drag, and shear will all tend to dislodge invertebrates from
the substrate, but their relative importance is poorly understood.
The boundary layers measured over the upper surface of the boulders were often extremely
thin (-1 mm), supporting the conclusions of Statzner and Holm (1982) that benthic insects in streams
are exposed to essentially free-stream velocities.

In fact, at locations on the stream bed where

velocity profiles were wedge-shaped rather than logarithmically distributed, benthic insects would be
exposed to velocities far greater than those predicted by normal boundary layer theory. Hart et al.
(1996) stated that, in shallow, high velocity stream flows the thickness of the benthic boundary layer
was of sufficient thickness that benthic invertebrates (in their case, larval blackflies) would not
experience free-stream current velocities. However, the velocity profiles recorded by Hart et al. (their
Figure 1) would generally be classified as log-normal by the conservative

0 (vertical velocity

variation) index used in this study.
Buffin-Belanger and Roy (1998) showed that the lee side of pebble clusters represents a
distinct hydraulic habitat with specific flow characteristics, and further suggested that stream
220

invertebrates might utilise this reg ion as a refuge .

Similarly, Bouckaert and Davis (1998)

demonstrated that distinct macroinvertebrate communities inhabit the horseshoe vortex region in front
of and the wake region behind stream boulders. The results presented here suggest that upper
surface of stream boulders may be utilised in a similar fashion by a distinct community of benthic
macroinvertebrates.

The locally high shear stresses, thin boundary layers, and high near-bed

velocities will act to exclude macroinvertebrate predators that are poorly adapted to high flow
environments (Hart and Merz 1998; Peckarsky et al. 1990), while at the same time providing
abundant grazing for herbivorous insects which are adapted to the high lift and drag of torrential
environments (Weissenberger et al. 1991; Vogel 1994; Delgado and Carbonell 1997). The turbulent
nature of shallow, high velocity flows might also prevent fish predators from predating on insects
grazing on the upper surface of torrential stream boulders (see Metcalfe et al. 1997). Salmonids ,
which are the primary fish predators of aquatic insects in many northern mountain streams, typically
inhabit pools below riffles and feed on drifting insects (Hughes 1998) in order to minimise the
energetic costs associated with swimming and maintaining position in high velocity flows . As such,
shallow, high velocity habitats may provide a refuge from fish predators as well.

Directions for future research
Increasingly, ecologists are adopting the view that flow is the most important factor in the
stream environment, and are incorporating measurements of near-bed hydrodynamic parameters into
their experimental designs. However, before stream ecologists can decide which factors to measure
and ascertain the most efficient way to measure them, they must have a comprehensive
understanding of the nature of near-bed flows in naturally occurring stream channels. Before this is
possible, there are two principal areas where substantial research is required . First, new models of
stream function that incorporate patterns of flow at small, organism-defined scales must be
developed. While currently utilised models of flow in stream channels can be used to adequately
predict large-scale phenomenon such as bedload movement and mean velocity, they are less useful
in cases when they are used to describe or predict the nature of the near-bed flows that define the
immediate environment of both macroinvertebrate taxa and periphyton. New models of stream flow
will provide not only an understanding of near-bed flow velocities, but also of turbulence, shear

221

stress, and the role of substrate/flow interactions. Second, stream ecologists must be prepared to
investigate the behaviour and ecology of benthic organisms at the scales at which they experience
flows.

While there are technical difficulties associated with this type of research , the potential

rewards are substantial. Lotic research at appropriate spatial scales will not only provide ecologist
with a greater understanding of the nature of biological-physical coupling in these environments, but
will also provide a theoretical basis that can be utilised to mitigate human impacts to rivers and
streams.

222

References
Bergeron, N.E. 1994. An analysis of flow velocity profiles, stream bed roughness, and resistance to
flow in natural gravel bed streams . Proc. Hydraul. Eng. (1994), ASCE . 1: 692-696.
Bouckaert, F.W., and J. Davis. 1998. Microflow regimes and the distribution of macroinvertebrates
around stream boulders. Freshwater Bioi. 40 : 77-86 .
Buffin-Belanger, T., and A.G. Roy. 1998. Effects of a pebble cluster on the turbulent structure of a
depth-limited flow in a gravel-bed river. Geomorphology 25: 249-267 .
Carling, P.A. . 1992. The nature of the fluid boundary layer and the selection of parameters for
benthic ecology. Freshwater Bioi. 28: 273-284.
Chance, M.M., and D.A. Craig. 1986. Hydrodynamics and behaviour of Simuliidae larvae (Diptera).
Can . J. Zool. 64:1295-1309.
Collier, K. 1994. Influence of nymphal size, sex and morphotype on microdistribution of De/eatidium
(Ephemeroptera: Leptophlebiidae) in a New Zealand river. Freshwater Bioi. 31: 35-42.
Collier, K.J., G.F. Croker, C.W. Hickey, J.M. Quinn, and B.S . Smith . 1995. Effects of hydraulic
conditions and larval size on the microdistribution of Hydrobiosidae (Trichoptera) in two New
Zealand Rivers. New Zealand J. Mar. Freshwater Res. 29: 439-451.
Davis, J. A. , and L.A. Barmuta. 1989. An ecologically useful classification of mean and near-bed
flows in streams and rivers. Freshwater Bioi. 21 : 271-282 .
Delgado, J.A., and R. Carbonell. 1997. Case features of caddisfly larvae (Sericostoma selysl) as
related to water velocity and potential to drift. J. Freshwater Ecol. 12: 193-197.
Denny, M. 1987. Lift as a mechanism of patch initiation in mussel beds . J. Exp. Mar. Bioi. Ecol. 113:
231-245.
Denny, M.W., T.L. Daniel, and M.A.R. Koehl.
organisms. Ecol. Monogr. 55: 69-102.

1985.

Mechanical limits to size in wave-swept

Golubkhov, S.M., T.M. Tiunova, and S.L. Kocharina. 1992. Dependence of the respiration rate of
aquatic insects upon the oxygen concentration in running and still water. Aquatic Insects 14:
137-144.
Hart, D.O., and C.M . Finelli. 1999. Physical-biological coupling in streams: the pervasive effects of
flow on benthic organisms. Ann. Rev. Ecol. Syst. 30: 363-395.
Hart, D.O., B.D. Clark, and A. Jasentuliyana. 1996. Fine-scale measurement of benthic flow
environments inhabited by stream invertebrates. Limnol. Oceanogr. 41: 297-308.
Hart, D.O., and R.A. Merz. 1998. Predator-prey interactions in a benthic stream community: a field
test of flow mediated refuges. Oecologia 114: 263-273.
Hughes, N.F. 1998. A model of habitat selection by drift-feeding stream salmonids at different
scales. Ecology 79: 281-294 .
Metcalfe, N.B., S.K. Valdimarsson, and N.H.C. Fraser. 1997. Habitat profitability and choice in a sitand-wait predator: juvenile salmon prefer slower currents on darker nights. J. Anim. Ecol. 66:
866-875.
Nowell, A.R.M ., and M. Church . 1979. Turbulent flow in a depth-limited boundary layer. J. Geophys.
Res. 84: 4816-4824.
223

Nowell, A.R.M., and P.A. Jumars. 1984. Flow environments of aquatic benthos . Ann . Rev. Ecol.
Syst. 15: 303-328.
Peckarsky, B.L. , S.C. Horn, and B. Statzner. 1990. Stonefly predation along a hydraulic gradient: a
test of the harsh-benign hypothesis. Freshwater Bioi. 24 : 181-191 .
Statzner, B., and B. Higler. 1986. Stream hydraulics as a major determinant of benthic invertebrate
zonation patterns. Freshwater Bioi. 16: 127-139.
Statzner, B. , and T.F. Holm. 1982. Morphological adaptations of benth ic invertebrates to stream flow
- an old question answered by means of a new technique {laser doppler anemometry).
Oecologia 53: 290-292 .
Statzner, B., and T.F. Holm. 1989. Morphological adaptation of shape to flow: microcurrents around
lotic macroinvertebrates with known Reynolds numbers at quasi-natural flow conditions .
Oecologia 78: 145-157.
Vogel, S. 1994. Life in Moving Fluids, 2"d Edition . Princeton University Press, Princeton, NJ .
Weissenberger, J., H.-Ch. Spatz, A. Emanns, and J. Schwoerbel. 1991 . Measurement of the lift and
drag forces in the mN range experienced by benthic arthropods at flow velocities below 1.2 m
s·1 . Freshwater Bioi. 25: 21-31.
Wetmore, S.H., R.J . Mackay, and R.W. Newbury. 1990. Characterization of the hydraulic habitat of
Brachycentrus occidenta/is, a filter-feeding caddisfly. J. N. Am . Benthol. Soc. 9: 157-169.
Young, W.J. 1992. Clarification of the criteria to identify near-bed flow regimes . Freshwater Bioi. 28 :

383-391 .

Young, W.J. 1993. Field techniques for the classification of near-bed flow regimes . Freshwater Bioi.

29 : 377-383.

1

White, F.M. 1999. Fluid Mechanics, 4 h edition . WCB I McGraw-Hill, Boston.

224

Chapter 5- Figures

~

-

0.90

E 0.80
..._..
E

0.70

ci

0.60

N
0

::::>

•

....>. 0.50

•

•

T5 0.40
0
Q)

> 0.30

"0
Q)

0.20

.!...

0.10

..c

ro
Q)
0.00
c
0.20

•

• •

2

R = 45.47%

0.40

1.00

0.80

0.60

1.20

1

mean velocity U (m s" )

A

.

~

(/')

1.4

E 1.2
E

N
0
0

ci
::::::1

-

>-

"(3

0

•

1
0.8
0.6

> 0.4

"'0
Q)

..c 0.2

~-

.!...
Q)

c

.
... •
•

•

~

Q)

('IJ

•

~
~

0
0

0.2

;

~

~

~

0.4

0.6

2

R =49.14%
0.8

1.2

1.4

1.6

1

mean velocity U (m s- )

8
Figure 5-1. The relationship between mean and near-bed flows. In both cases, the solid line
represents the regression line fitted to the data, while the dashed line represents the 1:1 regression
line (line of equality). A. Near-bed velocity (Ua.o2 m; velocity measured at 2 em above the bed) versus
mean velocity for 48 sites in two adjacent, high discharge streams (see Chapter 2). Predictably,
near-bed velocities are correlated with, but lower than, mean velocities . B. Near-bed velocity U0 .002 m;
velocity measured at 2 mm above the bed) versus mean velocity (U; measured at 50% of depth} for
21 sites above the upper surface of protuberant boulders (see Chapter 3). Near-bed velocities are
correlated with, but similar to, the mean velocities .

225

A
B
~ -- -- --- ----

C

-------

D

E
-------

~

Umean

Figure 5-2. Diagrammatic representation of the flow field above isolated roughness elements in
shallow, torrential streams (where D < 3d; e.g., rough-turbulent flow) . The approaching flow (A) has
two regions ; a bulk flow zone (Umean) where velocity is relatively homogenous due to high rates of
momentum transfer in the high Reynolds number flow, and a bed layer (BL) zone, where velocity is
retarded due to viscous and form drag effects at the bed/fluid interface. Flow in the stagnation region
(B) is characterised by a solenoidal ('horseshoe') vortex (HV), and a stagnation point in the front of
the stone. Over the leading edge of the stone (C), near-bed flows are influenced by the growth of a
boundary layer downstream of the 'leading edge' of the stone, and acceleration of the near-bed
portion of the water column due to constriction of the relatively shallow flow by the continuously
curving stone surface. The two processes combine to produce a small 'kink' or 'wedge' in the nearbed portion of the velocity profile. As the distance from the leading edge of the stone grows, so does
the boundary layer thickness. When flow is no longer being constricted (D), momentum transfer in
the free shear layer above the 'wedge' and the continuously velocity-retarding influence of the
boundary layer below the 'wedge' act to reduce the magnitude of the difference between mean and
maximal (at the tip of the 'wedge') velocities . As a result, the wedge begins to dissipate. At the rear
of the stone in the flow separation zone (E), the boundary layer separates, resulting in a backflow
(BF), or attached vortex.

226

Food resources
(periphyton)

<011111---

+
Predators

I_----------

-----.J

+

Figure 5-3. Diagram of the direct (solid arrows) and indirect (dashed arrows) relationships between
flow, Epeorus longimanus, primary producers, and predator abundance, with the effect of each
pathway indicated by a + (positive effect) or- (negative effect). The relative position of each factor
signifies their relative roles; flow is always a cause and is positioned on the left, E. /ongimanus
abundance is generally an effect and is positioned on the right, periphyton and predators both act as
causes and exhibit effects; as such , they are positioned in the middle (adapted from Hart and Merz
1998).

227

Appendix 1
Notation and Formulae
Notation and formulae used throughout the thesis are listed below.

A) Notation
2

p
\fl
K

A

cd
c1
d

dk
D

D/kave
Fd

F1

Fr
g
h
k

kave
ks

I

Lmax

Lave
Lso

m

p

Q

R

Re
Reb
Rex
Re*

u
0

Uz

Uo.so
Uo.oozm
Uo.ozm

U*
x*
y*

z

Zo

1

coefficient of kinematic viscosity (m s" )
boundary layer thickness (m)
viscous sublayer thickness (m)
critical boundary layer thickness (m)
wall shear stress, also denoted as -r 0 (Pa)
3
density of water (kg m" )
particle protrusion (as a proportion of diameter)
von Karman 's constant (empirically determined to be 0.40)
constant used in the determination of -rw (see chapter 3)
drag coefficient (dimensionless)
lift coefficient {dimensionless)
diameter of substrate elements (m)
functional diameter of substrate elements (m)
total channel depth (m)
relative roughness {dimensionless)
force due to drag (N)
force due to lift (N)
Froude number {dimensionless)
2
acceleration due to gravity (9.81 m s" )
head (m)
roughness height (m)
average roughness height (m)
Nikuradse's roughness length (m)
characteristic length scale (m)
maximum stone length (in quadrat) (m)
average stone length (of four largest stones in quadrat) (m)
standard deviation of stone length (four largest stones in quadrat)
Slope of a regression (used to determine U*; see below and Appendix 2)
2
pressure term (N m" )
3
1
discharge (m s" )
hydraulic radius (m)
Reynolds number {dimensionless)
channel Reynolds number {dimensionless)
local Reynolds number {dimensionless)
Reynolds roughness number {dimensionless)
1
mean channel velocity (m s" )
vertical velocity variation index (dimensionless)
1
velocity at height above the boundary z (m s· )
1
velocity measured at 50% of depth (m s" )
1
velocity measured 2 mm above the bed (m s" )
1
velocity measured 2 em above the bed (m s" )
1
shear velocity (m s· )
non-dimensionalised pressure difference {dimensionless)
non-dimensionalised shear stress {dimensionless)
distance from the boundary (m)
roughness height (m)

228

B) Formulae
(1-1) (also equation 2-1)
(1-2) (also equation 2-2)
(1-3)
(1-4)
(1-5)
(1-6)
(1-7)
(1-8)
(1-9)
(1-10)
(1-11)
( 1-12)
(3-1)
(3-2)
(3-3)
(3-4)
(3-5)
(3-6)
(3-7)
(3-8)
(3-9)
(3-1 0)
(3-11)

Re = Ullu
5
Fr = Ul(gD)0'
Rex= Uxxlu
2
tw = pU*
Uz = U*IK ln[(z-d)IZo]
U* = K [(u1 - u2)l(ln z1 - In z2)]
U* = K m
UIU* = 5.75 log(12 Dlk)
Re* = U*k5 1v
8' = 11 .5v I U*
2
Fd = Y:z Cd pSU
2
F1 = Y:z C1 pSU
Head= -0.00164 + (0.0508 x voltage)
P + 112pU2 + pgh = constant
u= ~
2
2
x* = log 10 ~ x d )1(4pu )]
2
2
y* = log 1o [(twX d )1(4pu )]
y* = -1 .394 + 718 x*

tw =A X ~
A= 1 I [10 1'394 x

~
= ~
0 = (Uo.so- Uo.oo2m) I 7Uo.so
x = Ocrrt I (0.16 I Re/ )
~

Dlkave = DI[0.5(Lave)]

note: assumes '¥=0.5

Note: the commonly determined bed roughness factor 'relative roughness' (Dikave) is more accurately
designated 'relative depth' or relative submergence (see Nowell and Church 1979), as depth (D) is
non-dimensionalised by the roughness factor kave· However, as is it an indicator of the degree of
protrusion of the bed elements into the flow, this ratio is often referred to in this work as relative
roughness, following the nomenclature established in Davis and Barmuta (1986).

229

Appendix 2
Near-bed velocity profiles and wall shear stress (tw) estimation
At a given point in a channel , the shear stress acting on the wall ('rw) can be estimated from
velocity profile data using the 'law of the wall ' (see Vogel 1994; Wilcock 1996; and Biron et al. 1998
for summaries). The law of the wall for the 'overlap layer' or 'log layer' is often expressed as
( 1)

Uz = U*/K ln{z!Zo)

where Uz is the horizontal velocity at a given elevation z above the boundary, U* is the shear velocity
(not a measure of velocity in the standard sense, but rather a nominal velocity related the rate at
which velocity changes in the vertical direction), K is von Karman's constant (empirically determined
to be 0.40; see White 1999), and Zo is the roughness height (a parameter whose value is dependent
on the roughness of the bed). As the roughness elements associated with the surface of the
boundary increase in size {i.e. a smooth plate becomes covered with an encrusting growth of
nodulous algae, or flat slate bedrock becomes covered with gravel), the rate of transfer of momentum
in the fluid , and thus shear velocity, is altered.
Shear velocity U* is defined as the square root of shear stress (-cw) divided as the density of
the fluid (p ),
(2)
From th is relationship, two aspects of the relationship between a moving flu id and a solid boundary
are apparent. First, shear stress can be estimated from a velocity profile {by rewriting equation 2 as
equation 3, and recognizing that, with care and appropriate equipment, all parameters in equation 1
can be measured in either a field or lab setting).
(3)
Second, wall shear stress increases directly with increases in the density and exponentially with
increases in the shear velocity of the fluid. Density, however, is not likely to change much in a given
channel, even if we account for fluctuations in temperature or solute load.
Despite the simple relationship between shear stress, shear velocity, and density, there can
be substantial variation in the fluid dynamic parameters as fluid shears due to friction at the fluid/solid
interface. If vertical transfer of momentum in the fluid is limited, the slope of the near-bed portion of
the velocity profile (the plot of z on U) is relatively steep . Adjacent 'sheets' of water are moving at
nearly the same velocity and 'slide' over one another only very slowly; shear stress in this case is

relatively low. If the near-bed portion of the velocity profile is relatively flat, however, the velocities of
adjacent 'sheets' of water are very different, and are 'sliding' over one another very quickly; shear in
this instance will be relatively high .
As the slope of the velocity profile is indicative of shear stress, the law of the wall can be
applied to estimate
However, the nature of the law of the wall varies depending on the relative
influence of viscosity which, in turn , varies with distance from the bed . When examining flows
immediately adjacent to the bed , the dominance of viscous effects {boundary layer flows are a
Reynolds-scalable phenomena) must be accounted for. The law of the wall as expressed in equation
1 only applies to the portion of the water column that is strongly influenced by viscosity, generally the
bottom 20% of the flow depth (see Biron et al. 1998 for a discussion). In the outer portions of the
boundary layer (beyond the so-called 'overlap layer'), the log-normal nature of the profile does not
apply, and the law of the wall as it appears in equation 1 no longer applies. Biron et al. (1998)
demonstrated that shear stress is underestimated when the entire velocity profile, rather than the
bottom 20%, is used.
As long as the log-normal portion of the boundary layer is used, equation 1 can be rewritten
as

•w·

U*/K = dUx I dlnz

(4)
230

which can be expressed as
U*/K =[slope of plot of U on lnz] = m

(5)

and further simplified as
U* =m K

(6)

where m is the slope of the plot of U on lnz. It is important to note that when making determinations
of shear velocity U*, the slope of the profile must be that of the plot of U (response variable) on lnz
(predictor variable); plots of lnz on U are only to be used for diagrammatic, rather than analytical,
purposes. In addition, Bergeron and Abrahams (1992) note that because U is generally measured
with greater error than is lnz, regressing lnz on U can lead to substantial errors in estimates of U*.
The principle of estimating shear stress using velocity profile data can be best illustrated
using an example. Using a small-scale Pitot-static tube, a velocity profile was obtained over a
boulder in a third class torrential stream (Figure 1; stone 7, Oct. 7, 1998; see also chapter 3). The
1
mean velocity in this location was 1.1 m s· . The velocity profile, and especially the near-bed portion
(Figure 1), appears to be log-normal. By visually inspecting the semi-log plot (z loge transformed), it
appears that the first eight points (or, within 1.28 em of the bed) comprise the log-layer (Figure 2). A
linear regression of this portion of the profile (remembering to employ the proper plot of dU I dlnz)
produces the following result (Table 1).

Table 1. Linear regression of the log-layer portion of the velocity profile (Oct.
7, 1998)
U = 1.52 + 0.1151nz
Predictor
Coeff.
Constant
1.523
Lnz
0.115
s = 0.0190

Stand. Dev.
0.0383
0.00680

t-ratio
39.77
16.97

p
0.000
0.000

MS
0.104
0.00036

F
287.95

2

R = 98.0%

Analysis of Variance
Source
DF
1
Regression
Error
6
Total
7

ss

0.104
0.00217
0.106

p

0.000

The slope of the regression is 0.115, which, when applied in equation 6, gives a value of U* of 0.046
1
m s· . This value, when incorporated into equation 3, provides an estimate of shear stress of 2.12 Pa.
A steeper slope, implying higher rates of shear in the log-layer (as adjacent 'sheets' of water slide
over one another at a higher rate), would produce a higher value of U*, and result in a higher value of
'tw.

231

-

E
..._...

0.3

~ 0.25

ro

"0

c::

::J

0

-e
.0

E
Q)

0.2
0.15

u

-

c::
ro 0.05

-~

"0

...
....... '

•
•
I

0.1

0
0.4

0.6

0.8

1.2

••
•••
•
••

1.4

1

velocity U (m s- )
Figure 1. Velocity profile obtained over stone 7 on Oct. 7, 1998 using a Pitot-static tube.

Figure 2. Logarithmic plot of the velocity profile in Figure 1, showing the linearity of the first eight
points (indicating the portion of the profile which corresponds to the 'log-layer' or 'overlap layer').

References
Bergeron, N.E., and A.D. Abrahams. 1992. Estimating shear velocity and roughness length from
velocity profiles . Water Resour. Res. 28: 2155-2158.
Biron, P.M., S.N . Lane, A.G . Roy, K.F . Bradbook, and K.S . Richards. 1998. Sensitivity of bed shear
stress estimated from vertical velocity profiles: the problem of sampling resolution. Earth
Surf. Processes Landforms 23: 133-139.
Vogel, S. 1994. Life in Moving Fluids, 2nd Edition . Princeton University Press, Princeton, NJ.
White, F.M. 1999. Fluid Mechanics, 41h edition . WCB I McGraw-Hill, Boston
Wilcock, P.R. 1996. Estimating local bed shear stress from velocity observations. Water Resourc.
Res. 32: 3361-3366.

232

Appendix 3

Ontogenetic changes in Epeorus deceptivus during the study
period (Sept. - Oct. 1998)
Approximately every three weeks during the study period (September to October, 1998;
specific dates shown below), 5 Epeorus deceptivus larvae were collected at the study site in order to
determine the ontogenetic changes in morphology. During this period, head capsule width increased
by 19.4% (from 1.29 mm to 1.54 mm), while total body length (not including cerci) increased by
67.3% (from 3.15 mm to 5.27 mm) (Figure 1).

6
5

"'

4

X

3

Q)

2

E
b

..-

.......
.c

head capsule width
c:=J body length

C>

c

1
0

Sept.02

Sept.26

Oct.14

date

Figure 1. Change in body dimensions {head capsule width and total body length excluding cerci) of
Epeorus deceptivus during the study period (dates shown). In all cases n = 5 {bars= SO).

233

Appendix 4
Preston-static tube
In natural systems, accurate measurements of hydrodynamic variables such as velocity,
shear velocity, and shear stress are often difficult to obtain very close to the boundary. For example,
constant temperature anemometers using hot-wire or hot-film sensors are fragile and susceptible to
rapid fouling (Hart et al. 1996), particle image velocimetry (PIV) techniques cannot be easily utilized
where a clear side-view of the flow is not possible, and acoustic-based techniques such as acoustic
Doppler velocimeters (ADVs) suffer from resolution problems due to resonance when used to
measure water velocity within 30 mm of the bed (Finelli et al. 1999). However, small-scale Pitot-static
tubes and Preston-static tubes do not suffer from these types of problems, and can be used to
accurately measure near-bed water velocities and shear stresses.
In this study, measurements of near-bed velocity profiles and shear stress were made using
a small-scale Preston-static tube (PST). The Preston-static tube used was configured to directly
measure wall shear stress when deployed on a surface (Ackerman et al. 1994). The Preston-static
tube was developed for measuring wall shear in situations where the incorporation of a wall-mounted
static pressure tap is impractical (i.e. around small-scale objects where flow lines may be curved), but
has been recently tested for use in a variety of field and lab conditions (Ackerman and Hoover, 2001 ).
The Preston-static tube (Figure 1) was constructed of a 90° bevelled tip syringe needle fixed
to a side-bored syringe needle (tube diameter= 0.88 mm; see Ackerman et al. (1994) for a detailed
description). The Preston-static tube differs from a standard Preston tube in that the dynamic
pressure port (which faces into the flow) is coupled with a static pressure port (on the sidewall) . As a
result, the difference between static pressure Ps and dynamic (or stagnation) pressure p 0 can be
measured, and velocity or shear stress can be determined (see chapter 3 for a description of the
relationship between differential pressure, water velocity, and shear stress). A Validyne model
CD379-1-2 membrane-type differential pressure transducer (Validyne Engineering; Northridge, CA),
connected to the Preston-static tube via 2.0 m of flexible pressure lines (inside diameter = 3 mm),
was used to measure differential pressure. Pressure differences were expressed as a voltage by a
Validyne portable indicator (model CD379). In order to obtain a mean differential pressure, 15
voltages (obtained over an approximately 30 second time period) were averaged to obtain each
measurement. Before measurements were taken, the pressure lines connecting the Preston-static
tube to the pressure transducer were inspected for the presence of trapped air bubbles which would
cause error in the differential pressure measurements. Preston-static tubes must be aligned parallel
to the flow for accurate measurement. As such, before each measurement was made, the horizontal
tube of the PST was oriented parallel to a fluorescein dye filament released immediately upstream
from a long thin needle (length = 200 mm, diameter = 1 mm) (see chapter 3 for an additional
description of Preston-static tube orientation and deployment).

234

8

A

'r--=--<

t

Ps

-

ball and socket
connector

pressure lines

u--.

-clamp
-

Preston-static tube

5mm

Figure 1. A) The Preston-static tube measures wall shear stress using the difference between the
total pressure (p 0 ) and static pressure (p 5 ). The PTot tap and ps tap are indicated (tube diameter d =
0.88 mm; the diameter of the 20G syringe needles has been exaggerated for clarity). B) Deployment
device used to position the Preston-static tube. The ball-and-socket connector allows the PST to be
oriented parallel to the stream flow. The deployment device is mounted on a vertical support and
attached to a xyz-positioning device (not shown; see chapter 3 for a description) (see also Ackerman
and Hoover 2001 ).

References
Ackerman, J.D., and T.M. Hoover. 2001 . Measurement of local bed shear stress in streams using a
Preston-static tube. Limnol. Oceanogr. 46: 2080-2087.
Ackerman, J.D., L. Wong, C. R. Ethier, D.G. Allen, and J.K. Spelt. 1994. Preston-static tubes for the
measurement of wall shear stress. J . Fluids Eng . 116: 645-649.
Finelli, C.M., D.D. Hart, and D.M. Fonseca. 1999. Evaluating the spatial resolution of an acoustic
Doppler velocimeter and the consequences for measuring near-bed flows. Limnol. Oceanogr.
44 : 1793-1801 .
Hart, D.D., B.D. Clark, and A. Jasentuliyana. 1996. Fine-scale measurement of benthic flow
environments inhabited by stream invertebrates. Limnol. Oceanogr. 41: 297-308.

235

Appendix 5
Periphyton sampler
The measurement of the biomass or chlorophyll content of periphyton, a community that
includes microbes, protists (protozoans, and algae), and fungi within a 'polysaccharide matrix' (Lock
et al. 1984 ), is one of the most important determinations in studies of the trophic structure of stream
environments (Steinman and Lamberti 1996). Analyses of periphyton have been used to infer
variations in photosynthetically active radiation (Steinman 1992), ultraviolet radiation (Bothwell et al.
1994 ), herbivory (Lamberti et al. 1987), and water velocity (Steinman and Mcintyre 1986), among
others. As such, it is important that any periphyton sampling technique incorporates the following
criteria: (1) that a constant proportion (preferably all) of the biomass be removed; (2) that a known
area be sampled; (3) that it is possible to apply the technique under a variety of conditions; and (4)
that the results are repeatable. Unfortunately, most periphyton sampling techniques (e.g. scraping,
brushing, or direct extraction of the substrate) are lacking with respect to one, or more, of the
aforementioned criteria (Cattaneo and Roberge 1991 ).
A device was developed to obtain algal samples from small, repeatable areas on the surface
of periphyton-covered natural and artificial substrates. The device consists of a central cylinder (PVC
pipe; inside diameter = 20 mm) 90 mm in length, sealed at one end and encircled by a high-density,
closed-cell foam rubber skirt at the other (Figure 1). The rubber skirt extends approximately 5 mm
from the end of the PVC cylinder, and is used to seal the device against the surface of the substrate
from which the sample is being obtained . The cylinder contains a circular brush cut from the head of
an Ultra-Stiff vegetable brush (Tupperware; Orlando, Florida) with nylon bristles (bristle length = 15
mm). This circular brush is attached to a brass shaft (diameter= 3 mm, length = 130 mm), which
extends upward through the device, and emerges through a hole in the upper end of the cylindrical
housing (hole diameter = 3.5 mm). The length of shaft that extends above the device was fastened in
the chuck of a hand-held, rotary carving tool (Ryobi® HT20VSK; Pickens, S.Carolina). The side of
the cylindrical housing was fitted with two tubing ports separated at 90°. The first port, fixed slightly
higher than the other, provides an entrance for water, while the second, lower port is attached to a
vacuum line, allowing for the removal, via suction, of the algae-water slurry created when the within
the device.
When the device is pressed tightly against the surface to be sampled, the foam rubber skirt
forms a seal against the substrate. Water is then introduced into the device via a line attached to the
upper port (either a squirt bottle or distilled water tap can be used to control the flow of the water into
the device), and the vacuum pump turned on. When the rotary tool is activated, the brush rotates,
scrubbing algae from the substrate. The slurry produced gathers in a filter flask attached to a vacuum
pump. The material can then be collected on a glass-fibre filter for determination of chlorophyll a
content or organic/inorganic biomass (ash-free dry weight).

236

A

E
F

Figure 1. Diagram of the apparatus used to sample periphyton from the surface of naturally
occurring and experimentally deployed substrates. The device consists of a plastic (PVC) cylinder
(A), sealed at the upper end . During use, the bottom end of the cylinder is pressed to the substrate.
A seal between the substrate and the cylinder is made by a high-density, foam rubber skirt (B) at the
base of the cylinder. A cylindrical, nylon-bristled brush (C), is attached to a hand-held rotary carving
tool (Ryobi® HT20VSK) via a brass shaft (D) (diameter= 3 mm) which extends upward through the
device. Water is drawn into the device at (E). When the rotating brush is held against the substrate,
the water and removed algae form a slurry that is carried off by suction (at F). The brush is shown
extended from device for clarity.

References
Bothwell, M.L., D.M.J . Sherbot, and C.M . Pollock. 1994. Ecosystem response to solar ultraviolet-B
radiation: Influence of trophic level interactions. Science 256:97-100.
Cattneo, A., and G. Roberge. 1991 . Efficiency of a brush sampler to measure periphyton in streams
and lakes. Can . J. Fish. Aquat. Sci. 48: 1877-1881 .
Hoover, T.M., J.L. McConnachie, and J.D. Ackerman . In prep . Measuring periphyton biomass on
natural and artificial substrates of differing roughness.
Lamberti, G.A., L.R. Ashkenas, S.V. Gregory and A.D. Steinman . 1987. Effects of three herbivores
on periphyton communities in laboratory streams. J. N. Am. Benthol. Soc. 6: 92-104.
237

Lock, M.A., R.R. Wallace, J.W. Costerton , R.M. Ventullo and S.E. Charlton . 1984. River epilithon :
toward a structural-functional model. Oikos 42 :10-22.
Steinman, A.D. 1992. Does an increase in irradiance influence periphyton in a heavily-grazed
woodland stream? Oecologia 91 :163-170.
Steinman , A., and G. Lamberti. 1996. Biomass and pigments of benthic algae. Pages 295-313 in :
Stream Ecology: field and laboratory exercises. Eds: Hauer, R., Lamberti, G. Academ ic
Press.
Steinman , A.D. and C.D. Mcintire. 1986. Effects of current velocity and light energy on the structure
of periphyton asemblages in laboratory streams. J. Phycol. 22 : 352- 361 .

238

Appendix 6

Stone survey form
The distributions of mayfly larvae on the upper surface of all experimentally deployed stones
in the study were recorded using a form (see below) that featured a grid that reproduced the
Cartesian grid marked on the surface of the stones (see chapter 3 and 4). Distributions of larvae on
the surface of naturally occurring stones were recorded (using circles marked on the diagram of the
stone; see below) on the same forms, by first measuring the stones (in situ) , and then sketching the
shape and extent of the stones on the form . Then , the distributions of larvae were recorded in the
same manner as in the pre-marked stones.

I

INVERTEBRATE POSITIONING
DATASHEETC

(Q ; ( s-

Time

AM/PM
\I

\ i

I

;

~

{

b.oQ

I

--- -

~ 'V-

..

!

I

___l _ - ~ ..,;>·- - · ;;tO

' ···
f-f - - -t-·-···

:::Q..C
-

f

~

!

'

:::()

. ---··

-- -~

~

~

--·--···-··-····-----

.---+-··-- -l 1
\
+Y
\~
--t:Y
--1\

I

:

:

1--,.

! /

l

t

Depth (at datum)=
Temperature=

.

:

h

~

·. ,.J

i

!

_, r ,

+t

-·+·· ~
I

I

\I

f'\.

I

~ ~

~

o'25'

- -~~

~

13

~

' if

~

'

J Date

~

Substrate 10 no.

----:-

·t•

i

Party

~

/

/

1

46. . . MVh

Notes

goc

'ne Wol tttr ~
~

~

tk a

.,., ... r"" t t;

"J "t-

d•v -

il. ti(J -..JC

,.,qtr (l?.pVi".j
d.o""" o.,.jo 1J..r
tt) t.

Mean Velocity=
ADF (Velocity) Files

239

~