INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI
films the text directly from the original or copy submitted. Thus, some
thesis and dissertation copies are in typewriter face, while others may
be from any type of computer printer.
The quality of this reproduction is dependent upon the quality of the
copy subm itted. Broken or indistinct print, colored or poor quality
illustrations and photographs, print bleedthrough, substandard margins,
and improper alignment can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete
m anuscript and there are missing pages, these will be noted. Also, if
unauthorized copyright material had to be removed, a note will indicate
the deletion.
O versize m aterials (e.g., maps, drawings, charts) are reproduced by
sectioning the original, beginning at the upper left-hand comer and
continuing from left to right in equal sections with small overlaps. Each
original is also photographed in one exposure and is included in
reduced form at the back of the book.
Photographs included in the original manuscript have been reproduced
xerographically in this copy. Higher quality 6" x 9" black and white
photographic prints are available for any photographs or illustrations
appearing in this copy for an additional charge. Contact UMI directly
to order.

A Bell & Howell Information Company
300 North Zeeb Road. Ann Arbor. Ml 48106-1346 USA
313/761-4700 800/521-0600

EVALUATION OF THE IMPACTS OF A SIMULATED IRRIGATION WITHDRAWAL ON THE
HABITAT AND POPULATIONS OF BROOK TROUT AND BENTHIC MACROINVERTEBRATES IN
HUNT CREEK, MICHIGAN

By
Edward Allen Baker

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Fisheries and wildlife

1995

UMI Number: 9605830

UMI MicroEorm 9605830
Copyright 1995, by UMI Company. All rights reserved.
This microform edition is protected against unauthorized
copying under Title 17, United States Code.

UMI
300 North Zeeb Road
Ann Arbor, MI 48103

ABSTRACT
EVALUATION OF THE IMPACTS OF A SIMULATED IRRIGATION WITHDRAWAL ON THE
HABITAT AND POPULATIONS OF BROOK TROUT AND BENTHIC MACROINVERTEBRATES IN
HUNT CREEK, MICHIGAN USING PHABSIM
By

Edward Allen Baker

The state of Michigan has many valued stream fishery resources.
However, the integrity of the stream resources of the state are
threatened by the increased use of water for the irrigation of lawns,
golf courses, and agricultural crops.

I diverted approximately 50% of

the summer streamflow from a treatment section of Hunt Creek, Michigan
to simulate an irrigation withdrawal.

I evaluated the Physical Habitat

Simulation System (PHABSIM) to determine its effectiveness in modeling
the impacts of the withdrawal on the brook trout Salvelinus fontinalis
and benthic macroinvertebrate habitat in Hunt Creek.

I used diurnal

habitat suitability criteria (HSC) for brook trout to model the
withdrawal impacts on diurnal brook trout habitat.

I also developed

alternative HSC for brook trout from nocturnal habitat use data and from
bioenergetic models of foraging microhabitats.

I used these HSC to

model the impacts of the withdrawal on the brook trout habitat in Hunt
Creek and compared the predictions of the PHABSIM model to brook trout
population data.

I also modeled benthic macroinvertebrate habitat from

habitat use data collected in Hunt Creek and compared the habitat
modeling output with population density data to determine if the
predictions of the PHABSIM model matched observed population responses
to the withdrawal.

The PHABSIM model indicated that the withdrawal did

not reduce the amount or quality of brook trout or benthic
macroinvertebrate habitat in the treatment section of Hunt Creek.
Population abundance data for brook trout and benthic macroinvertebrates
also indicated that the withdrawal had no impact on the numbers of
either group in the treatment section.

The PHABSIM model did indicate

that a withdrawal of greater than 50% would result in reductions in
benthic macroinvertebrate habitat but brook trout habitat would not be
reduced until discharge was reduced approximately 80%.

These results

suggest the PHABSIM modeling system may be effective in predicting
impacts of altered streamflows on the stream resources of Michigan.
However, I recommend continued study in Hunt Creek to validate the
effectiveness of the PHABSIM model.

For my wife.

Her understanding, patience, and hard work made this
possible.

iv

ACKNOWLEDGEMENTS
Z thank my guidance committee members: Dr. Gary Mittelbach, Dr.
Rich Merritt, and Dr. Scott Winterstein.

I especially thank my advisor,

Dr. Tom Coon, for his guidance, support and friendship.
I also am grateful to the personnel of the Michigan Department of
Natural Resources' Hunt Creek Fisheries Research station: Gaylord
Alexander, Andy Nuhfer, Jack Rodgers, Tom Adams, and Tim Smigelski.
Their cooperation during this study was invaluable.

Many undergraduate

and graduate students also helped with various aspects of the field and
laboratory work for this study.

I thank all of them but especially K.

Meyer, D. Simpkins, and J. Anicka.
Financial support for this study came from the Michigan Department
of Natural Resources, the Michigan State University Agricultural
Experimental Station, the Michigan Polar-Equator Club, the West Michigan
Chapter of Trout Unlimited, and the Kalamazoo Valley Chapter of Trout
Unlimited.
Finally, I thank my friends and especially my family for their
endless support and encouragement.

v

TABLE OF CONTENTS
Page
List of Tables

vii

List of Figures

ix

Introduction

1

chapter Is DEVELOPMENT AND EVALUATION OF ALTERNATIVE HABITAT
SUITABILITY CRITERIA FOR BROOK TROUT SALVELINUS EONTINALIS
Abstract
Introduction
Methods
Study Area
Bioenergetic-HSC Construction
Use-HSC Construction
Results
Bioenergetic-HSC
Use-HSC
Comparison of Diurnal Foraging HSC
Discussion
Literature Cited

3
3
5
8
8
11
18
21
21
34
37
42
49

Chapter 2s COMPARISON OF PREDICTED HABITAT CHANGE AND BROOK TROUT
POPULATION RESPONSE TO A SIMULATED IRRIGATION WITHDRAWAL IN
HUNT CREEK, MICHIGAN
Abstract
Introduction
Methods
Results
Characteristics of Study Reaches
Habitat Suitability Criteria
PHABSIM Model Results
PHABSIM Predictions
Fish Movement
Discussion
Literature Cited

55
55
56
59
62
63
64
64
73
80
83
89

Chapter 3: COMPARISON OF PREDICTED HABITAT CHANGE AND BENTHIC
MACROINVERTEBRATE RESPONSE TO A SIMULATED IRRIGATION
WITHDRAWAL IN HUNT CREEK,MICHIGAN
Abstract
Introduction
Methods
Results
Benthic Macroinvertebrate Assemblage
PHABSIM Model Results
Invertebrate Drift
Discussion
Literature Cited

94
94
95
97
102
102
107
126
131
137

Summary

142

Appendix A

144

vi

LIST OF TABLES

Table 1.

Parameters for the linear regression equation Log
(C)=I+S*FV, where C-the cost of swimming (mg 02*kg
^'hr- 1 ) for brook trout developed from data in
Beamish (1980).

I=y-intercept, S=regression slope

and FV=focal velocity (cm's- 1 ).

Fish weights were

estimated from length-weight regressions developed
from data collected in Hunt Creek.
Table 2.

Codes used to classify substrate and cover use and
availability in Hunt Creek.

Table 3.

Optimal and suitable habitat suitability criteria
values for diurnal and nocturnal periods.
all HSC are from Chapter 1.

Data for

Bioenergetic-HSC were

size specific and are presented here only for 7.5
and 15 cm fish (sizes equivalent to young of the

Table 4.

year and yearling and older fish respectively).
3 —1
Estimated discharges (Q, m *s ) for reaches B2 and
2

B4 and corresponding WUA (m *100 m

—1

) estimates

that would produce a statistically detectable
reduction in brook trout density or biomass in
section B.
Table 5.

Size and number of brook trout caught in inclined
screen traps during the treatment period (1991-94)
and for the summer prior to withdrawal from traps
at the upstream and downstream bulkheads.

Table 6.

Benthic macroinvertebrate taxa collected and
percent frequency of occurrence in benthic samples
in sections B and C of Hunt Creek,

1992-94.

Taxa

selected for habitat modeling are in bold face
type.
Table 7.

103

Mean column velocity (cm's- 1 ) and depth (cm)
habitat suitability ranges for the 13 families of
benthic macroinvertebrates selected for habitat
modeling in Hunt Creek.

Column headings (1, 0.5,

0.2, 0.1, and usable) are suitability values.
Table 8.

106

Substrate suitability values for the 13 families of
benthic macroinvertebrates selected
for habitat modeling in section B of Hunt Creek.
Percent embeddedness of the substrate was not
included in the substrate suitability calculations.

Table 9.

109

Weighted Usable Area (WUA, m 2 ’100 m - 1 ) in relation
to discharge for the 13 benthic macroinvertebrate
families selected for habitat modeling in section B
of Hunt Creek.

Table 10.

114

Benthic macroinvertebrate density estimates
(number'm

_2

, standard error estimates in

parentheses)

from random sample locations in

sections B and C for 1992-93.
Table 11.

117

Benthic macroinvertebrate density estimates
(number'm

_2

, standard error estimates in

parentheses) from one riffle each in sections B and
C, 1994.
Table 12.

121

Mean difference between pretreatment and treatment
period benthic macroinvertebrate density estimates
between sections B and C and BACI analysis resullts
for 1994 data (df=4 for all tests).

viii

127

LIST OF FIGURES
Page
Figure 1.

Map of Hunt Creek study area.

The upstream

bulkhead is the boundary between sections C and B,
the downstream bulkhead is the boundary between
sections B and A, and Fish Lab Rd. is the boundary
between sections A and Z.
Figure 2.

10

Parameters used in estimating bioenergetic benefits
of brook trout foraging microhabitats (adapted from
Hughes and Dill 1990).

The foraging area (FA) is a

two dimensional semi-circular plane perpendicular
to the direction of the current with radius equal
to the maximum capture distance (MCD).
OFA, and UFA
Figure 3.

RD, PL,

are defined in the text.

15

Invertebrate drift availability (calories*hr” 1 ) in
relation to mean column velocity (cm’s 1 ) from
drift samples collected in Hunt Creek at locations
where brook trout were observed feeding in section
B and on the

Figure 4.

fixed transect insection B.

22

Mean invertebrate length (mm) in relation to mean
column velocity (cm’s 1 ) for drift samples
collected in Hunt Creek at locations where brook
trout were observed feeding in section B and on the
fixed transect in section B.

Figure 5.

24

Reaction distance (RD) in relation to current
velocity (cm's 1 ) for foraging brook trout in Hunt
Creek.

25

ix

Figure 6.

Maximum capture distance (MCD) in relation to
current velocity (cm's- 1 ) for foraging brook trout
in Hunt Creek.

Figure 7.

26

Microhabitat benefit estimates based on length (cm)
of brook trout and current velocity (cm's- 1 ) for
foraging microhabitats in Hunt Creek.

Benefit

estimates are based on a microhabitat depth greater
than or equal to the fish's MCD.
Figure 8.

27

Brook trout length specific swimming cost estimates
(calories"hr- 1 ) versus current velocity (cm's- 1 )
based on equations given in Beamish

Figure 9.

(1980).

29

Length specific net caloric benefit (calories'hr- 1 )
estimates for brook trout foraging microhabitats in
Hunt creek in relation to current velocity (cm'sX ).

Figure 10.

30

Brook trout length specific bioenergetically
derived velocity habitat suitability criteria for
foraging microhabitats in Hunt Creek.

Figure 11.

32

Sample depth suitability criteria for 15 cm brook
trout in relation to current velocity in Hunt
Creek.

Figure 12.

33

Mean column velocity (cm's- 1 ) frequency-of-use data
and use-HSC for foraging young of the year (A) and
yearling and older (B) brook trout in Hunt Creek.
Histogram represents use data and line represents
suitability.

Figure 13.

35

Depth (cm) frequency-of-use data and use-HSC for
foraging young of the year (A) and yearling and
older (B) brook trout in Hunt Creek.

Histogram

represents use data and line represents
suitability.

36

x

Figure 14.

Mean column velocity (cm*a 3 ) frequency-of-use data
and use-HSC for resting young of the year (A) and
yearling and older (B) brook trout in Hunt Creek.
HiBtogram represents use data and line represents
suitability.

Figure 15.

38

Depth (cm) frequency-of-use data and use-HSC for
resting young of the year (A) and yearling and
older (B) brook trout in Hunt Creek.

Histogram

represents use data and line represents
suitability.
Figure 16.

39

Mean column velocity use-HSC and bioenergetic-HSC
for foraging young of the year brook trout in Hunt
Creek.

Figure 17.

40

Mean column velocity use-HSC and bioenergetic-HSC
for foraging yearling and older brook trout in Hunt
Creek.

Figure 18.

41

Logical sequence of the PHABSIM model process.
Data are entered in the form of habitat suitability
criteria and hydraulic status of the stream.

The

model predicts weighted usable area over a range of
discharges.

Weighted usable area is assumed to be

linearly related to fish abundance or biomass.
Figure 19.

58

2
—1
Total area (m *100 m
) in relation to discharge
(m3 *s-1 ) for reaches B2 (solid diamonds) and B4
(open squares)

Figure 20.

in Hunt Creek.

66

2.
—1 ) estimates derived from
Diurnal WUA (m ‘100 m
diurnal use-HSC as a function of discharge (m3 *s- 1 )
for young of the year (solid diamonds) and yearling
and older brook trout (open squares)

in reach B2 of

Hunt Creek.
Figure 21.

68
2

-1

Diurnal WUA (m *100 m

of stream) estimates

derived from diurnal use-HSC as a function of

XI

3 -1
discharge (m *s ) for young of the year (solid
diamonds) and yearling and older brook trout (open
squares) in reach B4 of Hunt Creek.
Figure 22.

2

Diurnal WUA (m *100 m

—1

69

of stream) estimates

derived from bioenergetic-HSC as a function of
discharge (m3 *s_ 1 ) for brook trout in reach B2 of
Hunt Creek.
Figure 23.

71
2

Diurnal WUA (m *100 m

—1

of stream) estimates

derived from bioenergetic-HSC as a function of
3 -1
discharge (m *s ) for brook trout in reach B4 of
Hunt Creek.
Figure 24.

72
2

Nocturnal WUA (m *100

—1

m

) as a function of

discharge (m3 ’s- 1 ) estimates for young of the year
and yearling and older brook trout in reaches B2
and B4 of Hunt Creek.
Figure 25.

74

Fall young of the year brook trout population
density (fish*ha- 3 ) estimates for sections A, B, C,
and Z of Hunt Creek for 1981-1994.
period was from 1991-94.

The withdrawal

Error bars represent 95%

confidence limits of the mean.
Figure 26.

76

Fall yearling and older brook trout population
density (fish*ha- 1 ) estimates for sections A, B, C,
and Z of Hunt Creek for 1981-1994.
period was from 1991-94.

The withdrawal

Error bars represent 95%

confidence limits of the mean.
Figure 27.

77

Fall brook trout population density (fish'ha- 3 )
estimates for fish 10-12.5 cm total length in
sections A, B, C, and Z of Hunt Creek for 19811994.

Figure 28.

The withdrawal period was from 1991-94.

Depth habitat suitability criteria for
Heptageniidae and Tipulidae calculated from habitat

xii

79

use data collected in section B of Hunt Creek,
summer 1992-93.
Figure 29.

110

Mean column velocity habitat suitability criteria
for Heptageniidae and Tipulidae calculated from
habitat use data collected in section B of Hunt
Creek,

Figure 30.

summer,

1992-93.

111

Substrate suitability criteria for Heptageniidae
and Tipulidae calculated from habitat use data
collected in section B of Hunt Creek, summer, 199293.

Figure 31.

112

2

Weighted usable area (m *100 m

—1 ) at discharge

(m^*s- 1 ) for Heptageniidae and Tipulidae in section
B of Hunt Creek.
Figure 32.

Total

123

macroinvertebrate

. —2
density (number m ,

with

standard error bars) in sections B and C of Hunt
Creek,
Figure 33.

Total

summer,

1992.

124

macroinvertebrate density (number m

_2

, with

standard error bars) in sections B and C of Hunt
Creek,
Figure 34.

Total

summer,

1993.

125

macroinvertebrate density (number"m

standard error bars) in

_2

, with

sampled riffles in sections

B and C of Hunt Creek, summer,

1994.

June 1

samples were collected immediately prior to the
Btart of the withdrawal period.
Figure 35.

128

Macroinvertebrate drift density (number*m- ^, with
standard error bars) in section B immediately
before and after the initiation of the withdrawal
period in 1993 and in section C.

Withdrawal period

began at approximately 12:40 on June 2, after the
12:00 sampling was completed.

Section C drift

samples were collected on June 6 and 7.

xiii

129

Figure 36.

Macroinvertebrate drift density (number‘m - 3 , with
standard error bars) in section B
1993.

on August 1-3,

Section B samples were collected on August 1

and 2 and section C samples were collected on
AuguBt 2-3.
Figure 37.

130

Macroinvertebrate drift density (number*m- 3 , with
standard error bars) in section B immediately
before and after the initiation of the additional
withdrawal period in August, 1993.

Withdrawal

period began at approximately 8:30 on August 22.
Section C samples were collected on August 25 and
26.
Figure 38.

131

Juvenile and adult brown trout WUA in relation to
discharge in reaches B2 and B4, Hunt Creek derived
from HSC data presented in Raleigh (1986).

Figure 39.

144

Juvenile and adult rainbow trout WUA in relation to
discharge in reaches B2 and B4, Hunt Creek derived
from HSC data presented in Raleigh (1984).

xiv

145

INTRODUCTION

The state of Michigan has abundant water resources that provide a
myriad of recreational opportunities for the state's residents and many
non-residents.

The stream resources of Michigan are particularly

valuable and are widely recognized as some of the finest in North
America.

However, the streams of the state and the Midwest are

threatened by the increasing use of water for out-of-stream uses
including irrigation of agricultural crops, golf courses and lawns.

The

use of water for seasonal irrigation has the potential to adversely
impact streams if water is removed directly from a stream or is removed
from an aquifer that supplies water to a stream.
The need to protect streams from excessive water withdrawals was
first recognized in the western United States in the 1970's.

This

recognition resulted in the formulation of several standard setting
methods which provide recommended flows in streams that are designed to
protect instream resources

(fisheries, recreational uses, wildlife) from

excessive degradation resulting from withdrawals.

The most common of

these methods is the Instream Flow Incremental Methodology (IFIM).

The

IFIM was designed to provide an estimate of the impacts of water
withdrawals from streams on instream fish, wildlife, or recreational
habitat.

The habitat modeling component of IFIM is the PHysical HABitat

SIMulation system (PHABSIM).

PHABSIM is widely used in the western

1

United States to evaluate the impacts of altered streamflows but has not
been widely applied in the midwest.
The objectives of this research were to: develop alternative
habitat suitability criteria for brook trout that could be used in a
PHABSIM analysis of stream habitat; evaluate the impacts of a simulated
irrigation withdrawal on aspects of the ecology of the brook trout and
benthic macroinvertebrate populations in Hunt Creek; to evaluate the
impacts of the simulated withdrawal on the brook trout and benthic
macroinvertebrate habitat using PHABSIM; and to compare the changes in
brook trout and benthic macroinvertebrate abundance predicted by PHABSIM
with observed changes.

2

CHAPTER 1

DEVELOPMENT AND EVALUATION OF ALTERNATIVE HABITAT SUITABILITY CRITERIA
FOR BROOK TROUT SALVELINUS FONTINALIS

ABSTRACT
I developed diurnal habitat suitability criteria from net bioenergetic
benefit models (bioenergetic-HSC)

for drift feeding brook trout in Hunt

Creek, MI and compared these to criteria developed from frequency-of-use
data (use-HSC).

I also constructed nocturnal use-HSC from frequency-of-

use data collected in Hunt Creek.

Bioenergetic-HSC were more

restrictive in predictions of optimal velocity: a single velocity was
optimal and depended on fish size, as opposed to a range of optimal
velocities predicted from frequency-of-use data.

The optimal velocities

predicted for yearling and older fish from bioenergetic-HSC (range 33-46
cm's"1) were greater than the highest optimal velocity (27 cm's"1)
predicted by use-HSC.

Optimal velocities for young of the year fish

predicted from bioenergetic-HSC (23 and 28 cm's 1 for 5 and 7.5 cm fish
respectively) were within the range of optimal velocities
predicted from use-HSC.

(6-30 cm's"1)

The predicted range of usable velocities was

narrower for bioenergetic-HSC than for use-HSC, regardless of fish size.
Use-HSC suitability scores for an independent set of habitat use
observations in Hunt Creek were significantly higher than bioenergeticHSC for young of the year fish but not for yearling and older fish.
This may indicate that use-HSC are too general and do not represent the
actual suitability of foraging microhabitats in Hunt Creek.

Young of

the year and yearling and older brook trout selected microhabitats with
lower mean column velocities and shallower depths at night than when

3

foraging in daytime.

Also, yearling and older brook trout used

microhabitats with higher mean column velocities and greater depths than
young of the year fish during diurnal and nocturnal periods.

4

Introduct ion
The construction and use of habitat suitability criteria (HSC) is
an important step in the evaluation of stream fish habitat, particularly
in conjunction with the use of
(PHABSIM).

the Physical Habitat Simulation System

The HSC used in stream

habitat

modelingare quantitative

models that represent the suitability of particular habitat parameters
for stream fish.

The value of HSC ranges between zero for an unusable

habitat state to on for an optimal habitat state (Bovee 1986; Thomas and
Bovee 1993).

The four habitat parameters typically used in a PHABSIM

evaluation of

stream habitat are water depth, water velocity, substrate

and instream cover (Milhous et

a l .1989). Previous

evaluations have been based on

HSC for the

stream habitat

species and life stage of

interest and constructed by use of one of three methods as suggested by
Bovee (1986) : l)the construction of HSC from expert opinion,
2)collection of frequency-of-use

data in the stream under investigation

and subsequent conversion of frequency-of-use data to HSC by one of
several methods and 3)frequency-of-use data corrected to reflect habitat
availability in the stream of interest so that HSC reflect the
preference of the species for specific microhabitat attributes.

The HSC

generated from these three methods are termed Category I, II and III
models respectively (Bovee 1986).

Category II criteria are the most

widely used in investigations of stream habitat.
Classifying the suitability of microhabitats based on frequencyof -use data alone may not be accurate.

It is possible for less

frequently selected microhabitats to be more suitable than those most
frequently selected if competition for microhabitats is intense in a
particular stream.

For example, a stream reach with N microhabitat

5

units that are truly optimal and 2N microhabitat units that are half as
suitable as the optimal microhabitats would have enough usable habitat
units for 3N fish.

If the stream supports 3N fish the microhabitats

that are less than optimal would be used most frequently and would
therefore be classified as optimal based on frequency-of-use data.
Also, in this simple case, if the HSC were corrected for habitat
availability the optimal and suboptimal habitats would be equally
suitable.
Several factors influence habitat use by drift feeding salmonids
in streams including energetic gains (Fausch 1984; Hughes and Dill 1990;
Hill and Grossman 1993), predation risk, and cover availability (McNicol
et al. 1985; Grant and Noakes 1987; Huntingford et al. 1988), all of
which may evoke territorial behavior (Grant and Noakes 1988; Hughes and
Dill 1990; Hill and Grossman 1993).

Previous studies on stream fish

have stressed the importance of energetic gains associated with drift
feeding and have demonstrated that drift feeding fish select
microhabitats that optimize energetic gains during foraging (Fausch
1984; Hughes and Dill 1990; Hill and Grossman 1993).

Juvenile coho

salmon Onchorhyncus kisutch form dominance hierarchies and experience
growth rate constraints inversely related to position in the dominance
hierarchy and related to microhabitat selected (Nielsen 1992).

Optimal

foraging theory (Schoener 1971) also predicts that, among other factors,
position choice for a drift feeding fish should be influenced by
energetic costs and benefits associated with the microhabitat, and that
drift feeding fish should select microhabitats that maximize the net
energetic gains during foraging.

This suggests the suitability of a

6

microhabitat location for a drift feeding stream fish should be related
to the energetic costs and benefits associated with the location.
Bioenergetic costs and benefits associated with microhabitats for
drift feeding stream fishes may be a more appropriate measure of the
suitability of microhabitats than data on frequency of habitat use.
Further, if HSC derived from bioenergetic models more accurately
represent the actual suitability of microhabitats, they may provide more
accurate predictions of the impacts of altered stream flows on drift
feeding stream fishes.

The use of bioenergetic modeling for

microhabitat suitability could also be used in individual based models
to predict growth rates of fish in particular microhabitats and to
evaluate the spatial array of microhabitats in a stream (Rose and Cowan
1993; Brandt and Kirsch 1993; Goyke and Brandt 1993).
The objectives of this work were to develop HSC derived from
bioenergetic cost and benefit models for foraging brook trout Salvelinus
fontinalis (bioenergetic-HSC in remainder of text) and to compare these
to HSC based on frequency-of-use data (use-HSC in remainder of text).

I

describe a method for developing bioenergetic-HSC based on water
velocity and depth and then test these against use-HSC on an independent
data set from Hunt Creek.

The hypothesis was that bioenergetic-HSC

suitability scores calculated from an independent data set of depth and
velocity use data would be lower than suitability scores calculated from
use-HSC.
This hypothesis was based on three assumptions.

The first

assumption was that the brook trout population in Hunt Creek was at or
near carrying capacity and competition for foraging microhabitats was
intense.

Evidence from Hunt Creek supports this assumption because Hunt

7

Creek is closed to fishing, and there are few piscivorous predators in
the research area of Hunt Creek.

In addition, an artificial increase of

the sand bed load in Hunt Creek reduced benthic invertebrate abundance
and brook trout abundance, presumably by reducing the habitat available
to both and food availability for trout (Alexander and Hansen 1986).
The second assumption is that the brook trout in Hunt Creek select
foraging microhabitats based on the net energetic gain available.

The

third assumption follows the second: substrate and cover in the
immediate vicinity of a foraging fish's position do not influence the
net energetic gain available from a microhabitat.
I suggest that the output of a PHABSIM analysis using
bioenergetic-HSC may be more biologically meaningful in terms of
expected changes in fish population parameters

(i.e. predictions of

growth rate) when changes in stream flow are modeled.

Use of

bioenergetically derived HSC may also lead to a better relationship
between the output of a PHABSIM analysis (WUA) and fish population
parameters in a hydraulically altered stream if they more accurately
reflect the suitability of microhabitats than HSC constructed from
frequency-of-use data.

Also, because the criteria are based on

bioenergetics instead of habitat use and availability the bioenergeticHSC may be easier to transfer to other streams (Thomas and Bovee 1993)
with possible adjustments for food availability in the target stream.
Methods
Study Area
This study was conducted at the Michigan Department of Natural
Resources'

(MDNR) Hunt Creek Fisheries Research Station in northern

Oscoda and southern Montmorency counties of Michigan's lower peninsula.

8

Hunt Creek, is a third order stream which drains glacial sands and
gravels deposited during the last glaciation of the region,
approximately 10,000 years ago (Dorr and Eschman 1970).

Hunt Creek and

surrounding watersheds have extremely stable discharge and temperature
regimes and are some of the most productive trout streams in Michigan
(G. Alexander, personal communication).

Hunt Creek was chosen as the

study stream for this research because the brook trout population in
Hunt Creek is self sustaining and has been monitored by the MDNR since
1949: a continuous record of population density estimates exists from
spring and fall mark-recapture electrofishing.

In addition, the entire

Hunt Creek research area has been closed to fishing since 1966.
Therefore, any response of the brook trout population to experimental
treatment should be attributable to an increase in the natural mortality
rate, emigration rate, or some other factor related to the treatment.
The portion of Hunt Creek that flows through the research area is
divided into four sections: three nontreatment sections (sections A, C
and Z) and a treatment section (section B; Figure 1).

Hunt Creek is a

second order stream upstream of the confluence with Fuller Creek and is
a third order stream through the remainder of the study area. Section C,
the source of the independent data set of habitat use observations, is
immediately upstream of the treatment section B.

Distribution of mean

velocities in section C are not significantly different from section B
(Mann-Whitney U, p=0.36) but depths are significantly shallower in
section C (Mann-Whitney U, p<0.001).
The brook trout population in Hunt Creek is composed primarily of
small fish; approximately 96% of the fish in section B are less than
17.7 cm total length (Alexander and Hansen 1986).

The only common fish

species in Hunt Creek are brook trout, mottled sculpin Cottus bairdi and

9

Flow

Hunt Creek
Fuller Creek
Section C

Upstream
Bulkhead

B2
Diversion Channel
Section B

B4

Downstream Bulkhead

Section A

Fish Lab Rd.
Section Z

Figure 1.

Map of

Hunt Creek study area.

The upstream bulkhead is the

boundary between sections C and B, the downstream bulkhead
is the

boundary between sections B and A, and Fish Lab Rd.

is the

boundary between sections A and Z.

10

slimy sculpin Cottus cognatus (Alexander and Hansen 1986).

In 1989-90

the MDNR excavated a diversion channel around the treatment section and
installed bulkheads at the upstream and downstream ends of the treatment
section (Figure 1, referred to as the upstream and downstream bulkheads
respectively in remainder of text). The bulkheads allowed us to control
the flow of water through the treatment section of Hunt Creek.

The MDNR

also installed inclined screen traps on the bulkheads to monitor
downstream fish movement into and out of the treatment section.
traps were operated during each summer of the treatment.

The

In addition,

to provide a baseline estimate of fish movement in Hunt Creek, the traps
on the downstream bulkhead were monitored during the summer of 1990
before the experiment was initiated.

The traps only caught fish moving

downstream, and prevented upstream fish movement.
Bioeneraetic-HSC Construction
I followed the methods described by Hill and Grossman

(1993) to

model the bioenergetic costs and benefits associated with specific
microhabitats based on water velocity and depth.

The net energetic

benefit

(Ex ) of a microhabitat is a function of the water velocity and

depth.

Water velocity is expected to affect the costs and benefits of

the microhabitat and the fish's foraging area, and depth is expected to
affect the foraging area of the fish.

Ex is equal to the difference

between the benefits gained by holding the position (Bx ) and the costs
of maintaining the position (Cx ) :
Ex = Bx - Cx .

(1)

I derived seven net benefit models for brook trout between 5 and
20 cm total length at increments of 2.5 cm.

11

Data necessary for

estimating model parameters came from a variety of sources given in the
description that follows.
I based estimates of microhabitat benefit on invertebrate drift
density data collected in Hunt Creek during 1993 and 1994.

I collected

invertebrate drift on a fixed transect in Hunt Creek during summer, 1993
from dawn to dusk.

I sampled the invertebrate drift for 20 minutes

every four hours at three locations across the transect approximately
every 30 days.

Preceding each sample I measured depth to the nearest cm

and measured mean column velocity to the nearest cm's’1 at each net
location with either a Marsh-McBirney electronic current meter or a
pygmy-Gurley mechanical current meter.

I compared velocity measurements

between the two meters in Hunt Creek on several occasions by measuring
velocity at specific points in the stream with both meters.

I found no

consistent differences in measurements of velocity between the two
meters and velocity measurements were always in close agreement.

I

sampled invertebrate drift with a rectangular drift net (64 pm mesh,
15.5 x 75 cm and 80 cm deep).

I also collected invertebrate drift

samples at locations where brook trout were observed feeding in 1993 and
1994.

The duration of drift sampling was 10 minutes at fish locations

as opposed to 20 minutes for the fixed locations.
I separated the invertebrates from the rest of the material
collected in the nets by floating the samples in a saturated sugar
solution (Anderson 1959).

I then preserved invertebrates in 95% ethyl

alcohol until they were identified and measured in the lab.
I identified aquatic organisms in the drift to family by use of
the keys in Merritt and Cummins (1984) and identified the terrestrial
invertebrates to order (Barnes 1987) .

12

I measured invertebrate lengths

to the nearest 0.1 mm using an ocular micrometer.

I only included

invertebrates >2 mm total length in the calculation of benefit because
this appears to be the smallest size prey item taken by other drift
feeding salmonids (Bisson 1978; Tippets and Moyle 1978) .

I converted

invertebrate lengths to weights with length-weight equations given in
Rogers et al.

(1977) and Smock (1980) and converted weights to caloric

values according to Cummins and Wuycheck (1971) .
I converted calories per sample to calories per hour and, because
the drift net sampled the entire water column, corrected the sample
caloric values to a standard depth of 25 cm (constant sampled area of
387.5 cm 2 ).

I made this correction by dividing 387.5 cm 2 by the area

sampled by the drift net and multiplied this by the calories per hour
for each sample.

I related calories per hour to mean column velocity by

use of linear regression in which the regression was forced through the
origin.

This yielded estimates of caloric benefit of microhabitats

2
based on current velocity for a constant sampled area of 387.5 cm .
Because fish size and water velocity are important in determining
the foraging success of a fish (Hughes and Dill 1990; Hill and Grossman
1993) , I adjusted the caloric benefit of a microhabitat by the maximum
capture distance (MCD) as defined in Hughes and Dill

(1990).

Maximum

Capture Distance (cm) is a function of fish size, water velocity, and
the size of the invertebrate prey:
MCD = VRD2- (V * RD / VMAX)2

(2 )

where:
RD=12*PL(1-e (-0 .2 * FL)

(3)

VMAX=17* FL 0 .58

(4) .

13

RD is the fish's reaction distance (cm), PL is p r e y length (mm), FL is
the fish's fork length (cm, Hughes and Dill 1990) ,

VMAX is the fish's

maximum sustainable swimming speed (cm's 1, Jones et a l . 1974) and V is
the microhabitat's mean column velocity (cm's 1) (Figure 2).
I estimated PL as a function of mean column velocity from the
invertebrate drift data collected in Hunt Creek by use of linear
regression.

I derived the regression equation by calculating the mean

invertebrate length for each drift sample and regressing these mean
invertebrate lengths against mean column velocity for the drift samples.
2

I used the MCD to estimate the fish's foraging area (FA, cm ) as a
semicircle, perpendicular to the current (Hughes and Dill 1990), with
radius equal to the MCD:
FA= 0 .5 (77MCD2 )

(5) .

I used a semicircle because over 95% of the fish observed foraging in
Hunt Creek were maintaining positions just above the substrate and
therefore could only feed on drift in an area defined as a semicircle
above the fish with radius equal to the MCD.
I adjusted the estimated caloric benefit of a microhabitat to
reflect the fish's FA as determined by the MCD.

I made this adjustment

by multiplying the benefit of the microhabitat (determined by the mean
column velocity) by the quotient FA/387.5 cm
drift samples).
the 387.5 cm
accordingly.

2

2

(standardized area for the

Therefore, if the fish's foraging area was greater than

the benefit estimate of the microhabitat was increased
The MCD increases with increasing velocity from 0 cm's 1

to a maximum, the value of which depends on fish size, and then
decreases to zero, again dependent on fish size (Hughes and Dill 1990).

14

R D = f(P L , fish len gth )
Prey
P L = f ( v e lo c it y )
M C D = f ( v e lo c it y , R D )
C u rre n t

F A = 0.5re M C D 2

_ ........... O F A - U F A
D e p th Suitability =— — ^ —

F A=0.5rc M C D 2

C urrent

Figure 2.

Parameters used in estimating bioenergetic benefits of brook
trout foraging microhabitats (adapted from Hughes and Dill
1990).

The foraging area (FA) is a two dimensional semi­

circular plane perpendicular to the direction of the current
with radius equal to the maximum capture distance (MCD).
RD, PL, OFA, and UFA are defined in the text.

15

I did not include the benthic prey available to the fish because
data indicated that brook trout in Hunt Creek foraged on the benthos
only rarely.

Behavioral observations collected in section B of Hunt

Creek demonstrated that only 10% of the feeding attempts were directed
at the substrate and none of the fish observed fed exclusively on
benthos (E. A. Baker unpublished data).
McNicol et al.

This is similar to results in

(1985), which showed that young of year brook trout in a

small stream in Manitoba, Canada directed only 3% of their foraging
effort toward the benthos.
I developed regression equations to estimate the cost of
maintaining position at a microhabitat location (Cx. cost of swimming)
from data in Beamish (1980, Table 1).

Because I only modeled summer

microhabitats I used the equations derived for brook trout swimming at
15° C.

This temperature is similar to the average daily maximum

temperature in Hunt Creek.

In the summers of 1993-94, average daily

maximum temperature for the period June l to August 31 was 15.l and 15.8
respectively.

The equations presented by Beamish (1980) related

swimming cost to current velocity and weight for brook trout at
velocities of 25, 30, 35, 40 and 45 cm's 1 .

I determined size specific

(i.e. 5, 7.5, 10 cm etc.) swimming cost estimates at each of these five
velocities for brook trout and calculated the linear regression of
swimming cost versus current velocity for fish of a specific size (Table
1).

I derived weight estimates for the brook trout in Hunt Creek from

length-weight data collected in the spring and fall, 1993 and 1994 in
Hunt Creek.

I estimated swimming cost in mg 02'kg’1 'hr’1 at velocities

from 0 to 100 cm's 1 from the equations in Table 1 and converted these
cost estimates to calories'hr'1 by use of the energetic equivalents
given in Elliot and Davison (1975).

I used a nonlinear regression

16

Table 1.

Parameters for the linear regression equation Log
(C)=I+S*FV, where C=the cost of swimming (mg

0 2

'kg 1 'hr 1)

for brook trout developed from data in Beamish (1980) .

I=y-

intercept, S=regression slope and FV=focal velocity (cm's'
1).

Fish weights were estimated from length-weight

regressions developed from data collected in Hunt Creek.

Length (cm)

Weight

(g)

I

S

r

2

P

5

1.0

2.605

0 .020

0.49

0.19

7 .5

3 .6

2 .366

0.019

0 .57

0 .14

10

8 .8

2.196

0 .019

0 .64

0.10

12 .5

17 .4

2 .064

0.019

0.70

0 .08

15

30 .5

1 .956

0 .019

0.75

0 .06

17 .5

48.9

1.865

0 .018

0.79

0 .05

20

73 .8

1 .786

0 .018

0 .82

0 .04

17

equation to predict focal point velocity as a function of mean column
velocity developed from data collected in Hunt Creek, and used the
estimates of focal point velocity to calculate Cx for specific
microhabitats by use of the equations in Table 1.
I modeled the net benefits of microhabitats withequation

1.

I

constructed the bioenergetic-HSC from the 7 net benefit curves by
standardizing each of the curves as outlined in Bovee (1986) .
Because the suitability of a particular microhabitat location
chosen by a drift feeding brook trout is dependent on the fish's
foraging area, I developed bioenergetic-HSC for depth which were
dependent on MCD and therefore, dependent on velocity.

Thesuitability

of the depth at a microhabitat is defined as follows:

if depth > MCD, HSCd=1.0
OFA - UFA
if depth < MCD, HSCd=-------OFA

(6)
,

(7)

where HSCd=depth suitability, OFA=foraging area available at optimal
depth (depth>MCD), and UFA=unavailable foraging area at depth<MCD.

This

assumes that depth is not limiting if it is greater than the MCD.
Because the MCD is dependent on the current velocity it is possible for
the suitability of a particular depth to vary as velocity changes.
Depth is defined as unsuitable when it is less than or equal to the
maximum body depth for brook trout (22.3% of total length, estimated
from data in Balon 1980) .
Use-HSC Construction
I collected diurnal and nocturnal brook trout habitat use data
during May through August, 1991-93 in Hunt Creek by instream observation

18

with mask and snorkel.

I randomly selected a 50 m reach for sampling

and moved from the downstream end through the entire reach to the
upstream end.

No 50 m reach was sampled twice during the diurnal or

nocturnal period in any year unless all reaches had been sampled
previously.

For each undisturbed fish observed, I estimated fish size

to the nearest 1.25 cm (0.5 in), classified substrate and cover at the
fish's location according to the codes in Table 2, measured depth to the
nearest cm, and measured mean and focal point velocity to the nearest
cm's 1 .
resting.

I also noted the fish's activity as either actively foraging or
A fish was considered resting if it was immobile and on the

substrate, and actively foraging if it was above the substrate and
swimming to maintain a position in the stream.

I only included data

collected from actively foraging fish for constructing diurnal use-HSC
and only included data from fish that were resting for constructing
nocturnal use-HSC.

Because Hunt Creek is a narrow stream in the study

section I was able to observe the entire stream width and thus sampled
all available habitats for habitat use data.
I constructed depth and mean column velocity use-HSC for the
diurnal and nocturnal periods for young of the year (<8.9 cm) and
yearling and older (.>8.9 cm) brook trout based on habitat use in Hunt
Creek using the nonparametric tolerance limits method (Bovee 1986).
Use-HSC were constructed using the formula:
NSI=2(1-P),

(8)

where NSI is the normalized suitability index and P is the central
proportion of the data distribution under the curve (Bovee 1986).

I

constructed use-HSC for depth and mean column velocity by this approach

19

Table 2.

Codes used to classify substrate and cover use and
availability in Hunt Creek.

Cover Code

Cover Description

l

No cover

2

Velocity shelter protruding out of substrate but not
providing visual isolation

3

Combination cover providing both visual isolation and
velocity shelter

Substrate Code

*

Substrate Description

1

Fines composed of sand and silt

2

Sand

3 .X

Small gravel, Diameter < 0.6 cm

4 .X

Medium gravel, diameter > 0.6 cm and less than 2.5 cm

5 .X

Large gravel, diameter > 2.5 cm

★
Substrate classifications for gravels included an estimate of the
embeddedness of the gravel, X=l,2,3, and 4 where l=up to 25%, 2=26-50%,
3=51-75%, and 4=76-100% embedded.

For example, a substrate

classification of 4.2 denotes medium gravel embedded between 26-50%.

20

and defined P as the 50, 75, 90, and 95% portions of the distribution
according to Somerville (1958) and a confidence level of 95%.
I compared microhabitat suitability values from the diurnal useand bioenergetic-HSC models for independent observations of microhabitat
use by brook trout in section C of Hunt Creek.

I collected the habitat

use data in section C of Hunt Creek during summer, 1991-92 with the same
methods employed in section B and calculated microhabitat suitability
scores by multiplying depth and velocity suitability values, as in a
PHABSIM analysis (Milhous et al. 1989).

I tested the null hypothesis

that median suitability scores for these observations would be equal
with these two methods with the Wilcoxon sign rank test for paired
samples (Zar 1984) .
ResuIts
Bioenergetic-HSC
I collected invertebrate drift samples over a range of mean column
_ i
velocities from 2 to 82 cm's
. Invertebrate drift availability
(corrected to a depth of 25 cm, calories ’hr" ■*■) ranged from 2.1 to
1428.2.

The regression of calories'hr"1 on mean column velocity was

significant

(F=41.6, df=l,l48, p<0.00l; Figure 3).

However, because I

forced the regression through the origin I was unable to calculate a
meaningful coefficient of determination.

These results are similar to

those of Hill and Grossman (1993) who also found a linear relationship
between energetic content of the drift and velocity.

Aquatic

invertebrates captured in the drift samples represented 28 families.
However, chironomid larvae and pupae represented 52.6% of the
invertebrates captured in the drift.

21

calories hr 1=5.234*m ean column velocity (cm s 1)
1600
1400 -j
!

1200

|
I

Calories hr

r

1000

I
800
600 j
400

200
fl

0
10

20

|° o ° ° o

30

8

—i
40

50

60

70

80

90

Mean Column Velocity (cm s*1)

Figure 3.

Invertebrate drift availability (calories'hr 1) in relation
to mean column velocity (cm's’1) from drift samples
collected in Hunt Creek at locations where brook trout were
observed feeding in section B and on the fixed transect in
section B.

22

The relationship between mean invertebrate length (mm) and mean
velocity was linear and positive over the range of velocities sampled
2

(Figure 4; F=26.4, df=l,l44, p<0.01, r =0.16).
equation

I used this regression

to predict prey length (PL) in the calculation of fish reactive

distance(RD).

Based on

the prey length and fish length the RD

increased linearly with water velocity (Figure 5).
fish length
velocity

The ratio of RD to

was greatest for the smallest fish, RD for a 5 cm fish at a

of 0 cm's"1 was 20.6 cm, 4.1 times the fish length.

In

contrast, RD for a 20 cm fish at a velocity of 0 cm's 1 was only 1.6
times the fish length or 31.9 cm.
MCD was equal to RD at 0 cm's"1, increased slightly for fish of
all sizes with increasing velocity to a maximum, and then decreased at
higher velocities to zero (Figure 6).

Maximum values of MCD occurred at

velocities between 5 cm's'1 for 5 cm fish and 25 cm's’1 for 20 cm fish.
In the model, benefit of a microhabitat location was related to
caloric value of the drift via a regression forced through the origin.
Thus, the model predicted Bx of all foraging microhabitats at 0 cm's’1
to be zero for fish of all sizes.

Bx of foraging microhabitats

increased with increasing velocity and reached maxima at mean column
velocities between 23 and 48 cm's 1, depending on fish size (Figure 7).
Small fish had a narrower range of mean column velocities that provided
a net caloric benefit than larger fish, and optimal velocities increased
as fish size increased.

The maximum velocity with benefit greater than

zero for a fish of any particular size was equal to the maximum velocity
at which the fish's MCD was greater than zero (Figure 6).

23

m ean invertebrate length (mm)=2.71 + 1,12*mean column velocity (cm s-1)

Mean Invertebrate Length (mm)

4 5 -i-

5 H

oo
3

Oo

0

10

20

30

40

50

60

70

80

90

Mean Column Velocity (cm s'1)

Figure 4.

Mean invertebrate length (mm) in relation to mean column
velocity (cm's 1) for drift samples collected in Hunt Creek
at locations where brook trout were observed feeding in
section B and on the fixed transect in section B.

24

45

Reaction Distance (cm)

40

35

30

25

Fish Length (cm)

--10

20

12.5
17.5

-20
15 t
0

•f
20

40

60

80

100

Mean Column Velocity (cm s'1)

Figure 5.

Reaction distance (RD) in relation to current velocity
(cm's-1) for foraging brook trout in Hunt Creek.

25

35 i

Fish Length (cm)

Maximum Capture Distance (cm)

30
12.5
17.5
25

20

0

10

20

30

40

50

60

70

80

Mean Column Velocity (cm s'1)

Figure 6.

Maximum capture distance <MCD) in relation to current
velocity (cm's"1) for foraging brook trout in Hunt Creek.

26

900 -

Benefit of Microhabitat (Calories hr'1)

Fish Length (cm)
800 1
- -

12.5

10

17.5

700 4

20
600 f
500
400
300
200

-

100 -J

0

10

20

30

40

50

60

70

80

Mean Column Velocity (cm s'1)

Figure 7.

Microhabitat benefit estimates based on length (cm) of brook
trout and current velocity (cm’s 1) for foraging
microhabitats in Hunt Creek.

Benefit estimates are based on

a microhabitat depth greater than or equal to the fish's
MCD.

27

A nonlinear regression equation related focal point velocity (FPV,
cm's 1) to mean column velocity (MCV, cm's 1), based on 287 observations
of foraging brook trout:
FPV=2.973*MCV°'5 , r2=0.782.

(9)

I used this equation to predict swimming cost for foraging microhabitats
from the mean column velocity and used the swimming cost estimate in the
calculation of the net benefit curves.
Brook trout swimming cost estimates for velocities from 0 to 100
cm's"1 demonstrated that the cost of swimming was small for fish of all
sizes at slow velocities

(Figure 8).

The cost of swimming increased as

velocity increased and exceeded the benefits gained at a velocity of 40
c m ’s 1 for 5 cm fish and at 71 c m ’s 1 for a 20 cm fish.

The cost of

swimming was greater than zero at zero velocity, because the equations
in Beamish (1980) included the standard metabolic rate.
The net benefit estimates for foraging brook trout
9)

(Figure

demonstrated that the cost of swimming exceeded the benefit of a

microhabitat at low, as well as high, mean column velocities for fish of
all sizes.

The influence of fish size on the net benefit curves was

dramatic: the minimum velocity at which a 5 cm fish received a net
benefit was slightly higher than 0 cm's"1, but the largest fish (20 cm)
required a microhabitat with a mean column velocity of at least 6 c m ’s"1
to receive a net benefit.

These curves indicated that as brook trout

grew they required microhabitats with greater velocities to maximize
their energetic gains during foraging.

Net energetic gains were

greatest for the largest fish: a 20 cm fish foraging in an optimal
microhabitat could potentially receive a benefit of approximately 650
calories'hr 1, but a 5 cm fish foraging at an optimal location could

28

1000 -

Swimming Cost (Calories hr'1)

900
800 -

Fish Length (cm)

700

--10

600

12.5
17.5

--20

500
400
300 -

200
100

0

20

40

60

80

100

Mean Column Velocity (cm s'1)

Figure 8.

Brook trout length specific swimming cost estimates
{calories'hr-1) in relation to current velocity (cm's"1)
based on equations given in Beamish (1980) .

29

700
Net Benefit of Microhabitat (Calories hr'1)

Fish Length (cm)
600
12.5

--10

17 5

500 4

•-■20
400
300 -

200

-

100

-

/:r

20

30

40

50

60

80

-100 ^

Mean Column Velocity (cm s'1)

Figure 9.

Length specific net caloric benefit

(calories'hr 1)

estimates for brook trout foraging microhabitats in Hunt
Creek in relation to current velocity (cm's"1).

30

receive a maximum benefit of 150 calories'hr'1 (Figure 9).

The maximum

velocity which could provide a net benefit also varied with fish size
and increased with increasing fish size.

The maximum suitable velocity

(net benefit>0) for a 20 cm fish was approximately 71 cm's 1 while the
maximum suitable velocity for a 5 cm fish was only 38 cm's’1 .
The bioenergetic-HSC for velocity generated from the net benefit
curves (Figure 10) again demonstrated the importance of fish size and
mean column velocity on habitat suitability.

As brook trout size

increased the mean column velocity that was most suitable
(suitability=l.0) for foraging increased as well.

The most suitable

microhabitat for 5 cm brook trout was at a velocity of 25 cm's 1 and the
most suitable microhabitat location for a 20 cm brook trout was at a
velocity of 46 cm's 1 .

The bioenergetic-HSC for velocity assumed the

depth at the microhabitat location was at least equal to the MCD for the
fish.

•
The bioenergetic-HSC for depth (Figure 11) at a particular

microhabitat location in a stream depended on current velocity for a
fish of a particular size.

As velocity increased from 0 cm's"1 the

suitability of a specific depth increased to a maximum until the depth
equaled the MCD.

Also, the suitability of depth at a particular

velocity was dependent on fish size.

Foi example, the suitability at a

depth of 20 cm and a velocity of 10 cm's 1 was 0.94 for a 5 cm fish, but
it was only 0.86 for a 15 cm fish.

The dependence of depth suitability

on the current velocity is contrary to the current method of calculating
habitat area in a PHABSIM analysis in which the habitat parameters are
considered to be independent in their influence on fish habitat
selection.

31

1.0

Fish Length (cm)
0.9 -!

5

------ 7 5
12.5

10

0.8

1 5 --------17.5

Suitability

0.7

20

0.6
0.5 0.4
0.3 -i
0.2

-

0.1

-

0.0

0

10

20

30

40

50

60

70

80

Mean Column Velocity (cm s'1)

Figure 10.

Brook trout length specific bioenergetically derived
velocity habitat suitability criteria for foraging
microhabitats in Hunt Creek.

32

1

0.9

0.8

Suitability

0.7

Mean Column
Velocity (cm s'1)

0.6
0.5

• • - 40

20
•-•6 0

25

30

0.4
0.3 ;

0.2
0.1

0

—i

+

10

15

20

35

Depth (cm)

Figure 11.

Sample depth suitability criteria for 15 cm brook trout in
relation to current velocity in Hunt Creek.

33

Use-HSC
I constructed diurnal use-HSC for mean column velocity and depth
from observations of 149 young of the year and 138 yearling and older
foraging brook trout in section B of Hunt Creek (Figures 12 and 13).
The optimal velocities

(suitabi1ity=1.0) predicted for young of the year

brook trout in Hunt Creek were

from 6 to 30 cm's 1 , almost identical to

the

optimal range for yearling and older brook trout (6 to 27 cm's’1).

The

range of usable velocities (suitability>0.0) predicted from use-HSC

for

young of the year fish was from 0 to 66 cm's"1 (Figure 13).

A

velocity of 0 cm's'1 had a predicted suitability value of 0.5 for
yearling and older fish and the maximum usable velocity predicted for
yearling and older fish was 98 cm's’1 .
The range of optimal depths based on diurnal use-HSC was narrower
ans shallower for young of the year fish (15-34 cm) than for yearling
and older fish (27-55 cm; Figure 13).

Usable depths for young of the

year fish more closely overlapped the range of useable depths for
yearling and older fish (3-67 cm and 12-85 cm, respectively).

The

minimum depths used by both young of the year and yearling and older
fish were close to the minimum suitable depths predicted using the
maximum body depth from Balon (1980), although no young of the year

or

yearling and older fish were observed in water equal to the minimum
depth predicted from the body depth.

It should be noted however, that

it was very difficult for the snorkeler to see in water less than about
8 cm deep.
I constructed nocturnal use-HSC from observations of 31 young of
the year and 62 yearling and older brook trout in sections C and B
combined.

The optimal velocity range was 5-23 cm's 1 for young of the

34

20

09

-

18

0
e
0
31
t
J
0>
w

(/)

0

9

18

27

37

45

55

64

73

82

90

09

21

-

07

>
uN
o
i
3
O’
0
1
uw.

06

e

05
04
03

02

0

9

18

27

37

45

55

64

73

82

90

Mean Column Velocity (cm's*1)

Figure 12.

Mean column velocity (cm's 1) frequency-of-use data and useHSC for foraging young of the year (A) and yearling and
older (B) brook trout in Hunt Creek.

Histogram represents

use data and line represents suitability.

35

09

Frequency

14

0 7
-*■

10

06
0 5

2a

03

02

01
0

9

18

27

37

45

55

64

73

82

09
08
10
07

Frequency

06
05

:£
15

04

cn

03

0

9

18

27

37

45

55

64

73

-

02

-

01

82

Depth (cm)

Figure 13.

Depth (cm) frequency-of-use data and use-HSC for foraging
young of the year (A) and yearling and older (B) brook trout
in Hunt Creek.

Histogram represents use data and line

represents suitability.

36

year brook trout, and was 4-22 cm's 1 for yearling and older brook trout
(Figure 14).

The range of usable velocities was narrower for young of

the year fish (0-39 cm's 1) than for yearling and older fish (0-54 cm's
1).

The range of optimal depths was from 12 to 29 cm for young of the

year and was from 20 to 46 cm for yearling and older brook trout (Figure
15).

Usable depths were from 2 to 75 cm for young of the year and were

from 6 to 72 cm for yearling and older fish.
Brook trout use of depth and mean column velocity micro-habitat
attributes differed between young of the year and yearling and older
fish (MANOVA, F=31.6, df=2,375, p<0.001).

Depth and velocity use also

differed significantly between the nocturnal and diurnal period (MANOVA,
F=6.9, df=2,375, p =0.001), but the interaction between life stage and
period was not a significant source of variation in the model
F=0 .11, df=2,375, p=0.90) .

(MANOVA,

Young of the year fish occupied

microhabitats that were shallower and had slower mean column velocity
than those occupied by yearling and older fish during both nocturnal and
diurnal periods.

Also, both young of the year and yearling and older

fish moved to microhabitat locations that had lower mean column velocity
during the nocturnal period but depth use was not different between
periods within life stage.
Comparison of Diurnal Foraging HSC
The optimal velocities predicted from the bioenergetic models were
in general greater and narrower than those predicted from frequency-ofuse data (Figures 16 and 17).

The optimal velocities predicted for 5

and 7.5 cm fish (equivalent to young of the year size range) from
bioenergetic-HSC were within the optimal velocity range predicted from
use-HSC.

However, the optimal velocities predicted from bioenergetic-

37

Frequency

05
04

0
0

6

12

18

24

30

36

42

40

54

10

09

Frequency

06
05
04

02

0

6

12

18

24

30

36

42

46

54

Mean Column Velocity (cm s'1)

Figure 14.

Mean column velocity (cm's 1) frequency-of-use data and useHSC for resting young of the year (A) and yearling and older
(B) brook trout in Hunt Creek.

Histogram represents use

data and line represents suitability.

38

C9

08

06

S’ 3

02

0

6

12

18

24

30

36

42

48

54

60

66

9 7

8

09
08

7 4

6
5
4
3
03
2

02

1

0

0

6

12

18

24

30

36

42

48

54

60

66

72

Depth (cm)

Figure 15.

Depth (cm) frequency-of-use data and use-HSC for resting
young of the year (A) and yearling and older (B) brook trout
in Hunt Creek.

Histogram represents use data and line

represents suitability.

39

0.9 -

0.8
0.7

Suitability

Fish Length or Aae
bioenergetic-HSC for 5 cm
brook trout
- - bioenergetic-HSC for 7.5 cm
brook trout
— use-HSC for young of the year
brook trout

0.4
0.3

i-

0.0

0

10

20

30

40

50

60

70

Mean Column Velocity (cm s )

Figure 16.

Mean column velocity use-HSC and bioenergetic-HSC for
foraging young of the year brook trout in Hunt Creek.

40

0.9

Suitability

Fish Length or Age
-bioenergetic-HSC for 12 5 cm
brook trout
- bioenergetic-HSC for 15 cm
brook trout
use-HSC for yearling and older
brook trout

0.4

0.2

0.0

—

0

10

20

30

40

50

60

70

80

i —

90

-1 v

Mean Column Velocity (cm s )

Figure 17.

Mean column velocity use-HSC and bioenergetic-HSC for
foraging yearling and older brook trout in Hunt Creek.

41

HSC for fish 10 cm and larger (yearling and older) were all greater than
the optimal velocities predicted from use-HSC.

Comparisons of the depth

suitability values are difficult because the suitability of depth
depends on velocity for bioenergetic-HSC.

I tested the null hypothesis of no difference in suitability
scores between the two methods for 146 habitat use observations
collected in section C of Hunt Creek that were independent of the data
used to construct the HSC models.

The null hypothesis that microhabitat

suitability scores calculated from both models were equal was rejected
for young of the year fish (Wilcoxon signed rank test, Z=5.167,
p=0.<001) but not for yearling and older fish (Z=1.087, p=0.277).

The

suitability scores for young of the year fish were significantly higher
based on the use-HSC model

(median=0.875) than for the bioenergetic-HSC

model (median=0.498) in section C.

This is in spite of the fact that

velocity availability distributions were similar between sections C and
B (see description of study area).

Also, habitat use distributions were

similar for young of the year fish between sections C and B (n=136 in
section B, n=14l in section C, Mann-Whitney U=8562.5, p=0.124, df=l).

Discussion

The bioenergetic-HSC differed from the use-HSC in several ways.
Bioeneretic-HSC predicted narrower renges of optimal and usable mean
column velocity for both young of the year and yearling and older
foraging brook trout.

From the bioenergetic-HSC, a single velocity

provided a maximum energetic gain and thus, was optimally suitable for
foraging brook trout.

In contrast, the use-HSC predicted a range of

optimal velocities.

42

The predicted use suitability scores at 0 cm's 1 seem unrealistic
based on the energy maximization principle for drift feeding stream
fishes (Smith and Li 1983; Fausch 1984; Godin and Rangely 1989; Hill and
Grossman 1993).

This is because drift feeding fish depend on the

current for food delivery, and the benefit of the drift is positively
related to the current velocity (Hill and Grossman 1993) .

Therefore, a

drift feeding fish occupying a foraging microhabitat with zero velocity
should not receive a benefit from foraging, and the suitability of those
microhabitat locations should be comparatively low.

The fact that foraging brook trout in Hunt Creek occupied
microhabitats that were less than optimal based on bioenergetic-HSC may
be an indication that optimal foraging sites are limited.

This would

result in competition for foraging stations forcing some fish to occupy
suboptimal sites.

Others have noted that a linear dominance hierarchy

exists in foraging salmonids and that the dominant fish select
microhabitats that provide the greatest benefit in foraging (Fausch
1984; Hughes 1992; Nielsen 1992).

Brook trout competition for optimal

foraging sites in Hunt Creek could also explain the differences in
predicted optimal velocities from the two sets of HSC.

One alternative

explanation is that net energy gain from foraging is not what determines
brook trout foraging habitat use in Hunt Creek.

Other alternatives are

that the measuring instrument is not accurate at low velocity or the
fish were feeding on non-drift food items.

Alexander and Gowing (1976)

determined that oligochaetes were an important component of the diet of
two and three year old brook trout in Hunt Creek.

Although I did find

oligochaetes in the drift samples I collected, they were a very minor
component of the drift.

43

The bioenergetic-HSC predicted a narrower range of velocities that
provide usable foraging habitat for brook trout in comparison to the
range predicted by use-HSC.

One potential explanation for this

difference is that foraging brook trout in Hunt Creek may have selected
foraging stations that were shielded from the current but were adjacent
to a region of high velocity where foraging occurs.
been well documented for other foraging salmonids

This behavior has

(Everest and Chapman

1972; Fausch and White 1981) and was also documented in observations in
Hunt Creek.

However, most of the foraging brook trout in Hunt Creek

maintained foraging stations just above the substrate and pursued food
items that were in the overlying water column.

Fewer than 1% of the

foraging brook trout I observed in Hunt Creek were holding position in
calm water and feeding in faster adjacent water.
Another difference between bioenergetic-HSC and use-HSC was that
optimal velocity was different for brook trout of different lengths.
Use-HSC predicted an optimal velocity range for yearling and older
foraging brook trout that was nearly identical to the range of predicted
optimal velocities for young of the year brook trout.

In contrast,

optimal velocities predicted from bioenergetic-HSC increased with
increasing fish size and the optimal predicted velocities for yearling
and older fi.sh from bioenergetic-HSC were greater than those predicted
from use-HSC.

Again, this could be explained through competition for

the most suitable microhabitats.

Competition for foraging microhabitats

in Hunt Creek could force subordinate fish to choose microhabitats with
mean column velocities that are either greater or less than optimal
velocity.

The results of that choice should also provide the

subordinate fish with the maximum energetic gain available.

44

The use-HSC

suggest that brook trout select microhabitats with a velocity that is
less than the optimal velocity (predicted by bioenergetic-HSC)

in

greater proportion than they select microhabitats with a velocity
greater than optimal.

The net benefit curves also suggest that

microhabitats with a velocity less than optimal are more suitable than
those with higher than optimal velocity (Figure 10).

For example, a 5

cm brook trout faced with the choice of occupying a microhabitat with a
velocity 15% lower than optimal or 15% higher than optimal velocity
should select the microhabitat with the lower velocity because it
provides a greater net benefit.

The observation that optimal predicted

velocity from bioenergetic-HSC increased with fish size suggests that,
in the absence of competition, the mean column velocity at a foraging
station selected by a drift feeding brook trout should be correlated
with the length of the fish.
Differences in the depth HSC between the methods are difficult to
assess because the suitability of a particular depth based on fish
foraging area is dependent on velocity.

Use-HSC for depth agreed

reasonably well with bioenergetic-HSC in the predictions of minimum
suitable depth although bioenergetic-HSC predicted suitable depths that
were slightly lower than those predicted from use-HSC.
The fact that bioenergetic-HSC yielded lower suitability scores
than use-HSC for young of the year brook trout observational data
collected in section C but not for yearling and older observational data
could also be explained by competition between young of the year fish.
If competition between young of the year fish was intense in section C
it could result in density dependent mortality or emigration of young of
the year fish.

This could reduce the density of the remaining fish to a

45

level low enough that competition between yearling and older fish for
foraging microhabitats is not as intense and therefore, a higher
proportion of the yearling and older fish can use foraging microhabitats
that have a relatively higher suitability.

It also could mean that

there is a greater availability of foraging microhabitats with high
suitability values for yearling and older fish than for young of the
year fish.
The tendency for both young of the year and yearling and older
fish to select microhabitats with lower mean column velocities during
the nocturnal period than during the diurnal period is further evidence
that the fish selected microhabitats that maximized energetic benefit.
Although there was no energetic gain during the nocturnal period because
the fish were not foraging, the fish minimized energetic expenditure
during the resting period.

By minimizing energy expenditure for

swimming, the fish maximized the amount of energy gained during active
foraging that was available for growth of soma and reproductive organs.
The differences between bioenergetic-HSC and use-HSC have
important implications for stream habitat analysis using the PHABSIM
modeling system.
Creek,

A PHABSIM analysis of summer foraging habitat in Hunt

(Chapter 2) documented differences in both the shape and

magnitude of the weighted usable area (WUA, the measure of habitat area
and quality calculated in a PHABSIM analysis) curves that were
calculated from bioenergetic-HSC versus use-HSC.

WUA values at a

particular discharge were generally lower when bioenergetic-HSC were
used in the calculations.

In addition, the PHABSIM model predicted a

reduction in discharge of 98% in section B of Hunt Creek would reduce
WUA 37-70% based on use-HSC and 75-99% based on bioenergetic-HSC.

46

I did not include cover and substrate components in the
bioenergetic-HSC model.

However, observational data collected in Hunt

Creek demonstrated that brook trout seek out velocity shelters and
combination cover types (those that provide both velocity shelter and
visual isolation).

I speculate that this results from the fish's desire

to evade predators and to increase the net benefit of microhabitats by
reducing the cost of swimming.

Therefore, the relationship between

focal point velocity and mean column velocity does reflect the use of
cover as it affects the fish's focal point velocity choice.

In

addition, I propose that substrate composition in the immediate vicinity
of the fish is of minor importance when the fish is selecting a feeding
station to maximize its energetic gain during foraging.

It is more

likely that substrate composition upstream of the fish is more important
because it influences upstream invertebrate abundance and drift
composition (Minshall 1984) .

Furthermore, the substrate in Hunt Creek

is composed almost entirely of small and medium gravels.
The tradeoffs between potential energetic gain and predation risk
have been implicated as an important factor in fish habitat choice
decisions
1988).

(Mittelbach 1984; Gilliam and Fraser 1987; Huntingford et a l .

However, it does not appear that predator avoidance influenced

position choice decisions for foraging brook trout in Hunt Creek.
Piscivorous fish are only rarely present in the study sections of Hunt
Creek (E. A. Baker, personal observation).

A variety of avian,

mammalian, and reptilian predators of trout (Alexander 1979) are present
in the study area but I only rarely observed great blue herons Ardea
herodias and belted kingfishers Megaceryle alcyon and never observed any
mammalian or reptilian predators.

47

The bioenergetic-HSC presented here were developed from modeling
principles that have already been shown to accurately reflect the
position choice preferences of drift feeding stream fishes under field
and laboratory conditions (Hughes and Dill 1990; Hill and Grossman
1993).

Therefore, HSC constructed by this methodology may offer an

improvement to those based on frequency-of-use data, and may accurately
represent the suitability of foraging microhabitats in Hunt Creek.
However, further research is needed to validate the mechanistic
basis for construction and use of bioenergetically-derived HSC and to
improve the predictive capacity of bioenergetic-HSC.

For example, the

estimates of RD presented here were based on data collected from Arctic
grayling Thymallus arcticus feeding on zooplankton in a laboratory under
controlled conditions (Schmidt and O'Brien 1982).

Research is needed

regarding the relationship between prey size, actual RD, water velocity,
and light intensity that would provide a more accurate prediction of
foraging area as a function of current velocity and light intensity.
Further, quantitative measures of swimming speed for brook trout
intercepting drift would also improve the estimates of MC D .

It is

possible that foraging brook trout may travel at burst swimming speeds
during foraging attempts.

If this were the case, it would be necessary

to increase the foraging area estimates used in the calculation of net
benefits of a microhabitat and to adjust the depth suitability estimates
based on MCD.

If foraging brook trout were found to swim at burst

speeds during foraging this would result in even higher estimates of
optimal velocities.

The importance of cover and substrate composition

for drift feeding fish as they affect predator avoidance and swimming
cost also needs to be addressed.

48

Literature Cited
Alexander, G. R. 1979. Predators of fish in coldwater streams. Pages
153-170 in H. Clepper, editor. Predator-prey systems in fisheries
management. Sport Fishing Institute, Washington, D. C.

Alexander, G. R., and H. Gowing. 1976. Relationships between diet and
growth in rainbow trout (Salmo gairdneri) , brook trout (Salvelinus
fontinalis) , and brown trout

(Salmo trutta). Michigan Department

of Natural Resources Fisheries Research Report No. 1841, Ann
Arbor, MI.
Alexander, G. R. and E. A. Hansen. 1986. Sand bed load in a brook trout
stream. North American Journal of Fisheries Management 6:9-23.
Anderson, R. O. 1959. A modified flotation technique for sorting bottom
fauna samples. Limnology and Oceanography. 4:223-225.
Balon, E. K. 1980. Early ontogeny of the brook charr, Salvelinus
(Baione) fontinalis. Pages 631-666 in E. K. Balon, editor. Charrs,
salmonid fishes of the genus Salvelinus. Dr. W. Junk, The Hague,
Netherlands.
Barnes, R. D. 1987. Invertebrate Zoology, Fifth Edition. Saunders
College Publishing, Philadelphia, PA.
Beamish, F. ,w. H. 1980. Swimming performance and oxygen consumption of
the charrs. Pages 739-748 in E. K. Balon, editor. Charrs, salmonid
fishes of the genus Salvelinus. Dr. W. Junk, The Hague,
Netherlands.
Bisson, P. A. 1978. Diel food selection by two sizes of rainbow trout
(Salmo gairdneri) in an experimental stream. Journal of the
Fisheries Research Board of Canada 35:971-975.

49

Bovee, K. D.

1986. Development and evaluation of habitat suitability

criteria for use in the Instream Flow Incremental Methodology.
Instream Flow Information Paper 21. U.S. Fish and Wildlife Service
Biological Report 86(7). 235 pp.
Brandt, S. B., and J. Kirsch. 1993. Spatially explicit models of striped
bass growth potential in Chesapeake Bay. Transactions of the
American Fisheries Society 122:845-869.
Cummins, K. W. and J. C. Wuychek. 1971. Caloric equivalents for
investigations in ecological energetics. Internationale
Vereinigung fur theoretishe und angewandte Limnologie,
Verhandlungen 18:1-58.
Dorr, J. A., and D. F. Eschman. 1970. Geology of Michigan. University of
Michigan Press, Ann Arbor. 476 pages.
Elliott, J. M. and W. Davison. 1975. Energy equivalents of oxygen
consumption in animal energetics. Oecologia (Berlin) 19:195-201.
Everest, F. H., and D. W. Chapman. 1972. Habitat selection and spatial
interaction by juvenile Chinook salmon and steelhead trout in two
Idaho streams. Journal of the Fisheries Research Board of Canada
29:91-100.
Fausch, K. D. 1984. Profitable stream positions for salmonids: relating
specific growth rate to net energy gain. Canadian Journal of
Zoology 62:441-451.
Fausch, K. D. and R. J. White. 1981. Competition between brook trout
(Salvelinus fontinalis) and brown trout (Salmo trutta) for
positions in a Michigan stream. Canadian Journal of Fisheries and
Aquatic Sciences 38:1220-1227.

50

Gilliam, J. F., and D. F. Fraser. 1987. Habitat selection under
predation hazard: test of a model with foraging minnows. Ecology
68 :1856-1862.
Godin, J. J.-G. and R. W. Rangeley. 1989. Living in the fast
lane:effects of cost of locomotion on foraging behaviour on
Atlantic salmon. Animal Behaviour 37:943-954.
Goyke, A. P., and S. B. Brandt. 1993. Spatial models of salmonine growth
rates in Lake Ontario. Transactions of the American Fisheries
Society 122:870-833.
Grant, J. W. A., and D. L. G. Noakes. 1987.

Escape behaviour and use of

cover by young-of-the-year brook trout, Salvelinus fontinalis.
Canadian Journal of Fisheries and Aquatic Sciences 44:1390-1396.
Grant, J. W. A. and D. L. G. Noakes. 1988. Aggressiveness and foraging
mode of young-of-the-year brook charr, Salvelinus fontinalis
(Pisces, Salmonidae). Behavioral Ecology and Sociobiology 22:435445 .
Hill, J. and G. D. Grossman. 1993. An energetic model of microhabitat
use for rainbow trout and rosyside dace. Ecology 74 (3) :685-698.
Hughes, N. F. 1992. Ranking of feeding positions by drift-feeding Arctic
grayling (Thymallus arcticus)

in dominance hierarchies. Canadian

Journal of Fisheries and Aquatic Sciences 49:1994-1998.
Hughes, N. F., and L. M. Dill. 1990. Position choice by drift-feeding
salmonids: models and test for Arctic grayling (Thymallus
arcticus)

in subarctic mountain streams, interior Alaska. Canadian

Journal of Fisheries and Aquatic Sciences 47:2039-2048.
Huntingford, F. A., N. B. Metcalfe, and J. E. Thorpe. 1988. Choice of
feeding station in Atlantic salmon, Salmo salar, parr: effects of

51

predation risk, season and life history strategy.

Journal of Fish

Biology 33:917-924
Jones, D. R., J. W. Kiceniuk, and O. S. Bamford. 1974. Evaluation of the
swimming performance of several fish species from the Mackenzie
River. Journal of the Fisheries Research Board of Canada 31:16411647 .
McNicol, R. E., E. Scherer, and E. J. Murkin. 1985. Quantitative field
investigations of feeding and territorial behaviour of young-ofthe-year brook charr, Salvelinus fontinalis. Environmental Biology
of Fishes 12 (3) :219-229.
Merritt, R. W. and K. W. Cummins, editors. 1984. An Introduction to the
Aquatic Insects of North America 2—

Edition. Kendall/Hunt

Publishing Company of America. Dubuque, Iowa.
Milhous, R. T . , M. A. Updike, and D. M. Schneider. 1989. Physical
Habitat Simulation System Reference Manual-Version II. Instream
Flow Information Paper No. 26. U.S. Fish and Wildlife Service
Report 8 9(16) .
Minshall, G. W. 1984. Aquatic insect-substratum relationships. Pages
358-400 in Resh, V. H., and D. M. Rosenberg, editors, The Ecology
of Aquatic Insects. Praeger Publishers, New York, NY.
Mittelbach, G. G. 1984. Predation and resource partitioning in two
sunfis.hes (Centrarchidae) . Ecology 65:499-513.
Nielsen, J. L. 1992. Microhabitat-specific foraging behavior, diet, and
growth of juvenile coho salmon. Transactions of the American
Fisheries Society 121:617-634.

52

Rogers, L. E., R. L. Buschbom, and C. R. Watson. 1977. Length-weight
relationships of shrubsteppe invertebrates. Annals of the
Entomological Society of America 70:51-53.
Rose, K. A., and J. H. Cowan, Jr. 1993. Individual-based model of youngof -the-year striped bass population dynamics. I. Model description
and baseline simulations. Transactions of the American Fisheries
Society 122:415-438.
Schmidt, D., and W. J. O'Brien. 1982. Planktivorous feeding ecology of
Arctic grayling (Thymallus arcticus). Canadian Journal of
Fisheries and Aquatic Science 39:475-482.
Schoener, T. W. 1971. Theory of feeding strategies. Annual Review of
Ecology and Systematics 2:369-404.
Smith, J. J. and H. W. Li. 1983. Energetic factors influencing foraging
of juvenile steelhead trout, Salmo gairdneri. Pages 173-180 in
Noakes, D. L. G., D. G. Lindquist, G. S. Helfman, and J. A. Ward,
eds. Predators and Prey in Fishes. Dr. W. Junk Publishers, The
Hague, Netherlands.
Smock, L. A. 1980. Relationships between body size and biomass of
aquatic insects. Freshwater Biology 10:375-383.
Somerville, P. N. 1958. Tables for obtaining non-parametric tolerance
limits. Annals of Mathematics and Statistics 29:599-601.
Thomas, J. A., and K. D. Bovee. 1993. Application and testing of a
procedure to evaluate transferability of habitat suitability
criteria. Regulated Rivers: Research and Management 8:285-294.
Tippets, W. E. and P. B. Moyle. 1978. Epibenthic feeding by rainbow
trout

(Salmo gairdneri) in the McCloud River, California. Journal

of Animal Ecology 47:549-559.

53

Zar, J. H. 1984. Biostatistical Analysis, 2—

Edition. Prentice -Ha 11,

Inc. Englewood Cliffs, New Jersey. 718 pp.

54

CHAPTER 2

COMPARISON OF PREDICTED HABITAT CHANGE AND BROOK TROUT POPULATION
RESPONSE TO A SIMULATED IRRIGATION WITHDRAWAL IN HUNT CREEK, MICHIGAN

ABSTRACT
I used three types of habitat suitability criteria (nocturnal, diurnal
frequency-of-use, and diurnal bioenergetically based) to evaluate the
impacts of a seasonal 50% withdrawal on brook trout habitat in Hunt
Creek, MI by use of the Physical Habitat Simulation System (PHABSIM).
Young of the year diurnal Weighted Usable Area (WUA) increased 16-31%,
regardless of HSC used.

Yearling and older diurnal WUA calculated from

frequency-of-use HSC decreased 1.5-1.9%.

Yearling and older WUA

calculated from bioenergetically based HSC increased for fish <.15 cm and
decreased for fish >15 cm.

Nocturnal WUA increased 18-29% for young of

the year fish and 9-15% for yearling and older fish.

Biannual brook

trout density estimates in treatment and control sections of Hunt Creek
were very similar for ten years preceding the withdrawal and during the
withdrawal period and not significantly different after the treatment.
The PHABSIM model predicted a summer withdrawal equal to 88% of baseflow
would produce a statistically detectable reduction in brook trout
density and that yearling and older brook trout habitat would be reduced
more than young of the year habitat at that level of flow reduction.

55

Introduction
Changes in stream flow regime can influence the ecology of stream
fishes in a variety of ways (Orth 1987).

Stream flows are important in

determining reproductive success (Starrett 1951), fish community
structure and habitat use (Bain et al. 1988), and habitat availability
(Kraft 1972) .

In midwestern trout streams the input of groundwater is

recognized as an important abiotic factor influencing trout populations.
For example, Latta (1965) found a significant positive relationship
between young of the year brook trout Salvelinus fontinalis density and
groundwater levels during a nine year study on the Pigeon River,
Michigan.

Similarly, White et a l . (1976) determined that trout streams

in Michigan and Wisconsin that had the most stable flow regime also had
the greatest trout abundance and standing crop.

Clearly, protecting

flows in midwestern trout streams is important for the maintenance of
healthy trout populations.
The Physical Habitat Simulation System (PHABSIM)

is the computer

based habitat modeling component of the Instream Flow Incremental
Methodology .(IFIM) that predicts stream habitat quality and quantity as
a function of discharge (Milhous et al. 1989).

PHABSIM was developed in

the western U.S. with the purpose of evaluating the impacts of changes
in streamflow on stream habitat.

The PHABSIM system is widely used in

the western U.S. to evaluate the impacts of water development projects
on stream resources and is a legal requirement in the state of
California (Reiser et al. 1989).

However, the PHABSIM system has only

recently been applied to streams in the midwestern U.S. where the
geology, hydrology, and composition of the fauna are distinctly
different from western streams

(Gowan 1984; Bovee et a l . 1994).

56

The PHABSIM system works on the premise that water depth, water
velocity, substrate, and cover are the four microhabitat parameters that
determine a fish's use of habitat.

Data input into PHABSIM are in the

form of habitat suitability criteria (HSC) and habitat availability data
for these four parameters for the species and life history stage of
interest and the stream under investigation (Figure 18; Milhous et a l .
1989).

The output of a PHABSIM analysis is a measure of habitat,

Weighted Usable Area (WUA).

WUA is a measure of the amount of habitat

in a stream that is suitable for the target species and life stage and
is calculated for a range of simulated discharges in the stream of
interest.

The resulting WUA versus discharge relation is used to

evaluate proposed changes in the flow regime in a stream and to predict
the impacts of altered flows on the fish population(s)

in the stream.

An assumption of the PHABSIM system is that WUA is linearly and
positively related to fish standing crop (Bovee 1970; Orth and Maughan
1982; Mathur et al. 1985).

Orth and Maughan (1982) and Milhous et al.

(1989) reviewed the computational procedures of the PHABSIM modeling
procedure and the assumptions associated with a PHABSIM analysis.
The PHABSIM system has been criticized for several reasons
including the technical simplicity of the habitat calculations and the
complexity and expense of its application.

More importantly,

it has

been critized for the assumption that WUA is positively related to fish
abundance (Mathur et al. 1985; Morhardt 1986; Scott and Shirvell 1987;
Morhardt and Mesick 1988; Reiser et a l . 1989; Armour and Taylor 1991).
Numerous studies have attempted to document a relationship between WUA
and fish population parameters with limited success (for example: Orth
and Maughan 1982; Gowan 1984; Shirvell and Morantz 1983; Conder and

57

H a b ita t Availability
(H ydraulic m o d e l of s tre a m )

H a b ita t Suitability Criteria
(H S C )

P H A B S IM Model

I

Discharge (Q )
Assumption

▼
<

§

Fish Abundance or Biomass

Figure 18.

Logical sequence of the PHABSIM model process.

Data are

entered in the form of habitat suitability criteria and
hydraulic status of the stream.

The model predicts weighted

usable area over a range of discharges.

Weighted usable

area is assumed to be linearly related to fish abundance or
biomass.

58

Annear 1987; Scott and Shirvell 1987) .

However, none of these studies

used experimental manipulation of streamflow to measure the response of
the fish population in a control site or a pretreatment period.

The

objectives of this study were to: 1) evaluate the impacts of a simulated
irrigation withdrawal on the brook trout Salvelinus fontinalis
population in Hunt Creek; 2) evaluate the PHABSIM system in Hunt Creek
by comparing the output of the PHABSIM analysis to the observed response
of the brook trout population; 3) evaluate nocturnal resting HSC and
bioenergetically derived HSC for foraging

microhabitats

(bioenergetic-

HSC, Chapter l) to determine if these alter the PHABSIM predictions.
Methods
Beginning on June 1 or 2 and continuing through August 31, 1991-94
I diverted approximately 50% of the summer stream flow around the
treatment section of Hunt Creek to simulate the. effects of a seasonal
water withdrawal for irrigation.

During the withdrawal period I

monitored the traps on the bulkheads at the upstream and downstream ends
of the treatment section (Figure 1, Chapter 1) to determine if the trout
moved in response to the dewatering.

I did not operate traps during the

period of full flow (September 1-May 31) and, brook trout were free to
move into and out of the treatment section.

In addition to the 50%

reduction of summer flow, I reduced flow to 25% of summer baseflow (0.11
m3 's 1) during a three day period in August, 1993 to collect hydraulic
data needed to calibrate the hydraulic modeling component of PHABSIM.
I modeled diurnal and nocturnal habitat of young of the year (<8.9
cm total length) and yearling and older (>8.9 cm total length) brook
trout habitat based on several different sets of habitat suitability
criteria (use-HSC, Chapter 1).

I used two sets of habitat suitability

59

criteria developed from frequency-of-use data collected in sections B
and C of Hunt Creek in 1991-93 over the range of flows from baseflow
- 1 ) to 25% of baseflow (0.11 m 3 's
- 1 ).
(0.46 m 3 's
observations of habitat use in diurnal conditions

One set was based on
(diurnal use-HSC) and

the other set was based on observations of habitat use in nocturnal
conditions

(nocturnal use-HSC).

I also modeled diurnal foraging habitat

from habitat suitability criteria based on bioenergetic cost and benefit
models (bioenergetic-HSC) from data collected in 1993-94 in section B
(Chapter 1) .

Bioenergetic-HSC were size -specific for brook trout and

were constructed for fish at 5 cm intervals between 5 and 20 cm total
length.
I used a representative reach approach for modeling the habitat in
section B of Hunt Creek with PHABSIM.

To select representative reaches

I first measured and marked section B into approximately 50 m contiguous
reaches, omitting the small area of impounded water at the downstream
end of the section as well as the short reach of disturbed habitat
immediately downstream of the diversion bulkhead (Figure 1, Chapter 1).
Two of the 50 m reaches in section B were randomly selected

to model by

use of PHABSIM (reaches B2 and B4 in Figure 1, Chapter 1).

I

established transect locations in each of the reaches, and used changes
in meso habitat
placement.

(riffle, run, pool) within the reach to guide transect

Substrate and cover along each transect were classified by

use of

the same codes used for the brook trout habitat use observations

(Table

2, Chapter 1).

PHABSIM cell.

The dominant cover type was recorded for each

I collected flow data in the two reaches in section B at

three discharges; 0.46, 0.23, and 0.11 m3 's_1: depths were measured to
the nearest cm with a wading rod and velocities to the nearest cm's’1

60

with either a mechanical Pygmy-Gurley or an electronic March-McBirney
current meter.

I compared velocity measurements obtained from both

meters at the same location in Hunt Creek on several occasions and could
not detect any differences between the meters.

I calibrated the PHABSIM

model and simulated habitat over a range of flows from summer baseflow
to 0.01 m ^ 's ^ (2% of baseflow) for the reaches in sections B.
Brook trout population abundance data were collected in
cooperation with Michigan DNR staff in all four sections of Hunt Creek
by conducting mark-recapture electrofishing in April and September of
each year of the study and following the same protocol that has been
followed for nearly five decades in Hunt Creek (Alexander and Hansen
1986).

The entire length of Hunt Creek was sampled from the downstream

end of section Z to the upstream end of section C (approximately 6.5 km
of stream).
72 hours.

Recapture sampling efforts always followed marking within

All fish were measured to the nearest 0.25 cm and a subsample

of 10 fish per 2.5 cm length interval from each section were weighed to
the nearest 0.1 g in spring and fall, 1993 and 1994.

I used Bailey's

modification (Bailey 1951) of the Petersen method to estimate brook
trout population density for 2.5 cm length intervals and used the
equations in Ricker (1975) to estimate 95% confidence limits.
I measured brook trout lengths and weights during electrofishing
sampling and developed length-weight regressions for each section and
season.

I tested the length-weight regressions developed from each

section, season and year with ANCOVA and determined there were no
significant differences between the regressions (F=0.97 df=5,2723,
p=0.43).

Therefore, the length-weight data were pooled and the

regression was recalculated and used to predict brook trout weights for

61

use in biomass calculations.

I multiplied the number of fish in each

2.5 cm length interval by the weight of a fish at the midpoint of the
interval, obtained from the length-weight regression, and then summed
biomass over all length intervals.
I formulated hypotheses concerning the impact of the withdrawal on
the brook trout population numbers and total standing crop in section B
(treatment section) based on the predicted relationship between WUA and
discharge for section B and the assumed positive linear relationship
between WUA and fish population standing stock and population density.
I tested these hypotheses by use of BACI statistics
al. 1986, 1992).

(Stewart-Oaten et

For the analysis I calculated spring and fall mean

population density estimates for sections A and Z combined (control
section) and determined differences between these estimates and the
density estimates for section B for the ten year period preceding the
treatment and the four years of treatment.
the total biomass data for the same period.

I repeated the analysis for
I used the Student's t-test

to compare pretreatment period mean difference with the treatment period
mean difference.
I did not include the upstream nontreatment section C in the
statistical analysis because the brook trout population in section C was
influenced by environmental factors that did not impact the population
in sections A, B, and Z.
Results
Summer baseflow in section C at the confluence with Fuller Creek
(Figure

1

, Chapter

1

) was approximately 0.23 m 3 's -1 .

Fuller Creek

delivered an additional 0.23 m 3 's -1 to produce a summer baseflow of 0.46
m 3 's -1 through the treatment section (B).

62

A small tributary enters Hunt

Creek just downstream of the treatment section and delivers
approximately 0.11 m 3/s, increasing discharge in Hunt Creek to
approximately 0.57 m 3/s.

Hunt Creek continues to gain flow as'it flows

downstream and out of the research area.
Characteristics of Study Reaches
Reaches B2 and B4 were 47.1 and 55.2 m long respectively.

I

established 19 transects and measured habitat availability at 385
locations (cells) in reach B 2 .

I established 21 transects and measured

habitat availability at 530 locations (cells) in reach B 4 .

Mean

distance between transects was 2.5 m in reach B2 and 2.6 m in reach B4,
and the maximum distance between any two transects was 6.6 m in reach B2
and 5.2 m in reach B 4 .
Several important differences existed between the two reaches used
in habitat modeling, and these affected the output of the PHABSIM
analysis.

First, the mean water surface slope at baseflow (0.46 m 3 ’s -1 )

in reach B2 (4.2 m'km"1) was nearly twice the slope (2.2 m'km’1) in
reach B 4 .

As a result, the mean water velocities were greater in reach

B 2 : at baseflow the mean of all mean column velocity measurements was 34
c m ’s 1 in reach B2 and 28 cm's’1 in reach B 4 .

Differences in velocity

distributions between modeled reaches were significant
test, p=0.002).

(Mann-Whitney U

A second difference between the modeled reaches was

that the mean channel width in reach B2 (4.15 m) was less than in B4
(5.00 m ) .

The substrate and cover composition in the two modeled reaches was
similar.

In both reaches the substrate was composed primarily of

gravels less than 2.5 cm diameter.
both reaches.

Sand and silt were also common in

Substrate composition in the reaches differed

63

2

significantly (X =64.1, df=9, p<0.001), primarily due to the presence
of more cells in reach B2 with large gravel than in B4 and a greater
number of cells in reach B4 with the substrate embedded more than 25%.
The majority of the cells (>93%) in both reaches had instream
cover present, either in the form of a velocity shelter or a combination
cover type.
the reaches

Availability of cover composition did not differ between
(X2=i.l3, df=2, p<O.Ol).
Habitat Suitability Criteria

The HSC developed from frequency-of-use data and from bioenergetic
models are fully presented in Chapter 1 and are only summarized here
(Table 3).

The diurnal use-HSC represent the suitability of foraging

microhabitats for brook trout in Hunt Creek because only data from
actively foraging fish were used (87% of all diurnal observations)
the construction of diurnal use-HSC.

in

In contrast, nocturnal use-HSC

were based on observations of inactive fish (93% of all nocturnal
observations) and represented the suitability of resting microhabitats
(Table 3).

All of the inactive brook trout observed at night were in

direct contact with the substrate, and some were burrowed into
vegetation (primarily watercress, Nasturtium officinale) or wedged
between sticks.
PHABSIM Model Results
The relation between surface area and discharge for the two
modeled reaches in section B indicated that a 50% reduction in summer
stream flow in the treatment section resulted in a very minor loss of
stream surface area (Figure 19).
2

reduced from 206 to 195 m '100 m

In reach B2, total surface area was
-1

, a reduction of 5.6%.

In reach B4,

total surface area decreased 292 to 275 m 2 '100 m -1 , a reduction of 5.7%.

64

Table 3.

Optimal and suitable habitat suitability criteria values for
diurnal and nocturnal periods.
Chapter 1.

Data for all HSC are from

Bioenergetic-HSC were size specific and are

presented here only for 7.5 and 15 cm fish (sizes equivalent
to young of the year and yearling and older fish
respectively).

Mean Column
Depth

Velocity

(cm)

(c m ‘s ’1)

Substrate

Cover

Optimal Range

15-34

6-30

1-5.4

3

Suitable Range

3-67

0-66

1-5.4

2 & 3

Optimal Range

27-55

6-27

1-5.4

3

Suitable Range

12-85

0-98

1-5.4

2 & 3

Optimal Value

Varies

28

1-5.4

3

Suitable Range

>1.7

1-46

1-5.4

2 & 3

Optimal Value

Varies

41

1-5.4

3

Suitable Range

>3.3

3-63

1-5.4

2 & 3

Optimal Range

12-29

5-23

1-5.4

3

Suitable Range

1-73

0-39

1-5.4

2 & 3

Optimal Range

20-46

4-22

1-5.4

3

Suitable Range

7-73

0-52

1-5.4

2 & 3

Diurnal U8e-HSC
young of the year

yearling and older

Diurnal
Bioeneraetic-HSC
7.5 cm

15 cm

Nocturnal Use-HSC
young of the year

yearling and older

65

300

«

B4

Total Area (m2100 m'1)

250

200

B2

150 •

100

50

1/2 baseflow
,

.005

0 1

0.15

0.2

iL

baseflow

— ^ n J/--- ]

______

0.25

0.3

0.35

0.4

045

05

Discharge (m 3s'1)

Figure 19.

Total area (m2 ‘ioo m'1) in relation to discharge (n^'s"1)
for reaches B2 (solid diamonds) and B4 (open squares) in
Hunt Creek.

66

Model results also predicted that reducing flow in section B by 98% to a
-1
discharge of 0.01 m 3 's

would decrease total surface area to 132 m 2
' ‘l00

m 1 in reach B2 (35.9% loss) and 176 m2 '100 m 1 in reach B4 (39.7%
loss).

The difference in the total surface area estimates between the

two modeled reaches is due to the greater width and lower slope in reach
B4 .
The 50% reduction of summer stream flow actually resulted in an
increase in WUA for young of the year fish based on diurnal use-HSC in
reaches B2 and B4 (Figures 20 and 21).

Suitable habitat area increased

27% in reach B2 and 16% in reach B4 with a 50% reduction of summer flow.
The maximum WUA value over the range of discharges modeled occurred at a
discharge of 0.17 m 3 's 1 (37% of baseflow) in both modeled reaches.

In

reach B2 maximum WUA was 160 m 2 ‘100 m -1 and in reach B4 it was 217
m 2 "100 m -1 .
section B

The PHABSIM model also predicted that if discharge

in

of Hunt Creek was reduced to 0.01 m3 's 1 young of the year
2

-

1

2

-

1

would only be reduced to 76 m '100 m

in reach B2 and 82 m '100 m

reach B 4 . This represents a reduction

in young of the year WUA

WUA
in

of37%

and 56% in reaches B2 and B4 respectively, with a 98% reduction of
summer baseflow.
In contrast to the young of the year WUA estimates, yearling and
older diurnal WUA in the two modeled reaches was slightly reduced with a
50% reduction in baseflow (Figures 20 and 21).

The yearling and older
2

diurnal WUA in reach B2 decreased from 134 to 132 m '100 m
1.5% of suitable habitat area.

-1

, a loss of

Yearling and older WUA in reach B4

2
-1
decreased from 162 to 159 m '100 m , a reduction of 1.9%.

The model

also predicted that a 98% reduction in summer stream flow would reduce
yearling and older WUA by 65% in reach B2 and 70% in B 4 .

67

250 -

I
I

WUA (m2 100 m'1

200

+

yearling
and older

150 ^

100

young of
the year

-r

1/2 baseflow

0

0.05

0.1

0.15

0.2

0.25

baseflow

0.3

0.35

0.4

0.45

0.5

Discharge (m3s*1)

Figure 20.

Diurnal WUA (m2 *100 m -1 ) estimates derived from diurnal useHSC as a function of discharge (m3 *s_1) for young of the
year (solid diamonds) and yearling and older brook trout
(open squares) in reach B2 of Hunt Creek.

68

250 1

young of
the year

WUA (m2iOO m

200

yearling
and older

150

100

1/2 baseflow

0

0.05

0.1

0.15

0.25

0.2

baseflow

0.3

0.35

0.4

0.45

0.5

Discharge (m3s'1)

Figure 21.

2

Diurnal WUA (m ‘100 m

-1

of stream) estimates derived from

diurnal use-HSC as a function of discharge (m3 's-1) for
young of the year (solid diamonds) and yearling and older
brook trout (open squares) in reach B4 of Hunt Creek.

69

Diurnal WUA estimates based on bioenergetic-HSC for brook trout 5
and 7.5 cm total length (length range equivalent to young of the year
fish) also increased with the 50% reduction in summer stream flow
(Figures 22 and 23).

In reach B2 WUA estimates increased approximately

24% and 21% for 5 and 7.5 cm fish, respectively.

The magnitude of the

increase was slightly lower than the 27% increase in WUA predicted for
young of the year fish from the diurnal use-HSC.

In reach B4, WUA

increased approximately 31% and 17% for 5 and 7.5 cm fish, respectively.
This increase in WUA estimates was slightly higher than the 16% increase
WUA for young of the year fish in reach B4 from diurnal use-HSC.
The maximum WUA estimates for 5 and 7.5 cm fish in reach B2
-1
occurred at Q=0.11 m 3 's

-1
and Q=0.17 m 3 's

respectively.

This is

similar to the predicted discharge of 0.17 m3 's 1 that yielded maximum
WUA for all young of the year fish from the diurnal use-HSC.

The

maximum WUA values for 5 and 7.5 cm fish in reach B4 occurred at Q=0.17
3-1
m ‘s

3-1
and Q=0.23 m 's

respectively.

This is also similar to the

results for reach B4 from the diurnal use-HSC which indicated maximum
WUA occurred at Q=0.17 m 3 ‘s -1 for all young of the year fish.

Finally,

the model predicted a 98% reduction in flow (Q=0.01 m3 ’s’1) would reduce
WUA in reach B2 for 5 and 7.5 cm fish approximately 75% and 86%,
respectively and 82% and 91%, respectively in reach B 4 .

The magnitude

of the WUA reduction based on the bioenergetic-HSC is nearly twice that
predicted on the basis of diurnal use-HSC.
For yearling and older brook trout, the PHABSIM model predicted
WUA would increase with the 50% reduction in flow for fish 10 cm to 15
cm total length in reach B 2 , but would decrease for larger fish in B2
and all fish 10 cm or larger in B 4 .

70

WUA increased 16% for 10 cm fish,

250

Fish Length (cm)

5
7.5
10

WUA (m2100 m'1

200

12.5
15
17.5
150

20

100

baseflow

1/2 baseflow

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

0.45

0.50

Discharge (m3s'1)

Figure 22.

Diurnal WUA (m2 *100 m -1 of stream) estimates derived from
bioenergetic-HSC as a function of discharge (rr^’s’1) for
brook trout in reach B2 of Hunt Creek.

71

Fish Length
5
- - 7.5

WUA (m2'100 m'1

200 4
12.5
- 15
17.5
150

100 !

1/2 baseflow

0.00

0.05

0.10

0.15

0.20

0.25

baseflow

0.30

0.35

0.40

0.45

0.50

Discharge (m3,s*1)

Figure 23.

Diurnal WUA (m2 *100 m-1

of stream) estimates derived from

bioenergetic-HSC as a function of discharge
brook trout in reach B4 of Hunt Creek.

72

( m ^ ' s 1) for

10% for 12.5 cm fish and 2% for 15 cm fish (Figure 22).

In reach B 2 ,

WUA decreased 5% and 11% for 17.5 and 20 cm fish, respectively.

The

reduction in WUA for fish in reach B4 was greater than in reach B2
(Figure 23), and ranged from 4% to 37% for fish 10 cm to 20 cm,
respectively.

Reductions in WUA were greatest for the largest fish in

both reaches (Figures 22 and 23).

The model predicted that WUA for fish

between 10 and 20 cm total length would be reduced between 75% and 98%
in reach B2 and from 82% to 99% in reach B4 if summer stream flow was
reduced 98% to 0.01 m 3 's"1 .
The PHABSIM model results for nocturnal habitat were similar to
the results from diurnal use-HSC.

Nocturnal WUA increased with the 50%

reduction in stream flow for both young of the year and yearling and
older brook trout in section B (Figure 24).

WUA for young of the year

fish increased 18% in reach B2 and 29% in reach B 4 .

At 98% reduction in

flow, the model predicted greater WUA than at baseflow in reach B2 and a
16% reduction in reach B 4 .
The 50% reduction in flow increased nocturnal WUA for yearling and
older brook trout 9% in reach B2 and 15% in reach B 4 .

In contrast to

the results for young of the year nocturnal habitat, the model predicted
that a 98% reduction in flow would result in substantial reductions (42%
and 56% in B2 and B4, respectively)

in yearling and older nocturnal WUA.

PHABSIM Predictions
The analysis of the PHABSIM model output yielded two different
sets of hypotheses concerning the impact of the 50% flow reduction on
the brook trout population.

The first hypothesis is that standing stock

and density would not change in rsponse to the withdrawal for young of
the year or yearling and older brook trout.

73

This conclusion stems from

250 T

WUA (m2100 m'1

200
B4 young of the year
150
B4 yearling and older
B2 young of the year

100

B2 yearling ana older
50

baseflow

1/2 baseflow

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

0.45

0.50

Discharge (m3 s'1)

Figure 24.

- 1
Nocturnal WUA (m2 "100
m

) as a function of discharge

(m3 's

1) estimates for young of the year and yearling and older
brook trout in reaches B2 and B4 of Hunt Creek.

74

the model predictions based on use-HSC that diurnal and nocturnal WUA
increased substantially in both modeled reaches for young of the year
fish, diurnal WUA decreased only slightly for yearling and older fish
and nocturnal WUA increased for yearling and older fish when summer flow
was reduced 50% (Figures 20, 21, and 24).

The second hypothesis is

based on the PHABSIM predictions generated with the bioenergetic-HSC.
The WUA data from reaches B2 and B4 predict no change in standing stock
or density of young of the year fish in section B2 (Figures 22 and 23).
However, the WUA data predicted that density or standing stock of
yearling and older fish in section B would be reduced, particularly for
fish 20 cm or larger.

The expected reduction in density was as much as

16-37%, depending on fish size (Figure 23).

The expected reduction in

brook trout abundance was based on an assumed one to one relationship
between WUA and fish standing stock (Bovee 1978).
The summer withdrawal had no effect on brook trout density or
biomass.

The pretreatment period mean difference in density between the

control and treatment sections was not significantly different from the
treatment period mean difference for young of the year (Figure 25,
Student's t=0.43, p=0.65, df=12) or yearling and older fish (Figure 26,
Student's t=1.21, p=0.28, df=12).

The pretreatment period mean

difference in total standing crop was also not significantly different
from the treatment period mean difference (Student's t, p=0.97, df=12).
I tested the predictions of the PHABSIM model generated from the
bioenergetic-HSC using BACI statistics.

I calculated the densities of

brook trout in the control and treatment sections of Hunt Creek for 2.5
cm length intervals from 5 to 20 cm (e.g. 5-7.49 cm) and then tested for
impacts on each of these length classes of fish.

75

I found a significant

8000

Section

7000
6000

Fish ha

5000
4000
3000 -

2000
1000

-

— t—
80

81

—

-t-

t ----------------- 1-------------------1—

82

83

84

85

86

—i

-t -

87

88

89

90

91

92

93

94

Year

Figure 25.

Fall young of the year brook trout population density
(fish'ha-1) estimates for sections A, B, C, and z of Hunt
Creek for 1981-1994.
94.

The withdrawal period was from 1991-

Error bars represent 95% confidence limits of the mean.

76

Section
_B — C

3500

- h- Z

3000

Fishha

2500

2000
1500
1000
500

1—
80

81

82

83

i

-t-

84

85

86

87

88

89

90

91

92

93

94

Year

Figure 26.

Fall yearling and older brook trout population density
(fish'ha’1) estimates for sections A, B, C, and Z of Hunt
Creek for 1981-1994.
94.

The withdrawal period was from 1991-

Error bars represent 95% confidence limits of the mean.

77

difference between the pretreatment and treatment period differences in
densities only for the fish in the 10-12.S cm length interval
27, Student's t=4.01, p=0.002, df=12).

(Figure

However, the change in this size

group was opposite the predicted change: from 1981-1988, density of 1012.5

cm brook trout decreased, and from 1989-1993 it increased (Figure

27).

For all

from

the withdrawal (p values between 0.09 and 0.78) .
Because

other length groups of fish there was no measurable impact

I was interested in the magnitude of the change in fish

density that would be needed to detect a difference I calculated the
minimum detectable difference and statistical power (Zar 1984) for the
BACI analysis.

A minimum difference of approximately 823.1 and 842.3

fish'ha 1 would be necessary between the pretreatment and treatment mean
differences to conclude that there was an impact on young of the year
and yearling and older fish, respectively.

Also, power estimates for

the BACI analysis were all less than 0.20, indicating that if there was
an impact of the experimental treatment, the chances of detecting it
were 20% or less at a=0.05.
I also used the minimum detectable difference estimates to predict
the reduction in discharge necessary to produce a measurable impact on
the fish population.

For this calculation I subtracted the minimum

detectable difference estimate from the mean density of fish in the
control section (sections A and Z combined) over the pretreatment period
to estimate the mean density of fish in the treatment section that would
produce a statistically detectable result.

I then calculated the

proportional change in density that this represents, relative to the
pretreatment density and multiplied the WUA at base flow by this
proportion to determine the change in WUA needed to obtain a detectable

78

Section
-A -® -B - * - C - * - Z

1400 T

1200 i

I

Fish ha

1000

\

800
[

600

j
!

400 f

200

—

80

81

82

83

84

85

t-------- 1-------- 1—

86

87

88

89

90

91

92

93

94

Year

Figure 27.

Fall brook trout population density (fish'ha 1) estimates
for fish 10-12.5 cm total length in sections A, B, C, and Z
of Hunt Creek for 1981-1994.
1991-94 .

79

The withdrawal period was from

change in density.

From this, I estimated the discharge reduction that

would produce this reduced WUA.
The reduction in WUA that would be needed to produce a
statistically detectable impact on both young of the year and yearling
and older brook trout densities in section B was approximately 50% of
the WUA available at baseflow (Table 4).

The discharge values that

would produce a statistically detectable result differed depending on
the type of HSC used to calculate WUA.

The WUA curves calculated from

diurnal use-HSC indicated that flow would need to be reduced to 0.02
m 3 ‘s'1 (reach B2) and 0.01 m ^ ’s'1 (reach B2) to reduce young of the year
3-1
densities 50% and to 0.03 m 's

3-1
(reach B2) and 0.05 m 's
(reach B4)

to reduce yearling and older densities 50%(Table 4).

This

represents a

reduction in flow of at least 88% to reduce fish densities a detectable
amount.

The discharge needed to produce a measurable impact on fish

densities based on the WUA curves calculated from bioenergetic-HSC was
between 0.02 and 0.0'6 m3 's 1 for

fish equal to young of the year size

and between 0.05 and 0.16 n^'s"1

for fish equal to yearling and older

size (Table 4).

Because the nocturnal WUA estimates at a 98% reduction

in flow were only slightly lower than those at baseflow, it was
impossible to evaluate a discharge which would produce a measurable
decrease in population standing crop from nocturnal WUA.
Fish Movement
Brook trout moved downstream and out of the treatment section
throughout summer, 1990, a pretreatment year (Table 5).

Fish movement

was relatively steady throughout the summer because the number of brook
trout caught on any date never exceeded four.

Trap data for the

treatment period (1991-94) were inconsistent between years and even

80

Table 4.

Estimated discharges (Q, m3s'1) for reaches B2 and B4 and corresponding WUA
(m 2 100 m-1) estimates that would produce a statistically detectable reduction in brook
trout density or biomass in section B.

U se-HSC
Q

WUA

Q

W UA

(m V )

(m 2 100 m ‘1)

(m 3 s ‘1)

(m 2 100 m '1)

<0.01

76

5 cm

0.02

43

7.5 cm

0.04

56

10 cm

0.05

53

12.5 cm

0.07

58

15 cm

0.08

55

17.5 cm

0.10

68

20 cm

0.12

59

5 cm

0.03

60

7.5 cm

0.06

76

10 cm

0.09

81

12.5 cm

0.12

82

15 cm

0.14

85

17.5 cm

0.16

85

20 cm

0.18

83

Fish S ize

Reach
B2

young of the year

yearling and older

B4

B io en eraetic-H S C

young of the year

yearling and older

0.03

66

0.02

100

0.05

83

81

Table

5.

S i z e a n d n u m b e r of b r o o k trout c a u g h t

in i n c l i n e d s c r e e n

traps during the treatment period (1991-94) and for the
summer prior to withdrawal from traps at the upstream and
downstream bulkheads.

Bulkhead

tips bream Bulkhead

mean length (cm)

n

10.9

69

39

10 .3

43

11.3

30

9.0

53

1993

12.6

199

12 .6

132

1994

8.5

183

8 .3

50

Year

mean length (cm)

1990

not recorded

1991

10.2

1992

n

82

between the upstream and downstream traps.

However, brook trout in

section B did not respond to the withdrawal consistently by moving
downstream (Table 5). The rate of fish movement into section B was
similar to the rate of movement out of the section in 1991-93, and
movement into section B exceeded the rate of movement out in 1994 by a
factor of four.

However, the number of trout that moved into and out of

the treatment section during any one summer was small in comparison to
the total number of trout in the treatment section (mean=l269 fish in
fall, 1991-94).

The mean length of fish captured in traps was also

similar among years and between the upstream and downstream traps.
Discussion
The brook trout population in Hunt Creek was not affected by the
50% reduction in summer stream flow and the habitat analysis predicted
no effects.

Hunt Creek is a groundwater fed stream with stable

discharge and high quality physical and biotic habitat under summer
baseflow conditions.

Given the high quality of the habitat under summer

baseflow conditions it is not surprising that habitat was not severely
impacted by a 50% reduction in baseflow.

The results of the PHABSIM

modeling support this conclusion because the diurnal WUA estimates for
young of the year fish were substantially higher at reduced flow and WUA
was only slightly reduced for yearling and older fish.

These results

were similar whether use-HSC or bioenergetic-HSC were used to estimate
WUA.

Furthermore, nocturnal WUA increased for young of the year and

yearling and older fish as a result of the reduced flow.
Because the PHABSIM model only varied the depths and velocities
with changes in discharge, these two parameters determined the shape of
the WUA-discharge relation.

Therefore, the increased WUA estimates at

83

half of the mean summer stream flow were due to either more locations
with improved depth suitability, improved velocity suitability, or both.
The change in the velocity availability was likely the primary cause of
the increased WUA estimates.

This conclusion is based on the

observation that the optimal mean column velocities from the use-HSC
were less than the mean of the mean column velocity measurements at
summer baseflow in section B of Hunt Creek for young of the year and
yearling and older fish.

Therefore, as discharge decreased,

locations

with greater than optimal mean column velocity at summer baseflow became
more suitable because mean column velocity decreased as discharge
decreased.

It is also likely that depth suitability in parts of the

modeled reaches decreased as discharge decreased.
The results from this study are very similar to those from a study
by Kraft (1972), who evaluated the impact of a seasonal withdrawal on
the brook trout habitat and population in a Montana stream.

He

dewatered a section of stream by up to 90% during the summer months and
monitored brook trout population density.

Brook trout moved from

shallow runs to pools as flow was reduced but the number of trout did
not change significantly (Kraft 1972) .

Brook trout did move out of the

dewatered section of the Montana stream, but not until the reduction in
flow was equal to 90% of mean flow.

Also, when fish moved out of the

test section it was in an upstream direction (Kraft 1972) .

Clothier

(1954) reported similar upstream movement of brook trout, brown trout
Salmo trutta, and rainbow trout Oncorhynchus mykiss during extreme
irrigation withdrawals in the Gallatin River, Montana.

Upstream

movement was not possible for the brook trout in the treatment section
of Hunt Creek, and downstream movement from the treatment section

84

generally matched movement rates from the upstream control section.

The

large number of brook trout caught in the traps in 1993 may have been a
result of the unusually wet summer.

Several heavy rains fell during the

summer and each was followed by large numbers of fish caught in the
traps.

For example, during the three day period of 7 to 9 June, 82 fish

were captured in the traps at the upstream and downstream bulkheads
fish at each bulkhead).

(41

These three days corresponded to a period of

heavy rainfall which ultimately caused the failure of a beaver dam just
upstream of section C.
Although physical habitat is important in determining fish
abundance and distribution in a variety of habitats, other biotic and
abiotic factors can influence fish abundance and distribution in streams
(Latta 1965; Chapman 1966; Sheldon 1968; Gorman and Karr 1978; Finger
1982; Bowlby and Roff 1986) .

Other factors which could change under

reduced flow conditions are predation risk, disease transmission rates,
water temperature, competitive interactions, and food availability (Orth
1987).

The magnitude of the changes in any of these parameters is

almost certainly dependent on the magnitude of the reduction in flow.
It does not appear that risk of predation was increased by the
reduction in summer stream flow in Hunt Creek because fish density in
the treatment section did not decrease.

Also, although I did not

measure disease occurrence during this study, I did not notice any
obvious differences in the occurrence of diseased fish between sections
during the spring and fall electrofishing sampling.

Water temperature

also did not appear to increase in the treatment section of Hunt Creek
due to the reduced flow.

Temperature recorders installed at the

upstream and downstream ends of the treatment section recorded mean

85

daily maximum temperatures for June 1-August 31 in 1993 that were 0.3° C
higher at the downstream end of section B and in 1994 were 0.4° C lower
at the downstream end (Michigan DNR unpublished data).

These

differences could be due to differences in calibration of the recording
devices or may be real differences.

In either case, there is no

evidence for an increase in temperature as a result of the reduced flow.
Finally, brook trout food was not reduced as a result of the reduced
flow because neither benthic invertebrate density or habitat were
impacted by the reduction in flow (Chapter 3).
Factors other than mean summer stream flow may serve to determine
the density of the brook trout population in Hunt Creek.

A large

reduction in fall young

of the year brook trout density in the upstream

section C of Hunt Creek

in 1993 was not evident in the other sections of

Hunt Creek (Figure 25).This reduction in fall

young of the year

density of brook trout may have been due to the intense rain which
caused the failure of a beaver dam just upstream of section C in June,
1993.

I was unable to measure stream gage or discharge during the

spate, however, the flow
during the first week

was over bank

of June when the

were approximately 2-3 cm total length.

fullin section C.

This occurred

youngof the year brook trout
This flood event may have

caused high mortality

or emigration of

C, but not in section

B, because half of the flood flow was diverted

through the diversion channel.

youngof the year fish in section

This high mortality or emigration of

young of the year in 1993 also apparently caused the reduction in
yearling and older density in fall 1994

(Figure 26).

The alternate HSC showed that an investigator's a priori choice of
HSC used in modeling habitat can affect the output of a PHABSIM

86

analysis.

However, the question still remains as to which type(s) of

HSC provide the best prediction of the impacts of a change in flow
regime in a PHABSIM analysis.

The magnitude of the withdrawal in this

study was insufficient to produce an impact on the brook trout.
However, the fact that the shape and magnitude of the WUA-discharge
curves differ indicate that the analysis based on bioenergetic-HSC may
provide a different prediction of

impacts.

The magnitude of

the

decrease in diurnal WUA predicted

at a reduced flow equal to 2% of

baseflow was only 37-70% when diurnal use-HSC were used in the
calculation of WUA but was 75-91%

when bioenergetic-HSC were used to

calculate WUA.

that a 98% reduction in discharge

It is more likely

would reduce the suitable drift foraging habitat area approximately the
same amount.

Therefore, the bioenergetic-HSC may be more accurate

predictors of the changes in foraging microhabitat availability in Hunt
Creek than diurnal use-HSC.
The differences in the magnitude of WUA between the two methods is
likely a result of the more conservative estimates of optimal velocities
based on bioenergetic modeling and the interdependence of the
suitability of depth to velocity.

However, the magnitude of WUA is less

important than the shape of a WUA-discharge curve in attempting to
assess the impacts of a proposed withdrawal on the fish population in a
stream (Bovee 1978).

In that respect, both types of HSC were accurate

predictors of the lack of impact from the reduction of summer stream
flow in Hunt- Creek.

However, this is not an adequate test of the

PHABSIM modeling procedure.

Rather, WUA curves developed from this

study could be used to establish a withdrawal level expected to produce
an impact on the brook trout population in Hunt Creek and the study

87

continued for four more years.

Only then can the predictions of the

PHABSIM model be tested sufficiently.
It is important to stress that the results of this study are
unique to Hunt Creek and are not necessarily applicable to other streams
in Michigan or the Midwest.

The fact that the 50% reduction in summer

streamflow did not reduce fish densities in Hunt Creek is probably
because Hunt Creek is a very stable stream with high quality brook trout
habitat under baseflow conditions.

If Hunt Creek was a marginal trout

stream the 50% reduction in summer baseflow may have resulted in a
reduction in WUA and fish densities.

For example, in an evaluation of

impacts of irrigation withdrawals on the brown trout population in a
marginal trout stream in southern Michigan, the PHABSIM model indicated
that a 50% reduction of summer baseflow would reduce brown trout WUA
approximately 40% (estimated from figures in Gowan 1984).

It is also

likely that a 50% reduction of summer stream flow in Hunt Creek would
have an adverse impact on the trout in Hunt Creek if the population in
Hunt Creek was brown trout or rainbow trout instead on brook trout.

I

modeled the habitat in section B of Hunt Creek with HSC for brown trout
(Gowan 1984; Raleigh et a l . 1986) and rainbow trout (Raleigh et a l .
1984).

The WUA-discharge curves indicated that if brown trout was the

only salmonid present in Hunt Creek, a 50% reduction in summer flow
would reduce adult habitat approximately 8% and would reduce juvenile
habitat 12-16%.

If rainbow trout were the only salmonid species in Hunt

Creek, juvenile habitat would be increased approximately 4% and adult
habitat would be reduced 14-23% with a 50% reduction in summer flow
(Appendix A ) .

88

Although this study could not evaluate the effectiveness of the
bioenergetic-HSC relative to the use-HSC used to calculate WUA estimates
for foraging microhabitats, bioenergetic-HSC offer several potential
advantages.

First, bioenergetic-HSC could be used to construct a

spatial model of foraging habitats in the stream of interest, which
could be used to predict the locations of suitable foraging
microhabitats.
predictions

This information in conjunction with territory size

(Grant and Kramer 1990) could be used to predict the actual

number of fish in a reach of stream and how that number may change with
reduced streamflow.

Also, the use of bioenergetic-HSC with a spatial

model of stream habitat could also be used to predict fish growth rates
(Nielsen 1992) as well as to predict expected changes in growth rates in
relation to changes in flow.

Information on the expected changes in

abundance and growth rates of stream fish in relation to flow could
therefore be used to predict changes in biomass as well.

89

Literature Cited
Alexander, G. R. and E. A. Hansen. 1986. Sand bed load in a brook trout
stream. North American Journal of Fisheries Management 6:9-23.
Armour, C. L. and J. G. Taylor. 1991. Evaluation of the Instream Flow
Incremental Methodology by U.S. Fish and Wildlife Service field
users. Fisheries 16(5):36-43.
Bailey, N. J. J. 1951. On estimating the size of mobile populations from
recapture data. Biometrika 38:293-306.
Bain, M. B., J. T. Finn, and H. E. Booke. 1988. Streamflow regulation
and fish community structure. Ecology 69(2):382-392.
Bovee, K. D. 1978. The incremental method of assessing habitat potential
for coolwater species, with management implications. Transactions
of the American Fisheries Society Special Publication 11:340-346.
Bovee, K. D., T. J. Newcomb, and T. G. Coon. 1994. Relations between
habitat variability and population dynamics of bass in the Huron
River, Michigan. National Biological Survey Biological Report
1994(21). 63 pp.
Bowlby, J. N., and J. C. Roff. 1986. Trout biomass and habitat
relationships in southern Ontario streams. Transactions of the
American Fisheries Society 115 (4) :503 -514.
Chapman, D. W. 1966. Food and space as regulators of salmonid
populations in streams. The American Naturalist 100(913) :345 -357.
Clothier, W. D. 1954. Effect of water reductions on fish movement in
irrigation diversions. Journal of Wildlife Management 18:150-160.
Conder, A. L. and T. C. Annear. 1987. Test of weighted usable area
estimates derived from a PHABSIM model for instream flow studies

90

on trout streams. North American Journal of Fisheries Management
7 :339-350.
Finger, T. R. 1982. Fish community relations in a small central New York
stream. Journal of Freshwater Ecology 1 (4) :343- 352 .
Gorman, O. T., and J. R. Karr. 1978. Habitat structure and stream fish
communities. Ecology 59 (3) :507-515.
Gowan, C. 1984. The impacts of irrigation water withdrawals on brown
trout (Salmo trutta) and two species of benthic macroinvertebrates
in a typical southern Michigan stream. M. S. thesis, Michigan
State University, East Lansing. 127 pp.
Grant, J. W. A., and D. L. Kramer. 1990. Territory size as a predictor
of the upper limit to population density of juvenile salmonids in
streams. Canadian Journal of Fisheries and Aquatic Science
47 :1724-1737.
Kraft, M. E. 1972. Effects of controlled flow reduction on a trout
stream. Journal of the Fisheries Research Board of Canada
29:14055-1411.
Latta, W. C. 1965. Relationship of young-of-the-year trout to mature
trout and groundwater. Transactions of the American Fisheries
Society 94:32-39.
Mathur, D., W. H. Bason, E. J. Purdy, Jr., and C. A. Silver. 1985. A
Critique of the Instream Flow Incremental Methodology. Canadian
Journal of Fisheries and Aquatic Science 42:825-831.
Milhous, R. T . , M. A. Updike, and D. M. Schneider. 1989. Physical
Habitat Simulation System Reference Manual-Version II. Instream
Flow Information Paper No. 26. U.S. Fish and Wildlife Service
Report 89(16) .

91

Morhardt, J. E. 1986. Instream flow methodologies. Research Project
2194-2. Prepared for the Electric Power Research Institute by EA
Engineering, Science and Technology, Inc., Lafayette, CA.
Morhardt, J. E. and C. F. Mesick. 1988. Behavioral carrying capacity as
a possible short term response variable. Hydro Review 7(2):32-40.
Nielsen, J. L. 1992. Microhabitat specific foraging behavior, diet, and
growth of juvenile coho salmon. Transactions of the American
Fisheries Society 121:617-634.
Orth, D. J. 1987. Ecological considerations in the development and
application of instream flow-habitat models. Regulated Rivers:
Research and Management 1:171-181.
Orth, D. J., and O. E. Maughan. 1982. Evaluation of the Incremental
Methodology for recommending instream flows for fishes.
Transactions of the American Fisheries Society 111:413-445.
Raleigh, R. F., T. Hickman, R. C. Solomon, and P. C. Nelson. 1984.
Habitat suitability information: rainbow trout. U.S. Fish and
Wildlife Service FWS/OBS-82/10.60. 64 pp.
Raleigh, R. F., L. D. Zuckerman, and P. C. Nelson. 1986. Habitat
suitability index models and instream flow suitabilitty curves:
brown trout, revised. U.S. Fish and Wildlife Service Biological
Report 82(10.124) .
Reiser, D. W., T. A. Wesche, and C. Estes. 1989. Status of Instream Flow
legislation and practices in North America. Fisheries l4(2):22-29.
Ricker, W. E. 1975. Computation and Interpretation of Biological
Statistics of Fish populations.

Bulletin of the Fisheries

Research Board of Canada 191. 382 pp.

92

Scott, D. and C. S. Shirvell. 1987. A critique of the instream flow
incremental methodology and observations on flow determination in
New Zealand. Pages 27-43 in J. F. Craig and J. R. Kemper, eds.
Regulated Streams Advances in Ecology. Plenum Press, NY.
Sheldon, A. L. 1968. Species diversity and longitudinal succession in
stream fishes. Ecology 49 (2) :193-198.
Shirvell, C., and D. Morantz. 1983. Assessment of the instream flow
incremental methodology for Atlantic salmon in Nova Scotia.
Transactions of the Canadian Electrical Association, Engineering
and Operating Division 22:83-H-l08.
Starrett, W. C. 1951. Some factors affecting the abundance of minnows in
the Des Moines River, Iowa. Ecology 32:13-27.
Stewart-Oaten, A., W. W. Parker, and K. R. Parker. 1986. Environmental
impact assessment: "pseudoreplication" in time? Ecology 67:929940.
Stewart-Oaten, A., J. R. Bence, and C. W. Osenberg. 1992. Assessing
effects of unreplicated perturbations: no simple solutions.
Ecology 73:1396-1404.
White, R. J., E. A. Hansen, and G. R. Alexander. 1976. Relationship of
trout abundance to stream flow in midwestern streams. Pages 597615 in Osborn, J. F., and C. H. Allman, editors, Instream Flow
Needs, Volume 2. American Fisheries Society, Bethesda, MD.
Zar, J. H. 1984. Biostatistical Analysis, 2—

Edition. Prentice-Hall,

Inc. Englewood Cliffs, New Jersey. 718 pp.

93

C H AP T E R

3

COMPARISON OF PREDICTED HABITAT CHANGE AND BENTHIC MACROINVERTEBRATE
RESPONSE TO A SIMULATED IRRIGATION WITHDRAWAL IN HUNT CREEK, MICHIGAN

ABSTRACT
I diverted approximately 50% of the summer flow from a 0.7 km section of
Hunt Creek from June 1-August 31, 1992-94, simulated the impacts of the
withdrawal on the benthic macroinvertebrate habitat in the treatment
section by use of the Physical Habitat Simulation System (PHABSIM), and
compared the changes in habitat with observed densities of benthic
macroinvertebrates in the treated section of Hunt Creek.

The withdrawal

of 50% did not decrease the benthic macroinvertebrate habitat of most of
the taxa examined, but did reduce habitat of riffle dwelling taxa (e.g.
Heptageniidae) by up to 38%.

The total density of benthic

macroinvertebrates in the treatment section of Hunt Creek did not change
as a result of the reduced flow in relation to the total density of
benthic macroinvertebrates in a control section.

However, the densities

of Heptageniidae in a riffle sampled in 1994 did decrease in relation to
a control riffle (p=0.05), indicating that reduced flow may have
resulted in a reduction of Heptageniidae density.

94

Introduction
The Physical Habitat Simulation System (PHABSIM) is the computer
based habitat modeling component of the Instream Flow Incremental
Methodology (IFIM) developed by the U.S. Fish and Wildlife Service that
models stream physical habitat as a function of discharge (Milhous et
al. 1989) .

PHABSIM is widely used to evaluate the impact of altered

flow regimes on stream fish habitat

(Reiser et a l . 1989) with the

assumption that stream physical habitat is directly related to stream
fish population abundance or biomass (Bovee 1978; Orth and Maughan 1982;
Mathur et a l . 1985) .

The PHABSIM is widely used in the western United

States for fish habitat evaluations

(Reiser et al. 1989), but has not

been as widely used for modeling stream benthic macroinvertebrate
habitat.

Also, the PHABSIM system has not been widely applied in the

midwestern United States where the geology, hydrology, and faunal
composition are different from western streams.
In addition to evaluating the effects of reduced streamflows on
fish populations, it is necessary to determine impacts on the benthic
macroinvertebrates because the benthic macroinvertebrates are the
primary source of food for fishes, particularly game species such as the
trouts and charrs (Elliott 1973; Alexander and Gowing 1976; Allan 1981;
Bechara, Morceau and Planas 1992; Nielsen 1992).

In many trout streams

the density of benthic macroinvertebrates and the occurrence of
macroinverte.brates in the drift may limit the growth rate of individual
fish or may limit the population size (Chapman 1966) .

Orth (1987) also

argued for consideration of ecological factors other than space occupied
by fish alone when evaluating the impacts of altered streamflows, yet

95

most PHABSIM analyses of altered streamflows are centered on the changes
in game fish habitat.

I am only aware of one published study which

applied the PHABSIM system to benthic macroinvertebrates.

Bovee (1985)

studied the impacts of a peaking hydropower operation on the benthic
macroinvertebrate habitat in a Colorado stream, but no comparison was
made between the output of the PHABSIM modeling and observed benthic
macroinvertebrate abundance.

Gowan (1984) also modeled the habitat of

two genera of macroinvertebrates in a marginal trout stream in Michigan
and determined that withdrawals for irrigation reduced habitat by up to
11%, but that fish habitat losses were more severe.

However, he did not

compare PHABSIM model output with observed benthic macroinvertebrate
densities.
Several studies have documented benthic macroinvertebrate habitat
use patterns and have published habitat suitability criteria

(HSC) which

could be used in a PHABSIM analysis (Gore and Judy 1981; Orth and
Maughan 1983; Gore 1989).

However, the availability of these data have

not led to an increase in the use of benthic macroinvertebrates in
predicting impacts of flow regulation in streams.
Very little is known about the relationships between natural
streamflow patterns and benthic macroinvertebrate communities and
abundance.

Ward (1976) reviewed the impacts of regulated streamflows on

the benthos below large dams and noted that streamflow regulation can
result in changes in community composition and enhancement or reduction
of standing crop, depending on flow regime.

Others (Minshall and Winger

1968; Corrarino and Brusven 1983; Poff and Ward 1991) have documented
the impacts of altered streamflows on invertebrate drift, noting that
altered streamflows can result in increased drift rates.

96

Current

velocity (Rabeni and Minshall 1977; Orth and Maughan 1983; Degani et a l .
1993), depth (Degani et a l . 1993) and substrate composition

(Minshall

1984) are important factors influencing the distribution and abundance
of benthic macroinvertebrates.

Of these three factors, depth and

velocity will change as streamflow is altered.

Also, substrates may

become embedded with fines if velocity is sufficiently reduced.
Therefore, altered streamflows can be expected to have an influence on
benthic macroinvertebrate communities.
Here I present the results of a PHABSIM analysis of benthic
macroinvertebrate habitat in a Michigan stream during a simulated
irrigation withdrawal. The objectives were to determine the impacts of a
simulated irrigation withdrawal on the benthic macroinvertebrates in
Hunt Creek, and to evaluate the PHABSIM model as a quantitative
predictor of the changes in the benthic macroinvertebrate assemblage
resulting from the altered streamflow.
Methods
I examined macroinvertebrate habitat and populations in two
sections of Hunt Creek: one nontreatment section (section C) and a
treatment section (section B; Figure 1, Chapter 1).

Prior to sampling

macroinvertebrates and macroinvertebrate habitat I measured and marked
sections B and C (Figure 1, Chapter 1) into approximately 50 m
contiguous reaches, and omitted the small area of impounded water at the
downstream end of sections B and C and the disturbed habitat at the
upstream end of section B immediately below the bulkhead.

I was left

with four 50 m reaches in section C and seven reaches in section B.
I sampled benthic macroinvertebrates from randomly selected
locations in sections B and C during May-September, 1992 and April-

97

September,

1993 to construct habitat suitability criteria (HSC).

I

moved upstream through the sections and collected 20 samples each month
in each section from randomly selected locations.

I sampled

macroinvertebrates using a modified Hess sampler or a petite Ponar grab
in water that was either too deep or too shallow to sample with the Hess
sampler.

The net on the Hess sampler was constructed of 500 pm mesh and
2

the area sampled by the Hess sampler was 0.023 m .
2

sampled 0.026 m .
alcohol.

The Ponar grab

I preserved benthic samples in the field in 95% ethyl

At each sample location, I also measured depth to the nearest

cm, water velocity to the nearest cm's"1, and visually estimated
dominant substrate composition using the codes in Table 2, Chapter l.
measured water velocity with either a mechanical pygmy Gurley or an
electronic Marsh-McBirney current meter.

There was no difference in

measurements between the meters (Chapter 2).
I separated macroinvertebrates from inorganic material in the
samples by floating the sample contents in a saturated sugar solution
(Anderson 1959).

I also thoroughly sorted a second time through 20

samples from a variety of habitats (depositional, erosional) after they
were sorted once to evaluate the effectiveness of the initial sorting.
I identified macroinvertebrates to family using the keys in
Merritt and Cummins (1984) and Pennak (1989) .

I counted the number of

organisms by family in the samples and converted these to density
estimates by dividing by the sampled area.
I constructed HSC from the combined data collected in 1992-93 in
section B using the nonparametric tolerance limits method (Bovee 1986).
HSC were constructed from habitat use data by use of the formula:
NSI=2(1-P),

98

(10)

I

where NSI is the normalized suitability index and P is the central
proportion of the data frequency distribution (Bovee 1986).

I

constructed the HSC for depth and mean column velocity with this
approach by defining P as the central 50, 75, 90, and 95% portions of
the distribution by use of the nonparametric tolerance limits table
found in Somerville (1958) and a confidence level of 95%.

I constructed

the HSC for substrate by normalizing the frequency-of-use data for each
of the substrate categories.

I normalized the substrate data by

dividing the frequency-of-use data in each category by the frequency for
the most commonly used substrate (Bovee 1986).

I combined the substrate

data by largest particle size and ignored the percent embeddeded
classification because sample sizes were small for some of the substrate
categories.

Because I was concerned about statistical independence of

the observations, I did not weight the value of the habitat measured at
a sample location by the number of organisms in the sample.
resulted in HSC constructed from presence-absence data only.

This
I also did

not correct the HSC based on habitat availability because I randomly
selected sample sites and therefore sampled all habitat types in
approximate proportion to their availability.
I used the same reaches in section B to model macroinvertebrate
habitat that were used to model brook trout habitat
1).

(Figure 1, Chapter

I simulated benthic macroinvertebrate habitat for selected families

over a range of flows from baseflow to 2% of baseflow for the reaches in
section B.

I selected families based on their frequency of occurrence

in the 1992-93 samples and based on habitat use characteristics.

I

modeled habitat for macroinvertebrate families that occurred in 20-80%
of the samples in 1992-93 and selected additional families to provide

99

data for habitat types that were not sufficiently represented in the
initial selection procedure.
Because early results of the PHABSIM modeling indicated that
riffle dwelling macroinvertebrates

(e.g. Hydropsychidae) were more

likely to be' impacted by the reduction in flow than macroinvertebrates
found more commonly in pool or depositional habitats, I altered the
macroinvertebrate sampling design during the final year of the study.
Prior to the withdrawal period in 1994 I selected two riffles, one in
section C and one in section B, that had similar microhabitat
characteristics under baseflow conditions.

The riffle selected in

section B was approximately 25 m upstream of the upper end of modeled
reach B 2 .

The depths, mean column velocities, and substrate

characteristics of the selected riffle were very similar to those found
in the riffle habitats in reach B 2 .

I measured the width and length of

each of the selected riffles and established a two dimensional grid of
cells using permanent markers in the stream bank.

The cells were

approximately the same size as the area sampled by the Hess sampler.

I

then collected benthic samples in seven randomly selected cells in each
riffle at three week intervals from May 12-August 23.

I used the same

sampling protocol as for the 1992-93 samples.
I formulated hypotheses concerning the impacts of the reduced
streamflow on the density of benthic macroinvertebrate families in
section B of Hunt Creek in 1992 and 1993 based on the relation between
WUA and discharge curves and the assumption that WUA is positively and
linearly related to macroinvertebrate abundance (Bovee 1978; Orth and
Maughan 1982; Mathur et a l . 1985).

I used profile analysis (Morrison

1990) to compare the total benthic macroinvertebrate density trends

100

between sections B and C during the summer for the 1992 data and used
Before After Control Impact

(BACI) analysis (Stewart-Oaten et al. 1986,

1992) to evaluate the predictions of the PHABSIM model for the 1993
data.

For the BACI analysis of the 1993 and 1994 macroinvertebrate

data, I determined the mean pretreatment difference between sections B
and C from the benthic samples collected prior to 1 June and compared
that mean difference to the mean difference between the sections from
the samples collected after 1 June.

Because the riffle sampled in 1994

in section B was close to reach B2 and was similar to the riffles in
reach B 2 , I evaluated the relation between WUA and discharge for riffle
transects in reach B2 to formulate hypotheses concerning the impacts of
the withdrawal on the macroinvertebrate assemblage in the sampled
riffle.

I evaluated hypotheses of withdrawal impacts on the total

macroinvertebrate density and on the densities of the most abundant
families in 1994 by use of BACI statistics by again comparing the
pretreatment mean difference to the treatment mean difference.
I also evaluated the impacts of reduced flow on the
macroinvertebrate drift density in Hunt Creek.

I collected

macroinvertebrate drift data across a fixed transect in the treatment
and the control sections of Hunt Creek.

I sampled the drift by use of

rectangular drift nets with a mouth opening 75 cm x 15.5 cm and 80 cm
deep.

Drift nets were constructed from 64 pm mesh nylon netting.

Three

nets were set across the transect which bisected a riffle, and the nets
sampled the entire water column.

I sampled the invertebrate drift for

20 minutes at four-hour intervals over 24 hours approximately every four
weeks during summer, 1993 in section C and sampled the drift in section
B on approximately the same dates but for up to 48 hours, depending on

101

the withdrawal schedule.

I measured depth and velocity at each net

location immediately prior to sampling and used these data to estimate
the volume sampled in the 20 minute period.

I sorted the drift samples

in a saturated sugar solution immediately after collection and preserved
macroinvertebrates in 95% ethyl alcohol as they were retrieved from the
samples.

I thoroughly sorted through 15 samples a second time to

determine the effectiveness of the procedure.
of invertebrates in the drift per m

3

I calculated total number

for each sample and determined the

mean number of invertebrates per m^ for each sample date and time.

I

used BACI analysis to evaluate the impacts of the withdrawal on the
macroinvertebrate drift densities by comparing pretreatment mean
difference in drift density between section B and C for the 24 hour
period prior to the withdrawal on June 2 with the treatment period mean
differences for the remaining sample dates.
Results
Benthic Macroinvertebrate Assemblage
I collected 199 benthic samples in 1992 and 237 benthic samples
during 1993.

I sorted 26 benthic samples a second time after sugar

floating and determined I was retrieving 88% (range 69-100%) of the
macroinvertebrates from the samples with the sugar floating procedure.
Because I retrieved nearly all invertebrates from the samples I did not
adjust the sample data by the efficiency of the initial sorting
procedure.

The benthic samples contained macroinvertebrates from 45

families representing 13 orders (Table 6).

I collected 83 benthic

samples in 1994, 42 from the riffle in section B and 41 from the riffle
in section C.

The 1994 samples contained macroinvertebrates from 29

families representing 11 orders (Table 6).

102

Oligochaetes, baetids,

Table 6

Benthic macroinvertebrate taxa collected and percent frequency of occurrence in benthic
samples in sections B and C of Hunt Creek, 1992-94

Taxa selected for habitat modeling

are in bold face type.

Class

Order

Insecta

Trichoptera

Plecoptera

Odonata

% Occurrence in

% Occurrence in 1994

1992-93 (n=436)

(n=83)

Baetidae

72.5

97.6

Ephemerellidae

37.6

53.0

Ephemeridae

5.1

Heptageniidae

38.8

83.1

Leptophlebiidae

1.6

2.4

Tricorythidae

0.2

Family

Hydropsychidae

31 9

47.0

Glossosomatidae

67.7

90 4

Limnephilidae

23.6

33.7

Rhyacophilidae

36.2

53.0

Philopotamidae

12.6

20.5

Brachycentridae

15.8

14.5

Lepidostomatidae

3.2

Hydroptilidae

1.6

Psychomyiidae

0.2

4.8

Pertodidae

23.2

32.5

Nemouridae

30.3

69.9

Taenyopterygidae

0.2

Leuctridae

6.0

Perlidae

0.2

Cordulegasteridae

9.2

Gomphidae

1.4

Aeshnidae

0.2

Calopterygidae

0.7

103

26.5

1.2

Ta bl e

6

(c o n t 1d)

Coleoptera

Diptera

Elmidae larvae

81 9

98.8

Elmidae adults

33.5

84 3

Dytiscidae (larvae)

0.2

Chironomidae

90.6

86.7

Tipulidae

16.3

3.6

Sim uliidae

34.9

51.8

Empididae

50.7

68.7

Ceratopogonidae

25.0

6.0

Tabanidae

9.2

2.4

Athericidae

3.7

10.8

Ptychopteridae

0.7

Muscidae

0.2

Dixidae

0.9

Stratomyiidae

0.2

Corydalidae

7.1

Sialidae

1.8

Lepidoptera

Pyralidae

0.2

Arachnida

Acari

Hydracarina

1.4

Bivalvia

Pelecypoda

Sphaeriidae

0.9

Malacostraca

Amphipoda

Gammaridae

91.7

97.6

Isopoda

Asselidae

6.7

14.5

Oligochaeta

Undetermined

Undetermined

100.0

100.0

Hirudinea

Undetermined

Undetermined

Megaloptera

1.2

10.8

15.7

1.2

104

gammarids, elmid larvae and chironomids were the most frequently
occurring taxa in 1992-93, when samples were collected at random
locations throughout sections B and C.

Oligochaetes were the most

frequently collected taxa in the 1994 samples in sections B and C.
Other frequently occurring taxa in 1994 were baetids, heptageniids,
glossosomatids, chironomids, elmid larvae, and gammarids

(Table 6).

I selected 13 families of macroinvertebrates for habitat modeling
using PHABSIM (Table 6).
frequency of occurrence,

In addition to the families selected based on
I selected the Tipulidae (present in 16.3% of

the 1992-93 samples) to provide an additional taxon that occurs
primarily in pool or depositional habitats.

I also excluded the

Limnephilidae because the genera of limnephilids found in Hunt Creek
included both scrapers (Goera sp.) and shredders-herbivores (Limnephilus
sp.) which occupy different habitats (Merritt and Cummins 1984).

The

majority of the taxa selected were found primarily in riffles (Table 6)
and the remaining taxa were habitat generalists found primarily in runs
and depositional habitats but also in riffles.
Habitat suitability criteria (HSC) for depth were very similar for
all the taxa irrespective of the habitat type selected most frequently
(Table 7).

The narrowest optimal (suitability=l.0) depth range was from

16-31 cm for the Hydropsychidae and the widest optimal depth range was
from 13-33 c.m for the Ceratopogonidae.

The range of usable

(0.0<suitability<l.0) depths was also very similar between taxa and was
also similar to the range of depths sampled in 1992-93.

The minimum

usable depth was 2 cm (Ceratopogonidae) and the maximum usable depth was
61 cm (6 taxa. Table 7).

The depth HSC did not differ between taxa that

primarily occupied riffles in comparison to taxa that primarily occupied

105

Table 7.

Mean column velocity (cm's 1) and depth (cm) habitat
suitability ranges for the 13 families of benthic
macroinvertebrates selected for habitat modeling in Hunt
Creek.

Column headings (1, 0.5, 0.2, 0.1, and usable) are

suitability values.

Mean Velocity (cm— s— )
1

0.5

0.2

0.1

Usabli

Baet idae

21-56

9-64

0-73

0-78

0-80

Ephemerel1idae

21-55

12-63

1-70

0-74

0-79

Heptageni idae

42-65

27-73

6-78

4-79

0-80

Hydropsychidae

34-61

23-69

15-77

7-78

4-79

Glossosomatidae

30-58

16-65

10-74

2-79

0-80

Rhyacophi1idae

34-63

27-70

12-78

0-79

0-80

Perlodidae

32-63

22-73

15-79

4-80

3-81

Nemouridae

27-58

10-66

2-73

0-79

0-80

Elmidae Adult

34-60

27-70

15-78

11-79

0-80

Tipulidae

24-58

5-64

0-69

0-70

0-71

Simuliidae

28-59

21-73

14-78

14-79

4-80

Empididae

27-57

14-65

8-77

0-79

0-81

Ceratopogonidae

4-43

0-54

0-73

0-77

0-78

Taxa

DeDth

(cm)

Baetidae

17-36

12-46

8-58

5-61

4-63

Ephemerellidae

17-37

11-47

5-58

4-58

2-63

Heptageni idae

20-38

13-49

11-58

6-61

4-63

Hydropsychidae

16-31

11-42

6-49

4-55

4-58

Glossosomat idae

17-37

12-47

6-55

4-61

4-63

106

Ta bl e 7

(c o n t 'd )

Rhyacophi1idae

17-36

13-44

11-50

7-56

6-61

Perlodidae

13-38

7-43

5-47

4-61

3-62

Nemouridae

13-31

8-40

5-47

4-58

3-59

Elmidae Adult

17-37

13-46

7-58

5-61

3-64

Tipulidae

15-35

11-43

9-49

6-56

5-57

Simuliidae

13-30

11-46

6-58

4-58

3-59

Empididae

16-37

12-48

8-56

4-61

2-64

Ceratopogonidae

13-33

10-48

3-52

2-56

1-57

107

pool or depositional habitats

(Figure 28).

For example, the optimal

depth range for Hydropsychidae (riffles) was from 16-31 cm and was from
15-35 cm for Tipulidae and 13-33 cm for Ceratopogonidae, both found
primarily in depositional areas (Table 7).
Although there was considerable overlap in usable mean column
velocities between the taxa, mean column velocity HSC were less similar
between taxa than depth HSC and were better indicators of the habitat
types selected by the taxa (Table 7).

For example, the optimal velocity

range for Hydropsychidae was 34-61 cm's 1 and 24-58 cm's 1 for Tipulidae
(Figure 29).

Usable velocity ranges also overlapped considerably for

the taxa selected for habitat modeling.
between 71-81 cm's"1 .

The maximum usable velocity was

The minimum usable velocity was 0 cm's"1 for most

taxa but was as high as 4 cm's 1 for the Simuliidae and the
Hydropsychidae, both filter feeding taxa (Merritt and Cummins 1984).
Substrate HSC also reflected the habitat selectivity of the taxa
(Table 8, Figure 30).

The optimal substrate size (medium gravel) was

the same for all the taxa selected for modeling.

However, this

undoubtedly reflects the fact that medium gravel was the most frequently
sampled substrate during 1992-93.

Large gravel was the most suitable

substrate after medium gravel for all taxa except Ceratopogonidae (Table
8).

Sand and silt in combination and sand alone were the second most

suitable substrates for the ceratopogonids.
PHABSIM Model Results
The 50% reduction in summer stream flow in the treatment section
resulted in a very minor loss of stream surface area in both modeled
reaches (Figure 19, Chapter 2).

In reach B2 total surface area

decreased from 206 to 195 m 2 '100 m -1 , a reduction of only 5.6%.

108

In

Table 8.

Substrate suitability values for the 13 families of benthic
macroinvertebrates selected for habitat modeling in section
B of Hunt Creek. Percent embeddedness of the substrate was
not included in the substrate suitability calculations.

Substrate Code
1 (sand

3 (small

4 (medium

5 (large

and slit)

2 (sand)

gravel)

gravel)

gravel)

Baetidae

0 .13

0 .13

0.17

1

0 .70

Ephemerel1idae

0 .11

0.11

0.2

1

0.65

Heptageni idae

0 .02

0

0 .02

1

0 .83

Hydropsychidae

0

0 .11

0.11

1

0.80

Glossosomat idae

0

0 .04

0.13

1

0 .55

0 .02

0 .08

0 .06

1

0 .67

Perlodidae

0

0 .04

0 .04

1

0 .88

Nemouridae

0.08

0 .24

0 .20

1

0 .76

0

0

0 .07

1

0 .65

Tipulidae

0 .23

0 .14

0.09

1

0 .82

Simuli idae

0.04

0 .12

0 .27

1

0 .96

Empididae

0.08

0 .12

0.17

1

0.61

Ceratopogonidae

0 .69

0.41

0. 14

1

0.38

Taxa

Rhyacoph i1idae

Elmidae AduLt

109

0.9 4
i

Heptageniidae
Tipulidae

0.8 -I

I
0.7 -t
0.6 4

0.5
0.4
0.3
0.2

■*

I

0

0

10

20

30

40

50

60

70

80

Depth (cm)

Figure 28.

Depth habitat suitability criteria for Heptageniidae and
Tipulidae calculated from habitat use data collected in
section B of Hunt Creek, summer 1992-93.

110

Heptageniidae
Tipulidae

Suitability

0.8 1

0.6

-

0.2

0

10

20

30

40

50

60

70

80

90

Mean Column Velocity (cnvs*1)

Figure 29.

Mean column velocity habitat suitability criteria for
Heptageniidae and Tipulidae calculated from habitat use data
collected in section B of Hunt Creek., summer, 1992-93.

ill

□ Heptageniidae

Suitability

Tipulidae

sand and
silt

sand

small
gravel

medium
gravel

large
gravel

Substrate

Figure 30.

Substrate suitability criteria for Heptageniidae and
Tipulidae calculated from habitat use data collected in
section B of Hunt Creek, summer, 1992-93.

112

reach B4 , total surface area decreased from 292 to 275 m 2 '100 m -1 , a
reduction of only 5.7%.

Model results indicated that reducing flow in

section B to a discharge of 0.01 m^ ‘s 1 would reduce total surface area
to 132 m 2 ‘100 m -1 m

reach B2 (35.9% loss in surface area) and 176

m 2 '100 m 1 i'n reach B4 (39.7% loss of surface area) .
-

This result

suggests that considerable pool habitat would remain in section B if
discharge was reduced to 0.01 m 3 's -1 .

The difference in the total

surface area estimates between the two modeled reaches is due to
differences in channel width.
The reduction of summer stream flow by 50% increased WUA for all
but three of the taxa in reach B2 (Heptageniidae, Rhyacophi1idae, and
Elmidae adults, Table 9).

The increases in WUA ranged from 1% to 22%

(Perlodidae and Tipulidae, respectively).

In general, the taxa that

were most commonly found in pool and depositional habitats had the
largest increases in WUA.

The Heptageniidae, Rhyacophi1idae, and

Elmidae adults all had reduced WUA as a result of the 50% reduction in
summer stream flow, although the losses were minor (4-15%) .

In

contrast, WUA decreased in reach B4 for all taxa except Ceratopogonidae
when summer stream flow was reduced 50%.

The WUA estimate at 50%

reduced flow for ceratopogonids remained virtually unchanged (from 279.8
2

m ‘100 m

-

1

2

to 280.3 m '100 m

-

1

) (Table 9).

For the remaining taxa, WUA

decreased between 9% and 38% (Nemouridae and Heptageniidae,
respectively).

The WUA estimates for taxa most commonly found in pool

and depositional habitats decreased less than for riffle dwelling taxa
(e.g. Figure 31).
Reducing flow to less than 50% of baseflow during the summer
months would result in reduced WUA for all taxa modeled (Table 9).

113

Table 9.

2

Weighted Usable Area (WUA, m '100 m
discharge

for t he

-

1

) in relation to

13 b e n t h i c m a c r o i n v e r t e b r a t e

fam ili es

selected for habitat modeling in section B of Hunt Creek.

Discharcre (m— 8— 1
Reach B2

0.01

0 .04

0.07

0.11

0.17

0.23

0.46

Baetidae

24 .8

60.0

91.3

123 .1

160.7

180.5

160.0

Ephemerellidae

26 .2

64 .5

96 .9

127 .8

162 .8

180.6

155.8

Heptageni idae

6 .7

20.8

35.9

59 .1

99 .2

130.7

153.4

Hydropsychidae

5.2

24 .7

50.6

85.7

130.9

155.1

147.8

Glossosomatidae

11 .0

37 .4

66.2

97 .8

138 .1

161 .2

148.0

Rhyacophi1idae

7.4

17 .8

33.7

62 .3

109 .1

140.0

146.5

Perlodidae

9 .1

36 .2

68.6

104 .8

145 .3

166 .6

165.4

Nemouridae

29.8

77 .0

115 .1

148 .4

180.2

193.1

164 .7

3.4

16 .6

35.5

65.0

111.0

14 1 .3

147.3

Tipulidae

30.8

66.0

99.7

135.8

173 .8

192.2

157.4

Simuliidae

4 .7

29 .6

65.7

108 .7

153 .0

175.8

169.2

Empididae

13 .9

42.1

72 .8

107 .4

148 .7

171 .0

156.6

Ceratopogonidae

88 .1

132.8

163.1

188 .9

212 .2

217.3

180.3

Baetidae

28.5

67 .5

108 .1

156 .7

197.2

216.7

249 .4

Ephemerellidae

30 .6

67 .9

108 .2

157 .5

201 .1

219.8

247 .0

Heptageniidae

6.5

22 .4

40.5

63.8

99.5

131.9

212 .8

Hydropsychidae

3.9

21. 5

43 .9

75.5

131. 6

166.6

225.7

Glossosomatidae

11 .8

36 .4

66 .9

108 .8

162 .1

190. 8

234 .1

Rhyacophi1idae

8.4

19.4

35.7

62 .0

108 .8

142 .6

220 .3

Perlodidae

8.5

34 .2

58 .8

91.1

153 .9

189.2

242.3

Elmidae Adult

Reach B4

114

T ab le

9

(c o n t 'd)

Nemouridae

34 .5

86 .0

125 .2

163.7

208.6

230.0

252 .2

Elmidae Adult

3.2

14 .6

30.6

57 .1

109.9

145.3

220.2

Tipulidae

33.0

80.2

123 .6

166 .1

201.3

220.8

254 .1

Simuli idae

3 .2

23 .5

54 .3

93 .8

156 .6

194 .6

241.7

Empididae

15 .0

42.8

79 .1

125.3

174.7

202 .3

244 .4

109 .1

171 .8

217 .1

249.0

265.6

280.3

279 .8

Ceratopogonidae

115

Also, the model predicted that reducing flow to 2% of baseflow (0.01
's ■*■) would reduce WUA between 51-96% in reach E2 (Ceratopogonidae
larvae and Elmidae adults respectively) and 61-99% in reach B4
Ceratopogonidae and Simuliidae larvae/Elmidae adults respectively).
Based on the WUA curves for reaches B2 and B4 I hypothesized that
the 50% reduction in flow should not have resulted in a decrease in
total benthic macroinvertebrate density in section B in 1992 or 1993.
The profile analysis of the 1992 density data support this hypothesis
because the test indicated the trends in total macroinvertebrate
densities

(Table 10, Figure 32) were not different between the sections

(F=0.3 0, df=4,194, p<0.001) .

Similarly, the results of the BACI

analysis of the 1993 data (Table 10, Figure 33) support the conclusion
that the withdrawal did not reduce total macroinvertebrate density in
section B (Student's t test, p=0.84).
The relation between WUA and discharge for the riffle transects in
reach B2 were virtually unchanged by excluding the run and pool
habitats, although the magnitude of the WUA estimates at any one
discharge did decrease.

Therefore, I hypothesized that the 50%

reduction in summer stream flow would have no impact on the abundance of
any of the benthic macroinvertebrate taxa in the sampled riffle in
section B during summer, 1994.

I tested these hypotheses on the taxa

for which I modeled habitat and also on the Gammaridae, Chironomidae,
and Elmidae larvae.

The 50% reduction in summer stream flow did not

impact the abundance of the families tested nor did it impact the total
macroinvertebrate abundance estimates in section B (Table 11).

Total

macroinvertebrate density in section B did not decline as a result of
the withdrawal and the trend in total density through the summer closely

116

Table

10.

Benthic m a c r o i n v e r t e b r a t e d e n s it y estimates
parentheses)

Year

Section

Baeti da e

1992

He pt ag e n i i d a e

117
H yd ro ps yc hi da e

G lo ss os om at idae

R h y a c o p h i 1idae

Perlodidae

_2

, standard error estimates

in

from r andom sample locations in sections B a nd C for 1992-93.

Family

Ephe me re ll id ae

(number'm

April

May

June

July

August

September

B

75 (21)

110(30)

273 (87)

321(106)

18(6)

C

136 (41)

641 (106)

427 (85)

318 (112)

185 (68)

B

138(38)

112 (27)

104 (24)

9(6)

2(2)

C

91(37)

40 (17)

47(13)

5(3)

11(5)

B

236(56)

11(8)

4(3)

9(4)

90(24)

C

460 (109)

65 (23)

14(8)

20 (20)

91 (37)

B

61(33)

24(11)

258 (93)

446 (222)

256 (145)

C

11(5)

9(5)

12(7)

241 (106)

510(307)

B

1042 (253)

661(190)

927(334)

410(141)

845(156)

C

173(57)

234 (63)

366(140)

310(90)

1108(265)

B

37 (12)

2 (2)

27 (9)

1 1 1 (35)

198(55)

C

33(10)

31(21)

23(10)

31(14)

107(30)

B

4 (3)

11(5)

18(12)

38 (14)

62 (18)

Table 10

(cont'd).

Ne m o u r i d a e

El mi da e adults

Tipulidae

S i m u li id ae

118

Em pi di da e

C er at op o g o n i d a e

Total

Ba e t i d a e

1993

C

5(3)

29(10)

42(15)

36(15)

111 (28)

B

44(23)

78(59)

62(30)

13 (7)

0 (0 )

C

20(8)

698(317)

206(104)

20 (9)

2 (2 )

B

94(41)

35(12)

36(20)

58 (20)

76 (29)

C

0(0)

0 (0 )

0 (0 )

0 (0 )

0 (0 )

B

18(6)

13(6)

11(4)

36(10)

13(5)

C

7(5)

2 (2 )

14 (7)

2 (2 )

13 (6)

B

13(7)

13(8)

322(202)

125 (58)

4 (3)

C

57(29)

546 (298)

330(157)

138(106)

120 (82)

B

24(8)

52(18)

79(21)

20 (9)

22 (7)

C

2(2)

80(25)

66 (24)

38(10)

18 (8)

B

20(14)

26(10)

37(14)

49(16)

44(11)

C

13(9)

2 2 (1 0 )

19(9)

29(17)

29(16)

B

4031(647)

5569(771)

8301 (1256)

6109 (940)

4148 (470)

C

2309(315)

6078 (1050)

8865 (1195)

6451(1001)

5443 (963)

B

277(58)

114 (25)

110 (29)

244 (33)

392(71)

130 126)

C

251(82)

78 (19)

180(72)

592(125)

435(117)

300 (70)

Table 10

(cont'd).

E ph e m e r e l l i d a e

Heptageniidae

Hydropsychidae

Glossosomatidae

R h y a c o p h i 1 idae

119
Per lo di da e

Nemouridae

El mi da e adults

T ip ul i d a e

B

317(81)

168(38)

191(39)

175(36)

21 (12)

13 (9)

C

49(16)

26(13)

26(15)

0 (0 )

2 (2 )

0 (0 )

B

239(134)

275(93)

38(18)

7(5)

12(8)

141(51)

C

378(93)

321(86)

236(81)

81(28)

40 (18)

18 (9)

B

146(81)

61(26)

38(18)

404(120)

442(202)

280(168)

C

45(24)

5(3)

0 (0 )

14(8)

25(16)

58 (24)

B

306 (72)

248 (71)

249(75)

245(79)

222(56)

630(182)

C

448(159)

109(36)

28(13)

236(77)

305(102)

117 (48)

B

131(38)

1 1 (6 )

9(7)

2 0 (1 0 )

59(18)

150(32)

C

36(12)

12(7)

21(9)

14(7)

29(12)

36(13)

B

11(6)

5(3)

5(5)

7 (4)

7(4)

49(16)

C

7(4)

2 (2 )

5(5)

2 (2 )

11(4)

16 (8)

B

70 (40)

90(39)

40(23)

54(25)

5(5)

0 (0 )

C

61(26)

31(14)

99(57)

36(13)

16(5)

2 (2 )

B

33(9)

90(27)

74(25)

52(24)

116(39)

146(42)

C

40 (20)

59(21)

45(19)

31 (12)

29(9)

119(54)

B

18 (7)

14 (7)

7(4)

9(5)

7(5)

0 (0 )

Table 10

(cont'd).

S imu li id ae

Empididae

C e r a t o p o g o n idae

Total

C

11(7)

0 (0 )

2 (2 )

5(3)

25 (22)

0 (0 )

B

16(10)

491(427)

5(3)

168(60)

113(52)

11 (4)

C

20(11)

83 (47)

5(3)

139(58)

94(39)

38(12)

B

’47(11)

105(30)

186(55)

197(57)

90(30)

61(21)

C

85(22)

59 (21)

66(29)

96(29)

88(33)

108(29)

B

45(29)

38(18)

29(11)

49(23)

38(13)

52(20)

C

2(2)

52(50)

2 (2 )

5(3)

9(9)

7(5)

B

6236 (1043)

4509(655)

5308(713)

9053(1131)

6312 (751)

4480 (727)

C

4241(783)

2403 (346)

2984 (737)

4036 (690)

4646 (751)

4030(918)

T a b le

11.

Benthic m a c r o i n v e r t e b r a t e density estimates
parentheses)

(number'm

_2

from one riffle each in sections B and C,

, standard error estimates

in

1994.

Date
Taxa
B ae ti da e

E ph e m e r e l l i d a e

121

Heptageniidae

Hydropsychidae

Glossosomatidae

Rhyacophilidae

Section

Hay 12

June 1

June 20

July 12

August 3

August 23

B

448(83)

91(33)

149(42)

1143(218)

2331 (576)

1052(210)

C

759(142)

442(92)

545(74)

992(187)

2182(389)

935 (215)

B

643(55)

350(78)

253(33)

117(41)

39(21)

19(14)

C

58(16)

26(13)

13(8)

0(0)

13(13)

0(0)

B

1149(136)

851(106)

273(49)

110(31)

39(15)

26 (13)

C

649(106)

539(83)

487(62)

326 (84)

130(30)

19(13)

B

195(55)

84(50)

208(67)

32(13)

636(148)

740(180)

C

6(6)

6 (6)

0(0)

0(0)

13 (13)

26 (13)

B

1818 (483)

1117(269)

1442(263)

234 (73)

312(38)

130(47)

C

571(141)

253(31)

221(53)

76(22)

65(17)

84(40)

B

52 (25)

13(8)

52(38)

13(13)

136(37)

52(15)

C

39(15)

39(6)

52(15)

38(18)

19(9)

19(9)

Table 11

(cont'd).

Perlodidae

Nemouridae

E l m i d a e adults

Tipulidae

Simuliidae

Empididae

Ceratopogonidae

Total

B

6(6)

0(0)

26(9)

0(0)

97 (48)

39 (12)

C

6(6)

6(6)

19(9)

23 (16)

45 (24)

6(6)

B

162(47)

104 (32)

435 (90)

45 (24)

97 (44)

26(13)

C

117(61)

162(78)

539(166)

159(66)

104 (45)

19(14)

B

377 (80)

214 (56)

273(44)

149(52)

299(62)

390(66)

C

78(32)

162(65)

182(65)

106(46)

84(27)

84(51)

B

0(0)

0(0)

0(0)

0(0)

19(14)

6(6)

C

0(0)

0(0)

0(0)

0(0)

0(0)

0(0)

B

52 (27)

32(22)

6(6)

104 (38)

156(119)

32 (26)

C

468(250)

156(80)

26(26)

60(19)

149(72)

65(40)

B

136(49)

130(55)

208(70)

117(28)

65 (34)

13 (8)

C

149(43)

136(43)

227(95)

98(36)

97(34)

32(19)

B

13(13)

0(0)

6(6)

0(0)

13(8)

6(6)

C

0(0)

0(0)

0(0)

0(0)

0(0)

0(0)

B

7188(485)

4188(713)

6195(892)

6948 (727)

7591(1031)

5409(726)

C

4844 (441)

2981(366)

4500(805)

3614 (524)

6325(805)

4149 (1262)

300

B2 Heptageniidae
B2 Tipulidae
B4 Heptageniidae

250

WUA (m2100 m'1

B4 Tipulidae
200

150

100

50

1/2 baseflow

baseflow

0
0

0.1

0.3

0.2

0.4

0.5

Discharge (m 3s‘1)

Figure

31.

Weighted

usable

Heptageniidae

area

(m

2

"100 m

and Tipulidae

123

-1

) at

discharge

in s e c t i o n

B of

(m

Hunt

3 - 1
’s
) for
Creek.

12000

Macroinvertebratesm

10000

8000

6000

4000

Section B
Section C

2000

- i
Apr

May

Jun

Jul

Aug

Sep

Month

.

Figure

32.

Total

macroinvertebrate

standard
Creek,

error

bars)

summer,

1 9 9 2.

density

in s e c t i o n s

124

(number'm
B and

o

C of

, with
Hunt

12000

Macroinvertebrates m

10000

8000

6000

4000

Section B
Section C

2000

—

Mar

Apr

May

Jun

■ -•-( -------

■

Jul

Aug

Sep

Month

_

Figure

33.

Total

macroinvertebrate

standard
Creek,

error

bars)

summer,

1993.

density

in s e c t i o n s

125

(number'm
B and

2

C of

with
Hunt

matched the trend in section C (Table 11).

The only instance in which

the trends in total macroinvertebrate numbers departed was between the
samples collected on July 12 when the total numbers in section C
declined slightly from the previous sample whereas the total number in
section B on July 12 increased over the previous sample period (Figure
34) .
The only taxa-specific BACI test that produced a significant
result was for the Heptageniidae (Table 12).

However, it is not clear

whether the difference between pretreatment and treatment mean
differences was due to the reduced flow because the densities of
Heptageniidae in both sections B and C were declining prior to the
withdrawal period and continued to decline through the summer.

However,

the rate of decline in Heptageniidae density was faster in section B
before the withdrawal was initiated and continued at a faster rate
through most of the summer, although the densities of Heptageniidae in
sections B and C were virtually identical at the time the final sample
was collected (Table 11).

This difference in rates of decline between

the sections resulted in the sign and magnitude of the treatment mean
difference changing after the withdrawal was initiated.

I also was

unable to detect a mean differences for the three most abundant taxa
(Gammaridae, Chironomidae, and Elmidae larvae, Table 12).
Invertebrate Drift
Invertebrate drift density (individuals'm

_

3

) was elevated in

section B immediately following the initiation of each water withdrawal
event (June 2 and August 22) but returned to prewithdrawal levels within
24 hours (Figures 35-37).

The elevated drift densities on June 2 were

126

T ab le 12.

M ea n difference betw een p retreatm ent and treatm en t period benthic m acroinvertebrate

density estimates between sections B and C and BACI analysis resullts for 1994 data
(df=4 for all tests).

P re tre atm e n t Mean

T re a tm e n t Mean

D iffe re n c e (# m 2)

D ifferen ce (# m ‘2)

t s ta tis tic

p value

Baetidae

-331.0

5.3

-2.5

0.09

Ephemerellidae

454.5

100.5

2.5

0 19

Heptageniidae

406.0

-128.5

4.9

0.05

Hydropsychidae

133.5

394.3

-1.5

0.21

1055.5

418.0

1.9

0.13

Rhyacophilidae

-6.5

31.3

-1.0

0.36

Perlodidae

o
CIO

17.3

-1.2

0.30

Nemouridae

-6.5

-54.5

0.8

0.52

Elmidae adult

175.5

163.8

0.1

0.94

Elmidae larvae

571.5

1198.8

-2.7

0.06

Simuliidae

-270.0

-0.5

-1.8

0.31

Empididae

-9.5

-12.8

0.3

0.79

Ceratopogonidae

6.5

6.3

0.04

0.98

Chironomidae

-90.9

29.5

-0.7

0.51

Gammeridae

-419.0

-736.7

0.5

0.63

Total Density

1775.5

1888.8

-0.15

0.89

Taxa

Glossosomatidae

127

12000

Macroinvertebrates m

10000

-

8000

6000

4000

Section B
■Section C

2000

21-Apr

11-May

31-May

20-Jun

10-Jul

30-Jul

19-Aug

8-Sep

Date

Figure 34.

Total macroinvertebrate density (number’m 2 , with standard
-

error bars)

in sampled riffles in sections B and C of Hunt

Creek, summer, 1994.

June 1 samples were collected

immediately prior to the start of the withdrawal period.

128

12

— Section B
- - Section C

Total Macroinvertebratesm

10

8

6

4

2
50% Diversion
Initiated @ 1240

o
6/1/93
1200

6/1/S3 6/1/93 6/2/93 6/2/93 6/2/93
16:00 20:00 000
4:00
8:00

6/2/93 6/2/93 6/2/93 6/3/93 6/3/93 6/3/93
1200 16:00 20:00 000
4:00
8:00

6/3/93
12:00

Dote/Time

Figure 35.

Macroinvertebrate drift density (number'm 3 , with standard
error bars) in section B immediately before and after the
initiation of the withdrawal period in 1993 and in section
C.

Withdrawal period began at approximately 12:40 on June

2, after the 12:00 sampling was completed.
samples were collected on June 6 and 7.

129

Section C drift

S ection B
S ection C
Total Macroinvertebrates.m-3

/J-

T /

i

o -l—

1--

8/2/93

8/2/93

8:00

12:00

8/2/93
16:00

8/2/93

8/3/93

20:00

0:00

1-8/3/93
4:00

8/3/93

8/3/93

8:00

12:00

8/3/93
16:00

Date/Time

_

Figure 36.

Macroinvertebrate drift density (number"ra
error bars) in section B

3

, with standard

on August 1-3, 1993.

Section B

samples were collected on August 1 and 2 and section C
samples were collected on August 2-3.

130

10

Total Macroinvertebratesm

— Section B
- - Section C

I

Additional Diversion
Initiated @ 8:30
8/21/93
8:00

8/21/93 8/21/93 8/21/93 E/22/93 8/22/93 8/22/93 8/22/93 8/22/93 8/22/93 8/23/93 8/23/93 8/23/93
12:00 16:00
20:00
000
4:00
8:00
1200
16:00
20:00
800
4:00
8:00

Date/Time

Figure 37.

Macroinvertebrate drift density (number"m
error bars)

_

3

, with standard

in section B immediately before and after the

initiation of the additional withdrawal period in August,
1993.

Withdrawal period began at approximately 8:30 on

August 22.

Section C samples were collected on August 25

and 26.

131

not significantly different from pretreatment levels (t=-1.39, df=10,
p=0.19) but the elevated densities on August 22 were significantly
different

(t=-2.98, df=lO, p=0.02). Also, total drift densities were

similar between sections B and C during the July and early August
sampling.
Discussion
Total benthic macroinvertebrate density in Hunt Creek did not
decline as a result of the 50% reduction in summer stream flow in the
treatment section.

Further, there is no evidence for an impact on any

of the individual families examined except, perhaps, the Heptageniidae
(Table 11).

The significance of the test result for Heptageniidae

density may be due to chance alone because I did not adjust p values for
multiple comparisons as suggested by Smith et a l . (1993).

The families

I selected for habitat modeling and the additional three families
selected for BACI analysis represented 93% of the benthic
macroinvertebrate fauna in Hunt Creek.

It is unlikely that any of the

remaining taxa were impacted by the reduced flow.

This result is

consistent with the observation that the withdrawal reduced the total
surface area of the stream less than 6%.

I recognize that I probably

did not adequately sample the first and second instars of many of the
taxa because of the mesh size used on the Hess sampler.

However, it is

unlikely that sampling the early instars would have influenced the
results of the study because the model predictions matched the observed
trends in density of the macroinvertebrates.
The PHABSIM model indicated that reducing stream flow more than
50% would reduce WUA in both reaches for all taxa and that WUA losses
would be substantial.

A PHABSIM analysis of brook trout habitat in

132

section B of Hunt Creek during the same period indicated that brook
trout WUA would not be substantially reduced by reducing the streamflow
until the discharge was equal to approximately 20% of baseflow, or to
0.09 m 3 ‘s

-

1

(Chapter 2).

These results indicated that brook trout food

resources may be impacted by a lesser reduction in flow, one that would
not be sufficient to reduce brook trout foraging or resting habitat.
The benthic macroinvertebrates in Hunt Creek comprise over 85% of the
diet of the brook trout, both by volume and caloric value (Alexander and
Gowing 1976).

A flow reduction in Hunt Creek greater than 50% could

adversely affect the brook trout population even if other measures of
brook trout habitat did not decline.
There were no obvious differences in the relation between WUA and
discharge among the various taxa.

The relation between WUA and discharge

for taxa that were most common in riffle habitats

(e.g. Heptageniidae and

Hydropsychidae) had a similar shape when compared to the relations for taxa
that were more common in depositional or pool habitats (e.g. Tipulidae and
Empididae)

(Figure 32).

The only clear difference in the relation between

WUA and discharge was between reaches B2 and B4 .

In general, the WUA

relations for reach B4 indicated a reduction in WUA for benthic
macroinvertebrates in Hunt Creek when summer stream flow was reduced 50%,
although the reductions were minor.

In contrast, WUA curves from reach B2

indicated a modest increase in WUA when summer stream flow was reduced 50%.
The differences in the WUA curves between reaches B2 and B4 are undoubtedly
due to the differences in velocity and depth between the two reaches.
The similarity between the HSC and the WUA-discharge relations for
the taxa modeled is probably due to the methods used to construct the HSC.
I did not weight the value of the habitat parameters for the individual

133

samples because the habitat parameters measured and modeled with PHABSIM
(depth and mean column velocity) may not be the most important parameters
influencing the abundance of macroinvertebrates at the sample location.
For example, Peckarsky (1984) noted that predation can be an important
factor influencing the distribution and abundance of both predator and prey
species of macroinvertebrates in localized areas.

Further, it is unlikely

that mean column velocity and depth are as important as the shear stress
and thickness of the boundary layer in determining the abundance of
macroinvertebrate taxa at specific locations in a stream (Osborne et a l .
1985).

The substrate composition at a location is undoubtedly important in

determining the abundance of specific taxa of macroinvertebrates at
locations in a stream (Minshall 1984).

However, the PHABSIM system assumes

substrate composition is constant and therefore, is not as important in
determining the shape of WUA-discharge relation as depth and mean column
velocity.
I did not conduct the PHABSIM analysis on the habitat of the
macroinvertebrate families that occurred in less than 20% of the samples
because these families represented less than 7% of the benthic organisms
encountered during sampling and habitat suitability criteria constructed
for extremely small sample sizes are generally not representative of the
actual habitat suitability requirements of the taxa (Bovee 1986).

If a

habitat analysis for the infrequently occurring taxa was necessary it would
require altering the sampling protocol to more adequately sample those
taxa.

In addition, I did not analyze the habitat for the most frequently

encountered taxa (found in more than 80% of the samples) because nearly all
microhabitats sampled were suitable for those taxa.

As a result, the HSC

for those taxa would have been too broad to be useful as a predictor or the

134

change in habitat for those taxa and the WUA-discharge relation would have
looked similar to the Total Area-discharge relation (Figure 19, Chapter 2).
In addition to macroinvertebrate habitat, invertebrate drift can be
affected by changes in discharge (Anderson and Lehmkuhl 1968; Minshall and
Winger 1968; Hooper and Ottey 1988; Poff and Ward 1991).

Poff and Ward

(1991) demonstrated that drift increased for most taxa during simultaneous
experimental streamflow reductions and elevations.

I documented an
_

increase m

invertebrate drift densities

(number'm

3

) in the 24 hour period

immediately following flow reduction but drift densities returned to
prewithdrawal levels quickly, indicating the response of the invertebrates
in section B to the reduced discharge was a short term, catastrophic drift
response (Waters 1972).

The only aspect of the total macroinvertebrate

drift that might have been altered by the reduction of flow was the rate of
downstream movement of the invertebrates due to the reduced velocity in the
treatment section.

This reduced rate of downstream movement may alter

macroinvertebrate colonization dynamics in dewatered streams, even if other
measures of habitat quality and quantity are not reduced by reduced flows.
Although I was not able to document a short term response to the
reduced flow in Hunt Creek, it is possible that reduced flows could have a
long term impact, particularly if stream edge habitats important for
oviposition are subjected to drying over successive years.

If stream edge

habitats were dewatered it could lead to reduced reproductive success for
taxa that use overhanging vegetation and stream edge substrates for egg
laying.

This in turn could lead to reduced production of

macroinvertebrates in the stream.
The results presented here also point out that habitat components
normally not considered in a PHABSIM analysis may be important in

135

determining the response of fish to reduced flows.

Orth (1987) argued for

including broader ecological analyses in impact assessment, yet most of the
PHABSIM studies conducted since 1987 have focused on fish microhabitat
alone (Conder and Annear 1987; Scott and Shirvell 1987; Bovee et a l . 1994).
In this study I found that brook trout food-producing habitat may be
decreased by a reduction in flow that would not produce a reduction in
brook trout WUA (Chapter 2).

For example, the PHABSIM model indicated

benthic macroinvertebrate WUA in section B would be reduced approximately
52% and 66% in reaches B2 and B4, respectively if summer flow was reduced
85% but brook trout WUA would only be reduced 10%.
Finally, I want to stress that the results presented here are unique
to Hunt Creek and should not be applied to streams throughout Michigan or
the midwest.

It is likely that a withdrawal of 50% of summer baseflow

would have a greater impact on both fish and benthic macroinvertebrate
habitat in marginal trout streams or streams that do not have high quality
habitat under summer baseflow conditions.

For example, a withdrawal of 50%

of summer baseflow in Fish Creek, a marginal brown trout stream in Ingham
County, Michigan would result in a 30% reduction of WUA for Hydropsyche
spp. and a 25% reduction of WUA for Ephemerella spp. (esimated from figures
in Gowan 1984) .

Clearly, the potential impacts of proposed water

withdrawals from streams must be evaluated stream by stream.

136

Literature Cited
Alexander, G. R. and H. Gowing. 1976. Relationships between diet and growth
in rainbow trout (Salmo gairdneri), brook trout

(Salvelinus

fontinalis) , and brown trout (Salmo trutta). Michigan Department of
Natural Resources Fisheries Research Report No. 1841.
Alexander, G. R. and E. A. Hansen. 1986. Sand bed load in a brook trout
stream. North American Journal of Fisheries Management 6:9-23.
Allan, J. D. 1981. Determinants of diet of brook trout (Salvelinus
fontinalis)

in a mountain stream. Canadian Journal of Fisheries and

Aquatic Sciences 38:184-192.
Anderson, N. H., and D. M. Lehmkuhl. 1968. Catastrophic drift of insects in
a woodland stream. Ecology 49:198-206.
Anderson, R. 0. 1959. A modified flotation technique for sorting bottom
fauna
Bechara, J.
trout

samples. Limnology and Oceanography 4:223-225.
S., G. Moreau, and D. Planas. 1992. Top-down effects of brook
(Salvelinus fontinalis) in a boreal forest stream.Canadian

Journal of Fisheries and Aquatic Sciences

49:2093-2103.

Bovee, K. D. 1978. The incremental method of assessing habitat potential
for coolwater species, with management implications. Transactions of
the American Fisheries Society Special Publication 11:340-346.
Bovee, K. D. 1985. Evaluation of the effects of hydropeaking on aquatic
macroinvertebtrates using PHABSIM. Pages 236-241 in Proceedings of
the Symposium of Small Hydropower and Fisheries. Olson, F. W., R. G.
White, and R. H. Hamre editors. American Fisheries Society, Bethesda,
MD.
Bovee, K. D. 1986 Development and evaluation of habitat suitability
criteria for use in the Instream Flow Incremental Methodology.

137

Instream Flow Information Paper 21. U.S. Fish and Wildlife Service
Biological Report 86(7). 235 pp.
Bovee, K. D., T. J. Newcomb, and T. G. Coon. 1994. Relations between
habitat variability and population dynamics of bass in the Huron
River, Michigan. National Biological Survey Biological Report
1994(21). 63 pp.
Chapman, D. W. 1966. Food and space as regulators of salmonid populations
in streams. The American Naturalist 100(913):345-357.
Conder, A. L. and T. C. Annear. 1987. Test of weighted usable area
estimates derived from a PHABSIM model for instream flow studies on
trout streams. North American Journal of Fisheries Management 7:339350.
Corrarino, C. A., and M. A. Brusven. 1983. The effects of reduced stream
discharge on insect drift and stranding of near shore insects.
Freshwater Invertebrate Biology 2:88-98.
Degani, G . , G. N. Herbst, R. Ortal, H. J. Bromley, D. Levanon, Y. Netzer,
N. Harari, and H. Glazman. 1993. Relationship between current
velocity, depth and the invertebrate community in a stable river
system. Hydrobiologia 263:163-172.
Elliott, J. M. 1973. The food of brown and rainbow trout (Salmo trutta and
S. gairdneri) in relation to the abundance of drifting invertebrates
in a mountain stream. Oecologia 12:329-347.
Gore, J. A. and R. D. Judy, Jr. 1981. Predictive models of benthic
macroinvertebrate density for use in instream flow studies and
regulated flow management. Canadian Journal of Fisheries and Aquatic
Sciences 38:1363-1370.

138

Gore, J. A. 1989. Models for predicting benthic macroinvertebrate habitat
suitability under regulated flows. Pages 253-265 in Gore, J. A. and
G. E. Petts, editors. Alternatives in regulated river management. CRC
Press, Boca Raton, FL.
Gowan, C. 1984. The impacts of irrigation water withdrawals on brown trout
(Salmo trutta) and two species of benthic macroinvertebrates in a
typical southern Michigan stream. M. S. thesis, Michigan State
University, East Lansing. 127 pp.
Hooper, F. F., and D. R. Ottey. 1988. Responses of macroinvertebrates of
two headwater streams to discharge fluctations. Verh. Int. Ver.
Limnol. 23:1159-1166.
Mathur, D., W. H. Bason, E. J. Purdy, Jr., and C. A. Silver. 1985. A
Critique of the Instream Flow Incremental Methodology. Canadian
Journal of Fisheries and Aquatic Science 42:825-831.
Merritt, R. W . , and K. W. Cummins. 1984. An introduction to the aquatic
insects of North America, second edition. Kendall/Hunt Publishing
Company, Dubuque, Iowa.
Milhous, R. T . , M. A. Updike, and D. M. Schneider. 1989. Physical Habitat
Simulation System Reference Manual-Version II. Instream Flow
Information Paper No. 26. U.S. Fish and wildlife Service Report
89 (16) .
Minshall, G. W. 1984. Aquatic insect-substratum relationships. Pages 358400 in Resh, V. H. and D. M. Rosenberg, editors. The Ecology of
Aquatic Insects. Praeger Publishers, New York.
Minshall, G. W . , and F. V. Winger. 1968. The effect of reduction in stream
flow on invertebrate drift. Ecology 49:580-582.

139

Morrison, D. F. 1990. Multivariate Statistical Methods, third edition.
McGraw-Hill Publishing Company, New York.
Nielsen, J. L. 1992. Microhabitat specific foraging behavior, diet, and
growth of juvenile coho salmon. Transactions of the American
Fisheries Society 121:617-634.
Orth, D. J . , and O. E. Maughan. 1982. Evaluation of the Incremental
Methodology for recommending instream flows for fishes. Transactions
of the American Fisheries Society 111:413-445.
Orth, D. J. and 0. E. Maughan. 1983. Microhabitat preferences of benthic
fauna in a woodland stream. Hydrobiologia 106:157-168.
Osborne, L. L., Herricks, E. E., and Alavian, V. Characterizationof benthic
microhabitat: an experimental system for aquatic insects.
Hydrobiologia,

123:153-160.

Peckarsky, B. L. 1984. Predator-prey interactions among aquatic insects.
Pages 196-254 in Resh, V. H. and D. M. Rosenberg, editors. The
Ecology of Aquatic Insects. Praeger Publishers, New York.
Pennak, R. W. 1989. Freshwater inbertebrates of the United States Protozoa
to Mollusca, third edition. John Wiley and Sons, Inc. New York.
Poff, N. L., and J. V. Ward. 1991. Drift responses of benthic invertebrates
to experimental streamflow variation in a hydrologically stable
stream. Canadian Journal of Fisheries and Aquatic Sciences 48:19261936 .
Rabeni, C. F., and G. W. Minshall. 1977. Factors affecting
microdistribution of stream benthic insects. Oikos 29:33-43.
Reiser, D. w., T. A. Wesche, and C. Estes. 1989. Status of Instream Flow
legislation and practices in North America. Fisheries 14(2):22-29.

140

Scott, D. and C. S. Shirvell. 1987. A critique of the instream flow
incremental methodology and observations on flow determination in New
Zealand. Pages 27-43 in J. F. Craig and J. R. Kemper, eds. Regulated
Streams Advances in Ecology. Plenum Press, NY.
Smith, E. P., D. R. Orvos, and J. Cairns, Jr. 1993. Impact assessment using
the before-after-control-impact

(BACI) model: concerns and comments.

Canadian Journal of Fisheries and Aquatic Sciences 50:627-637.
Somerville, P. N. 1958. Tables for obtaining non-parametric tolerance
limits. The Annals of Mathematical Statistics 29:599-601.
Stewart-Oaten, A., J. R. Bence, and C. W. Osenberg. 1992. Assessing effects
of unreplicated perturbations: no simple solutions. Ecology 73:13961404 .
Stewart-Oaten, A., w. w. Parker, and K. R. Parker. 1986. Environmental
impact assessment:

pseudoreplication

in time? Ecology 67:929-940.

Ward, J. V. 1976. Effects of flow patterns below large dams on stream
benthos: a review. Pages 235-253 in J. F. Osborn and G. H. Allman,
editors. Instream Flow Needs, Solutions to Technical, Legal, and
Social Problems Caused by Increasing Competition for Limited Stream
Flow. Symposium and Speciality Conference, Boise, ID, May, 1976.
American Fisheries Society, Bethesda, MD.
Waters, T. F. 1972. The drift of stream insects. Annual Review of
Entomology 17:253-272.

141

SUMMARY

The results presented here clearly illustrate that a 50% reduction of
summer baseflow in Hunt Creek did not reduce the population density of
brook trout or the benthic macroinvertebrate assemblage.

The results also

indicate that the PHABSIM modeling system may be appropriate for use in
Michigan streams for predicting the impacts of altered stream flows on
stream fish populations because the predictions of the model matched the
observed trends in fish abundance.

The model indicated that the withdrawal

of 50% did not reduce the foraging, resting, or food producing habitat
available for brook trout.

However, it is not certain that the PHABSIM

procedure is an accurate predictor of the changes in a fish population that
result from altered flows, because the flow reduction applied to Hunt Creek
was not large enough to produce an effect.

A much more rigorous test of

the model would be to reduce the flow in Hunt Creek to a level that would
be expected to produce a reduction of brook trout standing crop or
abundance and compare population response to the predicted effect size.
These results should not be interpreted to mean that all streams in
Michigan or the Midwest could withstand a 50% reduction in summer flow with
no adverse impact to the fish or other biota.

The response of the biota in

a particular stream to a flow reduction depends to a large extent on local
watershed attributes, the resident species assemblage, and stream
characteristics such as channel shape and water source.
results presented here are unique to Hunt Creek.

Therefore, the

Future water withdrawals

from streams in Michigan need to be evaluated on a case by case basis.
The alternative HSC presented here also offer a potential improvement
over existing HSC developed from frequency-of-use data.

142

However, the model

developed here could be improved by developing all model paramaters based
on performance of brook, trout (e.g. maximum capture distance) and by
verifying that bioenergetic criteria accurately describe the habitat
selection tendencies of foraging brook trout.

143

Appendix A

Appendix A

Brown trout and rainbow trout WUA (m2 '100 m -1 ) in relation to discharge
(m^'s

in reaches B2 and B4, Hunt Creek.

160 -7
140
120

E 100
80

B2
B2
B4
B4

60
40
20

brown
brown
brown
brown

trout
trout
trout
trout

juvenile
adult
juvenile
adult

-

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

0.45

0.50

Discharge (m 3s'1)

Figure 38.

Juvenile and adult brown trout WUA in relation to discharge in
reaches B2 and B4, Hunt Creek derived from HSC data presented
in Raleigh et a l . (1986; see Literature Cited fo Chapter 2).

144

300 j

!

WUA (m2100 m'1

250 +

B2
B2
B4
B4

rainbow
rainbow
rainbow
rainbow

trout juvenile
trout adult
trout juvenile
trout adult

2 0 0 J-

150 -

100 1

0 t------1------1
------1------(------1------ 1------ 1------1----- -f----- i
0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

0.45

0.50

Discharge (m3s’1)

Figure 39.

Juvenile and adult rainbow trout WUA in relation to discharge
in reaches B2 and B4, Hunt Creek derived from HSC data
presented in Raleigh et a l . (1984; see Literature Cited for
Chapter 2).

145