z:
x . {a .r .
r . . Em: 5.1.51,
ll‘x'lzn. $1.»: .55.... 8.5)".
via-Ingliis .316... I
5;... ‘3..- . .. k1 .
u. 4..
. .1}... .
x

.51
I

 

.Iti'zr.
as 1v}...

1.

. 3....
.i:......i a»...
by... I ft.

“5....
kit‘s." {8% .
:3...:..:......: exit SH.
.15....39Ln
.issi
5.08..)
l 3:
53:33.3...

..

.485

l
1;}.

a: .19

1.. tiny .

3.... mu... 3...
.
3.3.5....“ S

I .
3 I 11““ i
ad.n.roxti£
{4:51 :5.
1.5

 

I 7]. t
91.!
3..
is.) :5
z. 7.1.
£335.“... .
..: .511...
yx
)meta

...\
‘15.!
x . . ,
(a.

31......
1:31:
I}.
5...? . P

l . gist-.3... c at}
.. .v:

s». :2,

Stir...

 

 

 

 

I," .
I z 1». Lu...
1 u!
f
girl-Liaiu
tlw {If}!

cl

I‘ll.

.‘i .l
5. ’1‘ 9:13.41
1: aunt)...

.3.

£1.55...

31!! 'i
$25.13;..yl. ul .c.
II It: 9..

0.1-

 

This is to certify that the

dissertation entitled
Spawning Migration and Habitat Selection by Steelhead
an LongHOSe Suckers in the Pere Marquette and
1:. Joseph Rivers, Michigan

presented by
Robe r t Douglas Workman

ment
has been accepted towards fulfill
of the requirements for

W 11d].
. FiSh. & o
PhoD’ dc ee
J..— g

 

Major 1%

March 1, 2002
Date—J

'9? Action/Equal Opportunity Institution 0- 12771
MSU is an Amy-"wt“

 

 

 

LIBRARY
Michigan State
University

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE I DATE DUE ' DATE DUE

 

9 PDECJ' 107420041

 

 

 

 

 

 

 

 

 

 

 

 

 

6101 c:/ClRC/DateDue.965-p. 15

 

 

 

SPAWNING MIGRATION AND HABITAT SELECTION BY STEELHEAD AND
LONGNOSE SUCKERS IN THE PERE MARQUETI'E AND ST. JOSEPH
RIVERS, MICHIGAN
By

Robert Douglas Workman

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

2002

 

“I

ABSTRACT

SPAWNING MIGRATION AND HABITAT SELECTION BY STEELHEAD AND
LONGNOSE SUCKERS IN THE PERE MARQUETTE AND ST. JOSEPH RIVERS
MICHIGAN '

By
Robert Douglas Workman

I investigated the migratory behavior Of steelhead (Oncorhynchus mykiss) and

longnose suckers (Catostomus catostomus) in the Pere Marquette River using

radio telemetry to provide base-line information prior tO the construction Of a

proposed electrical sea lamprey (Petromyzon man'nus) barrier. Steelhead

arrived at the barrier vicinity on average within 8 days in 1997 and 18 days in

1998, and moved upstream quickly through the barrier vicinity, averaging 6

minutes (1997) and 32 minutes (1998). Longnose suckers arrived at the barrier
on average within 17 days in 1998 and passed through the barrier vicinity within
an average Of 20 minutes. Steelhead and longnose sucker migrations

corresponded with increasing water temperature and stream flows.

In addition to the Pere Marquette data, I used steelhead fishway passage data
from the St. Joseph River, Michigan tO develop a temperature-based movement
rule to quantitatively predict the probability Of Upstream movement. Exponential,

'09 istic and power functions were evaluated as a means tO express the

 

Probability Of movement. Of these, the power function provided the closest fit
between Observed and predicted movement. Stream flow was also evaluated as
a rTieans Of expressing the probability Of movement, but did not increase the

Predictive power Of the model. Therefore, I used water temperature to predict

w!

upstream movements. The temperature-based movement model incorporat$8
an increasing probability Of movement for increasing water temperatures aOQVe
the minimum temperature threshold for movement. By using data from two Lake
Michigan tributaries, I was able tO demonstrate that the modeling approach is
transferable to other Great Lakes tributaries and that the model consistently
demonstrates the upstream movement probability Of steelhead in systems where

upstream migration is governed by water temperature.

Finally, I evaluated features (groundwater presence, substrate particle size,
etc...) associated with the selection Of steelhead spawning habitat the Pere
Marquette River. Steelhead redds were evaluated for the presence Of
groundwater using a GIS-based groundwater prediction model and another
method based on intragravel temperature to provide insight to the importance Of
groundwater as a means Of spawning habitat selection. The GIS model and
Probe-based evaluation were inconclusive as means to identify groundwater
associated with steelhead redds. Steelhead preferred tO construct redds in a
substrate consisting Of small gravel, large gravel, and small cobble particle sizes

disproportionately to clay, silt, sand, and large cobble. Steelhead redds were

 

located in areas where the stream velocity was significantly higher (F = 97.77, P
< 0 .0001) than velocities that were recorded at reference sites, and redds were

located in water that was significantly shallower than what was typically found in
the study reaches (F = 113.84, P < 0.0001). Stream temperature did not appear

tO influence the selection Of redd locations.

Copyright by
ROBERT DOUGLAS WORKMAN
2002

 

 

For Krissy, Madeline, Ryan, and Logan.

 

ACKNOWLEDGEMENTS

Funding for this research was PFOVICIEG by the Michigan Department Of Natural

Resources and the Great Lakes Fisheries Commission.

I would like tO thank my advisor Tom Coon for providing me the Opportunity to

conduct this research, and for his guidance and assistance throughout this
learning process. I would also like tO thank Dan Hayes for his guidance and
assistance above and beyond the call Of duty. My gratitude is also extended tO

my committee members Stuart Gage and Rich Merritt for their guidance.

I would like to thank Bill Taylor for his assistance, Jim Bence for his statistical
advice, Amos Ziegler for GIS assistance, Jim Dexter for the St. Joseph River

data, and Ellie Koon for some Pere Marquette data and her guidance. In

addition, I would like tO thank Jim Willson and the College of Business and for

their assistance, and Chuck Wolverton and the staff Of Northern Ecological

Services, Inc. for their encouragement and support.

I need to thank my fishing buddies (they were also my field crew), Jamie

Ladonski, Mike Wilson, Aaron Snell, Dan Weisenreder, and Mike Wilberg. The

best part about having to fish for my research, was having your company. “Quick

wit”, and your expertise. Mmm, smoked BKD salmon.

vi

I would like to thank Ken Smith for providing me access to the Pere Marquette
River and the Pere Marquette Watershed Council for their interest and

assistance in this research.

Because of this research project, I was fortunate to meet and become good

friends with a family from Ludington. The world could use more people like the

Soberalskis. Thanks for your friendship and help.

Finally, l would like to thank my wife Krissy and our kids Madeline. Ryan, and
Logan for their love and support, and for tolerating my mood swings. It’s been
difficult raising a family while trying to complete school, but I couldn’t have done it

without you. I would also like to thank my mom and dad, and Gary and Hallie for

their love and support.

vii

TABLE OF CONTENTS

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

LIST OF TABLES .............................................................................................. x
LIST OF F'GURES .................................................................................................... Xiii
CHAPTER 1
iNTRODUCTION ................................................................................................. 1
CHAPTER 2
STUDY AREAS ...................................................................................................... .7
CHAPTER 3
A Description of the Spawning Migrations of Steelhead and Longnose Suckers in
the Pere Marquette River. Michigan ...................................................... -- ...... 10
Abstract ......................................................................... 11
introduction.........................................................'.'.'.'.'.'.'.'.'.'.'.'.'II'I .......................................... 12
Methods ............................................... --.1 5
Fish Capture- - ______ _,,--__ -.1 5
Radio Tag Implant Procedure _ ______ _______ _ - _-.19
Monitoring Movements of Radio- Tagged Fish. ___________ _ _ ___20
Study Area _______ - _ - .................... 2 3
Data Analysis............ " ................. 2 5
Environmental Monitoring ________________________ __ ............... 2 5
Results ..................... ......................................................................... 2 5
Fish Capture.....-_ ........................................ 2 5
Timing of Migration--- ......................................... 2 8
Speed of Movement Through Base Station Reception Areas ______________________ 2 8
. Percent Passage Through Bases Station Reception Areas ,,,,,,,,,,,,,,,,,,,,,,,,, 35
Dlat:ussion................_ ........................................................ 36
Timing of Migration ............................................................................................... 3 6
Percent Passage Through Base Station Reception Areas ,,,,,,,,,,,,,,,,,,,,,,,,,, 37
Evaluation of Capture Methods ................................... _ ______________ 39
Future CONSIderatIonsse
CHAPTER 4
A Model of Steelhead Movement in Two Lake Michigan Tributaries in Relation to
Water Temperature and Stream Flow ...................................................... 4 1
AbStra ct. .............................................................................. 4 2
”duct .............. 4 s
M9th0ds.-.... """ - ............... 4 6
M’Qration Data SouréZe’éIIILIXIII.....I.III .............................................................. 4 6
Mo de/ Development. ...................................................................................... 4 8
Resu'ts ............................... .................................................................................. 56
We tar Temperaturess

viii

Water Temperature and Stream Flow

 

 

 

 

 

 

................................................................ 53
Discussion...,.== ......... ...... _ - .......................... 55
Waiter Temperature................--- ::: ‘ 1:11:66
Waiter Temperature and Stream FIOW. ............................... I: ............................ 69
CHAPTER 5

A Description of Spawning Habitat Use and its Relation to Areas of Groundwater
Input in the Pere Marquette River. Michigan .................................................................

 

 

 

 

 

 

 

 

 

 

 

 

73
Abstract ................................................................................... 74
Introduction ................................................. 7 5
Methods ............................................ 7 5

Sample Locations____ _________________ __ ..... 8 0
Groundwater Evaluation ............................ " """"IZILZ .......... 1.5. ............. ao
Redd Characteristics - .. ---.88
Results........................... .................................... 91
Groundwater............................................. ..................................... 91
Substrate Particle Size ...... .. ...................................... 94
Velocity, Water Temperature, Water oeBthIIIIIIIII .......................................... 99
Discussion ..................................................................................................................... 104
Groundwater _ - __104
Substrate Particle Size __ __ ,,,,, 105
Velocity. Water Temperature, Water Depth .................................................... 105
CHAPTER 6
SUMMARY ...................................................................................................................... t 09
REFERENCES .................................................................................................................. 113

LIST OF TABLES

Table 3-1. Catch statistics of steelhead and longnose suckers from the Pere
Marquette River, Michigan, 1997 — 1998. The statistics include standard
deviation (0') of the mean weight and length of the captured fish as well as
the number of male and females that were radio-tagged each year .......... 27

Table 3-2. Number of fish arriving at base stations and time taken to reach base
station for steelhead in 1997 and 1998 and longnose suckers in 1998 ..... 32

Table 4-1. The sum of squared residual (RSS) values for the power, logistic and
exponential form of the temperature-based movement model for the Pere
Marquette (1998) and St. Joseph (1993 to 1999) Rivers, Michigan. A low
value indicates the best fit for that form of the model and is indicated by an
asterisk for each year ............................................................................................ 59

Table 4-2. Parameter estimates (1' 1 standard error) for the power function
version of the Temperature-Based Movement model (TBM) for the Pere
Marquette River and St. Joseph River, Michigan 1993 to 1999. The
parameter values and approximate standard error estimates for the
reduced model (combined 1993 to 1999 St. Joseph River data) that was
used in the F test to compare among years for the St. Joseph River data
are also included ................................................................... -- .......... 51

 

Table 4-3. The sum of squared residual (RSS) values for the power, 'OQlStiC and
linear function flow-temperature-based movement models, using relative
deviation from mean daily flow ((X — XV}? ) to represent flow for the Pete
Marquette (1998) and St. Joseph (1993 to 1999) Rivers, Michigan- A
power function was used to describe the temperature portion of the stream
flow * water temperature-based function. Power, logistic and linear refer

to the flow portion of function used to describe steelhead movement. A
low value indicates the best fit for that form of the model and is indicated

by an asterisk for each year ........................................................... ......... 65

 

Table 5-.1. Particle size classes that were used to classify substrate particle
Sizes of sample sites in the Pere Marquette River, Michigan 1997—1999,82

Table 5-2. Redd density by stream reach for the 1997, 1998, and 1999 in the

Pere Marquette River, Michigan ......................................................................... 91

Table 5‘3 - Correlation coefficients for _0 to 100 m categories of Darcy flow area
t"Estimates in study areas from the Pere Marquette River . Michigan ------------ 92

Table 5-4. The mean difference (AT i standard error) between intragravel
temperatures measured at 5 and 10 cm into the substrate and stream
temperatures, and the probe-based and Darcy groundwater flow
predictions in the Pere Marquette River, Michigan study reaches 1998 and
1999 ........................................................ -- ........ 94

Table 5-5. Redd and reference habitat characteristics in the Pere Marquette
River. Michigan 19911998 and 1999 ............................................................... 97

LIST OF FIGURES

Figure 2-1. Locations on the Pere Marquette and St. Joseph Rivers, Michigan
where steelhead migration, water temperature, and stream flow data were
collected. Radio-tagged steelhead passage and water temperature data
were collected at Custer and stream flow data was collected at Scottville
on the Pere Marquette River, Michigan. The bars on the St. Joseph River
map indicate where camera-recorded steelhead passage and water
temperature data (Berrien Springs), and stream flow (Niles) data were
collected .................................................................................................... -

 

Figure 3-1. The location of Pere Marquette River within Michigan (MI), and
notable sites of data collection within the River. The water temperature
sampling location is indicated by “T” .................................................................. 16

Figure 3-2. Hook-and-line steelhead capture area in the Pere Marquette River
1997-1998 ............................................................................................. 17

Figure 3-3. Fyke net set location in the Pere Marquette River, Michigan 1997
and 1998. Inset indicates the spatial arrangement of the Fyke nets ,,,,,,,,, 18

Figure 3-4. Approximate spatial arrangement of 1997 and 1998 radio telemetry
study sections and relative locations of receivers (black boxes) and
antennae placement along the Pere Marquette River, Michigan. The bar
across the river in section B indicates the location of the electric sea
lamprey barrier. The figure is not drawn to scale ........................................... 21

Figure 3-5. The number of radio-tagged steelhead counted/day at the Custer
reception site with mean daily stream flow (dashed line) and mean daily
water temperature (solid line) at the Custer site during the spring 1997
migration on the Pere Marquette River, Michigan ,,,,,, _ .............. 2 9

F'QUl’e 3-6. The number of radio-tagged steelhead counted/day at the Custer
reception site with mean daily stream flow (dashed line) and mean daily
Water temperature (solid line) at the Custer site during the spring 1998
migration on the Pere Marquette River, Michigan ,,,,,,,,,, ___________ 30

 

Figure 3—7, The number of radio-tagged longnose suckers counted/day at the
Custer reception site with mean daily stream flow (dashed line) and_ mean
daily water temperature (solid line) at the Custer site during the spring
1998 migration on the Pere Marquette River, Michigan ________________________________ 31

xii

Figure 4-1. Radio-tagged steelhead passage (number/day, solid bars) by the
Custer reception site on the Pere Marquette River, Michigan based on the
corncrdent stream flow (m3/s, dashed line) at the USGS stream flow gauge

downstream in Scottville and the mean daily mean water temperature (°C.
solid line) recorded at the Custer reception site ______________________

Figure 4-2. The number of steelhead counted using camera-recorded fish

passage video during 1993 in relation to water temperature in the St.
Joseph River. Michigan ........................................................................ 58

Figure 4-3. Plot of Power, Exponential and Logistic models of temperature-

based movement and the observed steelhead movement data in the St.
Joseph River. Michigan. 1996 ............................................................................. 60

Figure 4-4. Plot of the temperature-based probability of movement curves based
on models developed using a power function for the St. Joseph River,
Michigan, from data collected from years 1993 to 1999........ ........... 52

Figure 4-5. The residual plots of the difference between observed and predicted
movement using a temperature-based movement model and stream flow

for data from the Pere Marquette (A) and St. Joseph Rivers (B). Michigan,
1998 ............................................... - --.64

 

 

Figure 5-1. Stream reach segments in the Pere Marquette River, Michigan for
1 997 through 1999. Arrows indicate approximate study reach extents (A
through M). Note: the upstream extent of reach L is M-37 ,,,,,,,,,,,,,,,,,,,,,,,,,, 31

FlQUre 5-2. Reference sample site configuration where RL indicates river—left.
CH indicates Channel, and RR indicated river-right ........................................ 83

FlSlure 5-3. The cumulative percent frequency distribution of total percent
coverage of small gravel (sg), large gravel (lg), sand (8), small cobble (so),
large cobble (lc), silt (slt) and clay (clay) size classes within redd and

reference sites in the Pere Marquette River, Michigan for 1997, 1998, and
1 999 .......................................................................................................................... 96

Figure 5-4. Electivity values for steelhead spawning substrate particle size
descriptions in the Pere Marquette River, Michigan 1998 and 1999
Corhbined .................................................................................................................. 98

Figure 5-—5 - Percent frequency of stream velocity recorded at steelhead redd and
r efe rence sites during 1998 in the Pere Marquette River. Michigan ........ 100

Figure 5-6 - Percent frequency of stream velocity recorded at steelhead redds
duri ng 1997, 1998, and 1999 in the Pere Marquette River, Michigan ....... 101

xiii

Figure 5—7. Percent frequency of water depth recorded at steelhead redd sites
from 1997 to 1999, and from reference sites from 1998 to 1999 in the Pere
Marquette River, Michigan, 103

xiv

CHAPTER 1

INTRODUCTION

 

INTRODUCTION

The general topic for my research is the migration and spawning habitat use of
steelhead (Oncorhynchus mykiss) and the migration of longnose suckers
(Catostomus catostomus). l evaluated steelhead and longnose sucker
migrations in the Pere Marquette River, Michigan to address concerns for the
successful upstream passage of both species beyond an electrical sea lamprey
(Petromyzon man'nus) barrier and fishway in the Pere Marquette River, Michigan.
Previous operations of another electrical barrier in the Pere Marquette River did
not permit the upstream passage of either species (Rozich 1989), and created

concern among recreational anglers who targeted these species.

Rainbow trout (steelhead are the migratory form of rainbow trout) have been a
popular sport fish among Pere Marquette River anglers for many years. Rainbow
trout were first introduced to the Pere Marquette River in 1887, and were
naturalized in this and other Lake Michigan tributaries shortly after initial
introduction (Krueger et al. 1985). The steelhead was among the strains

established in Lake Michigan tributaries.

The predominant life history form of steelhead expressed in Lake Michigan
tributaries is one that includes 2 years of growth in the river after hatching,
followed by a spring migration of smolts to the lake, and then 3 years of growth in

the lake before returning to spawn in the spring of the fifth year (Harbeck 1999).

 

Other variants are expressed in some Lake Michigan tributaries, including some
fish that out-migrate after 1 year of stream growth and others that grow 3 years in .r
the stream before out-migrating. Lake growth ranges from 1 to 5 years. One
other variant, the Skamania strain, was introduced in Michigan waters in the past
20 years (Dexter and Ledet 1997). Fish of this strain also spawn in the spring
(March - May), but unlike the original strain, these fish migrate into their

spawning stream late in the summer preceding spawning (Behnke 1992). Fish of
the original or winter strain delay migration to their spawning stream until late

winter or early spring of its spawning year (March - April).

I evaluated longnose sucker movements with steelhead movements to provide a
comparison between a native non-game species (longnose suckers) and an
introduced game species that migrate at similar times in the Pere Marquette
River. Although sucker species were not considered in the design of the fishway
around the barrier, it was important to document the movement of a non-game
Species and determine if they migrated upstream before it began operation, and

to have a reference for comparison after operation began.

Similar to steelhead in Michigan waters, longnose suckers have been observed
sDawning in April and May in Wisconsin waters (Eddy and Surber 1947), and
initiated spawning migrations in rivers when water temperature rose above 5° C
(Seen et al. 1966). In addition to spawning in rivers, longnose suckers may

Spawn in some shoal waters of the Great Lakes (Bailey 1969). Although juvenile

 

steelhead Spend one to two years in the river after hatching, larval longnose
suckers spend little time in rivers and drift to lakes soon after hatching (Bailey
1969). Longnose suckers are a slow-growing fish that reach maturity in 4 to 9

years (Bailey 1969).

There is a strong body of literature that suggests stream flow and water
temperature may regulate the upstream movements of steelhead and other
salmonids (Northcote 1962; Shephard 1972; Miller 1974; Power and McCleave
1980; Power 1981; Smith et al. 1994). The degree to which water temperature
and/or stream flow regulate upstream migrations depends on the many factors
including watershed size, geology, and weather patterns (Jonsson 1991). To
further evaluate the effect of stream flow and water temperature on fish
migrations, I developed a model to predict the upstream movement of steelhead

based on stream flow and water temperature.

For the final segment of my research, I examined steelhead spawning habitat
Use in the Pere Marquette River. Pacific Coast steelhead exhibit preferences for
Specific water velocities and water depths (Smith 1973), and for coarse-
Particulate substrates small enough to be moved during redd construction, yet
large enough to accommodate sufficient water flow through the redd for
OXYgenation of eggs and alevin emergence (Cooper 1965; Fraser 1985; Sowden

and Power 1985, Kondolf 1988). Although spawning habitat selection by

 

steelhead is well understood in the Pacific Coast region, little is known about

spawning habitat selection in the Great Lakes region. L.

In addition, little is known about the influence of groundwater on the selection of
spawning habitat by steelhead. Groundwater is important to the reproductive
success of the brook trout (Salve/inus fontinalis) and chum salmon (O. keta) by
protecting eggs from ice and infiltrating surface water (Kogl 1965; Fraser 1985;
Curry and Noakes 1995; Curry et al. 1995), and may be important to reproductive
success of steelhead by providing a thermal refuge to developing alevin from
warm water in the summer and more stable flows (Shepherd et al 1986).
Locations where groundwater is abundant in a river may be preferred by
steelhead as spawning areas. Identifying groundwater locations within a river
system where steelhead reproduce could provide insight to the importance of

groundwater and its role in the selection of steelhead spawning sites.

i examined spawning microhabitat (substrate particle size, flow velocity above

the redd, etc...) use characteristics of steelhead in relation to their occurrence in

. selected sections of the river. Identifying habitat features that are important to
SDawning site selection in a Great Lakes tributary can be used to help identify
other areas within the same tributary that may be used by spawning steelhead,
Or could be improved to accommodate steelhead by habitat improvement

Projects that commonly occur throughout the watershed.

 

In summary, I addressed three topics in this dissertation:

o Steelhead and longnose sucker movements in the Pere Marquette River L.
were examined as part of a preliminary evaluation of movement in the
vicinity of an electrical sea lamprey barrier prior to its construction and

operation.

0 The upstream migration of steelhead was modeled in relation to stream

flow and water temperature in two Lake Michigan tributaries.

o Steelhead spawning habitat was described in relation to areas of
groundwater input and was Characterized according to microhabitat

characteristics such as substrate particle size, flow velocity.

By addressing these topics, I will provide baseline information for future
evaluations of steelhead and longnose sucker movements in the vicinity of the
electrical barrier and of the effectiveness of the steelhead passage device that
Was installed at the barrier. I will also further evaluate the influence of stream

flow and water temperature on steelhead spawning migrations and provide

information on steelhead spawning habitat use within a tributary of the Great

Lakes.

 

CHAPTER 2
STUDY AREAS

 

STUDY AREAS

Movement data from radio-tagged steelhead in the Pere Marquette River (Figure
2-1) were used in the studies presented Chapters 3 and 4, and count data of fish
passing through a fish ladder on the St. Joseph River (Figure 2-1) were used in
the study presented in Chapter 4. Steelhead spawning habitat use data in the

Pere Marquette River was used in the study presented in Chapter 5.

The Pere Marquette River is located in west-central Michigan and the main stem
of the river is approximately 154 km long (MDNR/IFR 1988). The river drains
1,955 km2 of watershed and is one of the last large free-flowing Great Lakes
tributary streams in Michigan. The Pere Marquette River is primarily dominated
by a cold-water fish community, and receives spawning migratory steelhead

d 1.: ring the fall, winter, and spring.

The St. Joseph River is located in southwest Michigan and northwest Indiana.

rhe mouth of the St. Joseph River is approximately 200 km south of the Pere

Marquette River and is separated by at least three major tributaries (Muskegon,
grand, and Kalamazoo Rivers) of Lake Michigan. The St. Joseph River is 493
km long and drains a watershed of approximately 11,098 km2 in Michigan and
Indiana (Brown 1944). Six dams located along the river's mainstem regulate
stream flow. In 1975, a fishway was constructed in the downstream-most darn at

Berrien Springs, located 39 km upstream of Lake Michigan (Figure 2-1).

 

 

Scottville Site
PM
River
Lake Custer Slte
Michigan
16 km N T
Lake
Michiga
St. Joseph
River
. Niles MICHIGAN
Berrien Spring
INDIANA

16 km
Figure 2-1. Locations on the Pere Marquette and St. Joseph Rivers, Michigan

where steelhead migration, water temperature, and stream flow data were
collected. Radio-tagged steelhead passage and water temperature data were
collected at Custer and stream flow data was collected at Scottville on the Pere
Marquette River, Michigan. The bars on the St. Joseph River map indicate where
camera-recorded steelhead passage and water temperature data (Berrien Springs),
and stream flow (Niles) data were collected.

 

CHAPTER 3
A description of the migratory behavior of steelhead (Oncorhynchus mykiss) and
longnose suckers (Catostomus catostomus) in the Pere Marquette River,

Michigan

10

 

Abstract
I evaluated steelhead (Oncorhynchus mykiss) and longnose sucker (Catostomus
catostomus) movements in the vicinity of a proposed electrical sea lamprey
(Petromyzon marinus) barrier and fishway, and throughout the Pere Marquette
River during 1997 and 1998, using radio telemetry. Radio-tagged steelhead
moved upstream quickly (a? = 6 minutes, n = 4) through the barrier reception area
from 23 March to 4 April 1997, and from 6 January to 17 April ()7 = 32 minutes, n
= 26) during spring 1998, and the upstream passage times between 1997 and
1998 were not significantly different (F = 0.84, P = 0.37). There was no
significant difference in the mean time from release to arrival at the barrier
between the spring 1997 and spring 1998 radio-tagged steelhead (F = 1.12, P =
0.30). Radio-tagged longnose suckers also were recorded moving quickly (I =
20 minutes, n = 8) upstream from 23 March to 10 April 1998. The upstream
passage of radio-tagged steelhead did not differ from longnose sucker for the
1 9 97 and 1998 data (F = 0.09, P = 0.77). The percent passage of radio-tagged
steelhead through the 150-m long barrier reception area was not significantly
d i1ziterent (P=0.22) between 1997 (57%, n=14) and 1998 (63%, n=54). At least
L<K°lo of the radio-tagged longnose suckers passed through the barrier area.
padio-tagged fish movement indicated that there was a general correspondence
between increasing stream flow and water temperatures. My telemetry data
demonstrated that radio-tagged fish moved upstream quickly through the electric
sea lamprey barrier vicinity, and provided a basis for evaluation fish movement in

the vicinity of the barrier following the construction and operation of the barrier.

11

 

Introduction

Migration barriers can be detrimental to the reproductive success of adult
steelhead (Oncorhynchus mykiss) and other salmonids (Raymond 1979; Powers
and McCleave 1980). Barriers such as hydroelectric dams that do not provide
adequate passage for migrating anadromous steelhead will significantly
decrease the number of adult returns and delay their upstream migration,
resulting in a decrease in the number of steelhead that successfully reproduce
(Raymond 1979). Because barriers can be detrimental to the reproductive
success of steelhead, barriers should be designed to facilitate swift migrations

through barrier areas.

An electric sea lamprey (Petromyzon man'nus) barrier and fishway were
constructed in the Pere Marquette River in 1999, and were designed to
selectively prevent sea lamprey spawning migrations while allowing steelhead to
m 59 rate upstream to spawning locations. Previous operations of an electric
la m prey barrier in the same location as the current barrier did not allow steelhead
to s uccessfully migrate around or through the barrier (Rozich 1989). Although
the newly constructed barrier and fishway were designed facilitate steelhead
gpawning migrations, the operation of the electric barrier may cause delays or

block the spawning migrations of fish species.

The passage of sucker species was not considered as a design element of the

proposed electrical barrier, but suckers are an exploited species in the Pere

12

 

Marquette River. Shortly after the river ice has melted, fishers can be found in

i

abundance along the banks of the river near Branch, Michigan, and downstream
to the mouth of the river, pursuing migratory suckers well into the month of April.
The operation of the electrical barrier may adversely affect future populations of

migratory suckers in the Pere Marquette River.

Steelhead and longnose suckers (Catostomus catostomus) have similar
migration patterns making the two species a likely combination for a dual
examination of their migration patterns. Steelhead and longnose suckers share
spawning migrations during the first half of the year (primarily February through
May) in Michigan rivers. Successful introductions of steelhead into the Great
Lakes region have created opportunities to examine the interactions of non-
native fish with native fish species such as the longnose sucker. The study of
steelhead and longnose sucker migrations in the Pere Marquette River provided
a u nique opportunity to examine the movement and behavior of two fish species
i n a Michigan river where there were no physical barriers to migration for either

3 D ecies.

 

«rhe goal of my study was to evaluate steelhead and longnonse sucker
migrations prior to the construction of an electrical lamprey barrier and fishway
on the Pere Marquette River. I evaluated three aspects of adult movement:
migration timing, movement rates of fish over large distance (103 m), and the

proportion of migrants that passed upstream of the barrier. Specifically, my

13

 

objectives were to:

1) Describe the timing of migration initiation by adult steelhead and
longnose suckers by describing the temporal distribution of steelhead
capture in the reach immediately upstream of Pere Marquette Lake and by
describing the date of arrival of radio-tagged fish at the proposed electric

barrier site.

2) Determine the speed of upstream movement by adult steelhead and
longnose suckers through selected sections of the Pere Marquette River,

and;

3) Determine the percent passage of radio-tagged fish at the proposed

electric barrier.

The information from this study was used to evaluate the effectiveness of the fish

passage device located at the barrier during 2000 and 2001 (Snell 2001).

‘ “formation on the speed and duration of the spawning migrations of steelhead
and longnose suckers was used to determine if the electric barrier had an effect
on the number and timing of steelhead and longnose suckers that passed

upstream after the barrier was constructed and went into operation.

14

 

Methods

Fish Capture

I collected steelhead in the Pere Marquette River (Figure 3-1) for use in the
telemetry study of fish movement from February to May 1997 and collected
steelhead and longnose suckers from November 1997 to April 1998. I used
hook-and-line fishing continually throughout the study within an 8 km reach of the
river immediately upstream of Pere Marquette Lake (Figure 3-2). I fished with
steelhead spawn, Wiggle Warts ®, and Hot-N-Tots ® in stretches of river that
were thought to contain steelhead. I also captured fish in fyke nets measuring

122 by 183 cm with a 5-cm stretch mesh.

A 33-m long lead with a 5-cm stretch mesh was attached to each fyke net. I
fished the fyke nets at various locations within 1 km upstream of Pere Marquette
Lake during the spring 1997 field season. By the end of the spring 1997 field
Season, I chose a location 0.25 km upstream of Pere Marquette Lake, where
“'1 I‘ee fyke nets placed adjacent to each other blocked almost the entire width of
0 h e channel of the river (Figure 3-3). I used the adjacent net configuration for
\“Q spring 1998 field season and I placed the nets in the river from February to
April as soon as ice conditions permitted their use. Fyke nets were not used
during the 1997 fall months (October - December), due to heavy amounts of leaf
material in the river. I checked the nets daily for the presence of steelhead and
longnose suckers. All fish that I captured in the nets were identified and I

measured the length and weight prior to their release.

15

.l‘
it;

 

 

Pere Marquette River

Scottville USGS
Stream Gauge

Bowman Bridge
Receiver Location

/

 

 

2E Custer Electric
-‘-’ Sea Lamprey
E B I
a, arrler
x
(U
_l
Nonh
MI
16 km
PM River

 

Figure 3-1. The location of Pere Marquette River within Michigan (MI), and
notable sites of data collection within the River. The water temperature
sampling location is indicated by T’

16

 

 

 

.mmmesmme 52m osmium—2 Eon 9: E coco oSEoo omo£o9m 3:65.200: .N-m 959“.

cosmoou
3233. :0:me
comm :3 >593th
etoz

acouxm c.5390 :mE

 

comics.

oxm4 -

Stiles Rd
U. S. 31 ,»

 

 

 

//r)
l\ 9.3

H uses—Ems. when

 

 

17

.Eo: min. on“ so EoEomcmtm _mzmam
o£ wofioEE Home. .mmme new home :mmEoS .521 «3:30.62 mood on. E 5:82 How so: 93”. .m-m 0.59“.

cozmoo; uoz

 

 

30—“—

 

 

   

oxmu 83:ng 9.0m

:toz

 

Pere Marquette

$52 as \I/\Iv

18

99519
MW
tag (lit
Monti
tecgitl
to ma
and It
“the
dele
prono
M'f/I //'
by vis
dete
dunng
hand
fish :

and

301

 

the

remc

QXT

(h

 

 

 

   

Ratdio Tag implant Procedure
' Selected steelhead and longnose suckers larger than 1 kg for radio
tag implants. | anesthetized fish that were suitable for implants in a 150 L tank
filled with 60 mg/L tricaine methanosulfate (MS-222) dissolved in river water. I
r eCorded the weight, length, and sex from the fish when they were no longer able
to maintain an upright posture within the anesthetic tank. I also recorded time
and location of capture for each fish that was suitable for radio tag implant. I
attempted to implant radio transmitters at a ratio of 50% males to 50% females. l
dti‘terr'n i ned the sex of the steelhead using several factors: the presence of a
pro/you nced kype (indicative of males); the ease of scale removal (scales detach
With little effort indicates female, not readily detachable indicative of male); and
by visua l inspection of the gonads during the surgical implant procedure. I
detem’i h ed the sex of longnose suckers using visual inspection of the gonads
dutihshe surgical implant procedure, and the presence of pronounced lateral
Wand coloration and tubercles on the anal fin (indicative of males). l inverted the
(79" and placed them in a V-trough surgical table lined with indoor-outdoor carpet,
a(\§‘emoved scales from a longitudinal row for 4 cm along the abdomen
roximately 2 cm posterior of the pelvic fins, and from a small area adjacent to

39' p
dorsal fin. I made a 4-cm long incision in the abdomen region where l

the
removed the scales and implanted the fish with a LOTEK Engineering, Inc.
CF R 'T-7A digitally encoded radio transmitter into the peritoneal cavity. The

digitally encoded radio transmitters were used because of the ability of the

receiver to simultaneously monitor multiple radio signals. The radio tags

19

measmed 16.0 mm in diameter by 83.0 mm long, and weighed 29 gm in air and
12.8 gm in water. I made an incision the width of a surgical blade posterior to the
larger incision and pulled the radio transmitter antenna through. The radio tags
Were rated to last for 282 days at a 5-second burst rate. I closed the incisions
With 000 gut suture and inserted a numbered floy tag in the area where the
Scales were removed adjacent to the dorsal fin. l irrigated the gills of the fish with
r iver water during the surgical procedure. I transferred the radio-tagged fish to
anOthe r tub with river water where the fish remained until it maintained an upright
POStu re within the tub. I released the radio tagged fish as close to the point of
caPI‘Ure as possible and I released fish captured in the fyke nets approximately
200 m upstream of the nets. | used a portable hand-held SRX-400 receiver to

verify that each radio tag was functioning prior to and following surgical

implanla tion.

W O”170/7779 Movements of Radio- Tagged Fish
4119 movements of steelhead that were radio-tagged during the spring 1997 were
@\§\\o‘ed at the site of the electrical barrier in Custer, Michigan (Figure 3-4).
Ra dig tagged fish were monitored using an SRX-400 receiver with W-17
fir ”(aware that allowed for scanning of multiple channels and numerous
ned“ encies deveIOped by LOTEK. Two yagi antennae were mounted 100 m

apart to detect direction of movement, the date and time of arrival of a radio-

tagged fish into the reception area, and the date and time of departure when

20

if

 

c._ 5>c .2me 9 cgmcu “o: m_ 239., one .8252 $563 mom 2:86 9: ho c0380. m5 882?: m :28me
. 9: wwocom can 9: .cmmEoS cozm 26:ng 9mm 9: 92¢ EoEoomE 8595 cam Awmxon x038 23609
B @8280. 3:32 ucm wcozoom 33w 99:29 069 am? can .33 he EwEwmcmtm _mzmaw QwE_xo.aa< 41m 239”.

8282 8:9. $2
255 cmEsom $28.2 co=9w
33 __fl EEOQEB.

3:63 mmmwwma cw: ucm
5:ch 2.;on uwwoaoi

   
    
 

U zO_._.Umm

ctoz

< ZOFUmm

m
20:.omm cozficmag

330282 242;

mg 665 3cm 2:38 do if

 

21

ixmea.
l\\\ «32
penc
Nani

\EC

lrnor
05/279 '
ponab

sefies
appnx
probe
near
aLL
ens

//) 5&9

the d

 

(Same

‘Du

'
Q

 

i‘

the radio-tagged fish left the reception area. I programmed the receiver to scan

it»...

antennae sequentially and record the presence of all coded transmitters that
Were in the listening range of the antenna. The use of coded transmitters
inCreased the time resolution possible for detecting movements of individual fish.
All data were logged in the memory of the receiver, and then downloaded
periodically to maintain a continuous record. The Michigan Department of

Natural Resources provided me a power supply and a structure to house the

receive r.

I monitored the movements of fish radio-tagged during the 1997-98 field season
using two base stations (one was loCated at the Custer site and one was
portab le) , each equipped with a scanning, continuously recording receiver and a
series of fixed-direction antennae. l positioned a portable base station

399\$\“‘lately 12 km downstream of the proposed electric barrier on private
preperty (43° 55’ 9.5”N, 86° 21’ 33.7” W, WGS-84), on the north side of the river,
()fiarthe end of Stiles Road (Figure 3-2). The portable base station consisted of

a \‘Q‘EK SRX-400 receiver with W-17 firmware and a weatherproof secured
en G‘osure- I used a 50'6" panel to supply power to the receiver at the remote
sit %' I mounted two yagi antennae 20 m apart to detect direction of movement,
the date and time of arrival of a radio-tagged fish into the reception area, and the
date and time of when the radio-tagged fish left the reception area. The primary
purpose of using the downstream base station was to gain more information

about the movement of fall radio-tagged steelhead. I left the portable base

22

 

Station at this location until radio—tagged fish were detected at the Custer Site, /

H!
I,

r 8located the portable base station approximately 40 km upstream of Pere
Marquette Lake to a site 2 km upstream of the Bowman Bridge public access site
that marks the downstream end of a reach that contains much of the spawning

habitat used by steelhead in the Pere Marquette River (Figure 3-1). The portable
bElse station remained at this location for the remainder of my study. I used a
Portable hand-held SRX-4OO receiver to verify the location of the radio-tagged

Steelhe ad downstream of the fall-placed base station.

8’00? Area
' estab l ished three study sections as a means of assessing movements of radio-
tagged fi sh to provide base-line data for comparison with a follow-up study after
the elect tic lamprey weir was in operation (Figure 3-4). Some of the key features

of “\9‘3‘3 atial arrangement were:

’Section A (from a point approximately 28 km downstream of the barrier to a
vo’mt 100 m downstream of the barrier) is unaffected by the barrier
operation. Thus, rates of fish movement and percent of fish passing
through this zone acted as a reference site for the duration of this study and

follow-up studies.

~Section 8 (within the 150 m-long detection zone including the barrier and

extending 100 m downstream from the barrier) provided a control during

23

 

years the barrier was not OPBrating and provided the data needed f0f

A.
estimating the impact of the barrier during years of barrier operation.

—Section C (approximately 28 km upstream of the barrier) provided a
reference during years when the barrier was not operating. During years of
barrier operation, data from this zone could be used to determine if fish
compensate movement rates if the barrier facility delays their migration. If

no compensation was observed, data from this zone could also serve as a
co ntrol.

' ”39d the time for passage from the release location of the radio-tagged fish to

the downstream antenna at the barrier to calculate movement rates through

SeCf‘O’I A. To estimate fish movement rate through the barrier section (Section

Bit Wfi‘asured the time required for a fish to move from the lower antenna to the

dime it arrived immediately upstream of the barrier. For the final movement rate

flfifimate (Section C), I used the time from the last record of the fish being

.“wedwtely upstream of the barrier to the time it was first detected at the

up {ream base station. A determination of the direction of movement was
5
ne éessary to assess movement rates through each section. A minimum of two

0‘)de from one base station (one record from each antenna) were necessary to
re

determine the direction of movement for a radio-tagged fish.

From these antennae arrangements, l was able to determine the dates when

24

 

"'1th 0‘ (he migration of steelhead and longnose suckers occurred, as well as

11::

the time required for transit between the fixed base stations and the proportion of
fish that completed each segment of the migration. Radio-tagged fish whose
direction of movement could not accurately be determined from the record of the
base station receivers were not used in the speed of movement through section

analyses.

Data Analysis
' Used 8 General Linear Model (SAS v. 8.01) to compare movement rates of
steelhe ad and longnose suckers between sections and years. I determined the
paSSage through each section as the number of radio-tagged fish that reached
the Upper limit of a section as a percent of those that moved upstream of the
\OWer (C1 ownstream) limit of the section. I used Fisher’s exact test (Zar 1984) to

6.0th Q the percent passage of radio-tagged steelhead in 1997 and 1998.

E’W’O’Vn ental monitoring

@\Q3“5e stream discharge and stream temperature data were likely to be
irh portant cues in the migratory process, I monitored stream discharge, and
s t (fiam temperature. Stream discharge data were obtained from the US.
Ga, ological Survey gauge station at located in the river at Scottville (Figure 3-1).
Stream temperature was monitored by means of an Onset Hobo electronic

therr—‘nograph located at Custer (Figure 3-1).

25

 

Results

Fish Capture

I captured 18 steelhead and implanted 14 with radio transmitters from 18

February to 6 April 1997 (Table 3—1). Six of the 14 steelhe ad (eight males-5i"

females) Were caught by hook-and-line and Bight We“: caugtfl“ tyke “6‘5 (Table
3'1). I radio-tagged 54 steelhead (26 hook‘aNG—linng fyke “a 039mm. and ‘6
malesz38 females) from fall 1997 to spring 1998. Ten of the 54 steamead were
captured and implanted from 24 October to 16 December 1997’ 300 a\\ta\\ ladio-
tagged fish were captured by hook-and-line. l captL. r e d the Spring (99%

steelhead from 6 February to 8 April.

The steelhead that I captured dUring the spring 1997 ranged er
long, and the steelhead captured during the fall 1997 a n d Sprin 'M 270 to BOO—mm

steelhead that was radio-tagged weighed 1.6 kg. and the largQ

The movements of longnose suckers were not evaluated dUri S6,"?
”9 19 '

97.
200 longnose suckers were captured in fyke nets from 24 February t 4 (Ora
o I Of
1998, and 33 longnose suCkerS Were radio tagged from 24 Faeraryt 4DP”
0 2
8
mm long and the- March

rr .
ranged from 1,0 to 2.0 kg (Table 3-1). Eighteen of the 33 longnose s we'ght

1998. The longnose suckers ranged from 420 to 550-

. . . er
implanted with radio transmitters were males and 15 were females (T S that I
a

 

ble 3-1).

26

——:

 

   

 

 

mmfiu:

 

 

 

 

 

 

 

O.N|O.—. Km.“ own-Q KQFHC
9 WV F k962m
2 3 mm 08 $895..
wmwr 9..th
.. 82.QO
S . . emu emu: 82-32
mm o E e mm 83% E E E 3.8-552,
erC
tn: 82.36
c a weed 2 838 So 3 M: 52 .9:an
\illfifllililmmé X
«Emu mais— omcme A9: mmcfi «E95 Emma... 2930 wflomnmEOtmm
mo.
II/
Edam Hag flag .74... .o .8252 29:3
i iii/illlilii
llllllli

ng comm bwmmfléfig 0.52, «m5 mflmEm» “Em
29: B .595: 9: mm =95 mm :m: 8538 Etc sci new E .63 SEE ms 0 a . |
w b E 3.3% ms em \

32 $953 .55 953322 men as so: 293% $892 Em 8252 .6 8e
. e5» :38 .F n o
- 3m...

 

Timing of Migration

Radio-tagged steelhead were recorded moving upstream through the Custer
reception area (Section B) from 23 March to 4 April 1997 (Figure 3-5), and were
recorded moving upstream through the Custer reception area from 6 January to
17 April 1998 (Figure 3-6). Although the 1998 movement occurred over alonger
period than 1997, much of the upstream movement occutted over a “(a period‘

from 25 March to 17 April 1998. Radio-tagged long nose suckers were recotded

moving upstream from 23 March to 1 0 April 1 998 (Figure 3-7).

Radio-tagged steelhead and longnose suckers were recorded at the Custer base

station when water temperatures we re greater than 35°C and Stream flows were

greater than 20.8 m3ls, for 1997 and 1998 (Figures 3-5, 3-6, and 3:7). in «997
and 1998, temperature and discharge data indicated that there Was a gener3\
correspondence between fish movement and in creasing stream flow and water

temperatures.

Speed of Movement Through Base Station Reception Areas
Section A
The mean number of days between release of marked fish and arrival at Custer
was 7.8 days (n = 8, s = 11.1 days), with a range of 0.5 to 33 d ays for steelhead
radio-tagged during spring 1997 (Table 3—2). The mean number of days between
release of marked fish and arrival at Custer W85 18 days (n = 26, s = 265 days ),

with a range of2 to 126 days, for the fall 1997 and 59"”9 1998 ’ad’O‘tagged

28

 

em

.6
m
E
w
H
O
M
w...
W

h
ram 9.5 05:30 Mummy-HQQONu u

v con 5.? 9..

66 LIB/V

:2

i

8

Vm:

9:.

mV..

 

Fm .i.

'

66MB]?
66”le

E9 . Emma
am 6:: 20$ 22%”? poor/Iowan oommflésg so .883: 9? m-m 059W
5 . 9e .
o So 28%
995 L oumo
co m... R N m
w .,. emlmm
W m ”v I/v ../v m m 8 L my. \v lr
iv Ir L .L I- l Ir U H \v 6 6 6
6 u 6 n 6 6 6 n 6 n 6 u i! to
+~
n: v
.. m
I l l \ \ L. w
I I I |I I I| I s \
/ \\ \ .fiOw
, , t l . 82882.3 .633
,. .. . X 26E Emefi - .. -
8385'
anf

 

. ms to
dog—#25. 521 9.63ng 3mm team
.062, Zfio some can 3:: 8:83 26: EVE

 

 

 

 

 

(Do) 91 "IBJadLUOJ 1612M pue Keplpelunoo peel-"991$ 1° 19C! 114 mg

29

 

ma
2 9.5m 05 @550 my

>

Slew MOH UJBans

__NU waE 5%.; Qfim

 

o. m m
6 w e
or.

8..
omittl.
e...

8..

co.

2%

0x
A 3.69 eaeoowcesuéoflw 889-069 625:: 9: .m.m
m 9:. . so
.m .mcmsocucmfiso 9: a 32%,;
. 00¢
:93
38
R m W Ah 9 Z Z Z L L I:
w m new w m w m m w m m a m m m
8 8 8 6 6 file a 6 6
s 9 8 8 8 e 8 e 8 8 8
n r - II
Il\|ll\ lllllllll Illlll \\l| I
If ll C \II
I \~ ’1‘ II
an r

 

domino—.5— ;mZW. 9635: 2mm 2.: co to
.662, 260 :85 new 3:: 8;me 26:

 

oSHmEQEmH 5am>>|
26¢ 88% u - -

vessel

.CWKQNE

Ewwzw

930E

~
<r N O ()1

fl
‘
I
fit
(0

n
I
oo

4_
I
O
‘—

0':
(0°) alnlmadura; 1618M pue flap/pawnoo peaqlaaus ;o JaqmnN

 

L
U
V-
v—

30

 

.3952? cozy, 9.53ng 2mm 2: co tongue
.. awe... ..___ s... 5.. 5......
$0: :5 @3562 3.22:: 9.: . n- a

>=mv

 

 

 

 

9:
as 6:; Bag 929» am
.5 one on . a o
5.. m we 9:8 95 56:0 0 m5 6 328228 m c, m 05m.
9: . cmum: .u_
5.? 05w cozamowe 3&0
V l l l I l 8 I Z
r v W. H m m u w u w u m m m w. m
u l l I I I / l I /
m m we. % % m 6 6 6 a m m e 6 m a N
w 8 e 3 9 3 8 8 8 n
o . . e 8 m
I iii - o q
m
S f m.
I W n
a
m
cm . a
S .. v e
w .. . I . . . . . . . . p
e y ( \ I I n\ \ I m
w 8 , x .. I . n
H— I; ~ .‘I 0 w
o .. . m.
M I! \ I l I I h I
. P
m .. .. s a
he CV l / u I l K II m M
S e . U
— p
om .. x . 929880» .325 .. or m.
a .. 25.“. Emmbw .. u I J
.. . memxozm owocmcool .. we W.
‘ ~~ d
om ._ f . m
4 1?
.. E m
a
\d
r b
n we

 

on.

31

 

Table 3-2. Number of fish arriving at base stations and time taken to reach base
station for steelhead in 1997 and 1 998 and longnose suckers In 1998.

 

 

VM/

 

 

 

 

 

Mean no. of Mean no. of
days from days from
release to Custer to

No. of fish Custer and Bowmane
No. of fish counted at range No. of fish at and {fife
implanted CuitL (Section AM
Spring 1997,
Steelhead
1 4 8 8 N 0t monitored
(0.5 - 33)
Fall 1 997 to
Spring 1998,
Steelhead 18
54 34
(2 - 126) 14 (1.928)
§pring 1998,
Longnose
Suckers
33 8 17 O
[_eL \f__/

 

steelhead. However, if the fish tagged in fall 1997 were excluded the me an

number of days between release of marked fish and arrival at C ”star (”3% 11
days (n = 24, s = 12] dayS). with a range of 2 to 51 days. The mean tin‘ f
e rom

release to arrival at Custer was not significantly different between the Spring

1997 and spring 1998 radio-tagged steelhead (F = 1 .12, P = 0-30).

Three of the 10 fish that l radio-tagged during the fall 1997 were recorded at the
temporary base station located downstream 0f Custer, TWO 0f the three fish
recorded at the temporary base station were later recorded Upstream at the

CUSter base station. Of the seven radio-tagged steelhead that were not recorded

32

 

 

at the temporary base station, five were not found in the river within 3 days after

radio tag implantation, and using a portable receiver, I observed two
anted

approximately 0.25 km upstream of Pere Marquette Lake 1 day after I impl

them with radio transmitters.

Section B
When steelhead arrived at the Custer station. they moved through the sectio“

 

quickly. The mean duration of passage through the 1 50 m section during 39mg
1997 was 7 minutes (n = 8. S = 5-9 minutes) for Upstream-bound fish and 7

minutes (n = 4, s = 6.0 minutes) for downstream-bou nd fish (kelts returning to the

 

lake after Spawning). Six of the eight fish that arrived at Custer Passed upstream
through the monitored section quickly (less than 1 day), and arrived within a '
day period from 23 March to 4 April. This suggests that the fish Were resp0“dmg

to similar cues to initiate upstream migration and they can make the passage

from near Pere Marquette Lake to Custer in one day or less.

The mean duration of passage for the fall 1997 and spring 1998 radio-tag S d
e

Steelhead was 32 minutes (n = 27, S = 72 minutes) for Upstream—bou“d fi§h, and

11 minutes (n = 9, s = 1 3 minutes) for downstream—bound fish. One 83%| head

took 11 hours and 12 minutes to move downstream through the (L‘Ut‘rtfi‘r reception

area. I excluded data from this fish in the calculation of downstream mOVement

time. because I believed that the longer duration was atypical Wh en compared to

the other nine fish that were moving downstream. There was No significant

33

 

 

difference (F = 0.84. P = 0.37). in the time of upstream passage between the

spring 1997 and fall 1997-spring 1 998 data,

The mean duration of passage for longnose suckers through the Custer

reception area was 20 minutes (n = 8, s = 16 minutes\ tor upstream-hoimd “Sh

and 90 minutes (n = 1) for downstream-bound fish. The mean numbel oi days
between release of tagged suckers and arrival at CUSter was 16.6 days i“ = 8, s

= 9.1 days), with a range of 2 to 24 d ays. There was no Significant difference in
the mean recorded time to pass thro ugh the 150 m lo ng Custer Station (F _____ 0.09,
P = 0.77), or in the mean time from release through the Custer Stati on (F = 2_46.

P = 0.12) between the radio-tagged suckers and steelhead.

Section C

The mean number of days between leaving Custer and arriving at the Bowman

Bridge base station was 9 days (n = 10. S = 9). With a range of 1 to 28 day

during 1998. The mean time radio-tagged fish moved through the 600 m 8

reception area of the base station was 22 minutes (n=7, s = 17 minutes). going
upstream and 186 minutes (n=1) in a downstream direction. No fa“ radiu\tagged
steelhead were recorded at the Bowman Bridge receiver Site- Radio-$99 ed
longnose suckers were not recorded at the Bowman Bridge base station during

1998.

 

 

Percent Passage Through Base Station Reception Areas

At least 570/0 (n = 14) of the radio-tagged steelhead passed upstream of the
C3uster barrier location in the spring 1997 and 63% (n = 54) of radio tagged fish
Passed upstream of the Custer location in the spring 1998 (Table 3-2). The
Percent passage of radio tagged fish through the Custer reception area (Section

B) was not significantly different between 1997 and 1998 (P = 0.22).

Dtiring 1998, there were seven radio-tagged steelhead recorded at Custer who’s

d" ecti on of travel I could not determine using only the Custer base station data.
HQWever, using the Bowman Bridge receiver data, I determined all but one of the
Steel/Ne ad that reached Custer moved upstream of Custer. One of the seven fish
was with in receiver range at Custer for 10 days. The receiver at the Bowman

3‘3“” recorded 26% of 54 radio-tagged steelhead (Table 32).

‘east 24% (n = 33) of the radio-tagged longnose suckers passed by the Custer
t
5(anion (Table 3-2). Ten of the 33 radio-tagged longnose suckers were presumed

dead after they had been repeatedly located at their release location for one

month or longer.

35

 

 

Discussion

Timing of Migration
The timing of migration initiation corresponded to increasing water temperature

and increasing stream flow for steelhead and longnose suckers. Shepard (1 972)
and Miller (1974) found water temperature and stream flow influenced the timing
0f steelhead migrations. Northcote (1962) found water temperature regulates the
uDstream movement of rainbow trout. Steelhead initiate upstream migrations
°Ver a range of temperatures and have been reported to spawn in water
temperatures from 5.0 to 125°C (Shepard 1972; Miller 1974; Beschta et al.
198‘7) - Another salmonid, the Atlantic salmon (Salmo salar) also initiates
migrations over a range of water temperatures and may enter rivers during
periods of increased flows (Power and McCleave 1980; Power 1981; Smith et al.
19941 Longnose sucker migrations intensified as a result of increasing

\%\\\‘{’\§ture in a Wisconsin stream (Bailey 1969), and Geen et al. (1965) found

at rising stream temperature may be associated with the onset of upstream

migration.

Because steelhead and longnose sucker migrations were influenced by water

temperature and stream flow, the shorter duration of upstream migration for
steelhead in 1997 than in 1998 was probably due to the colder water temperature
during the months of January and February. Water temperature was less than

35°C in January and February 1997, and water temperature exceeded 3.5°C for

periods in January and February 1998. Although the duration of upstream

36

 

 

migration differed from 1997 to 1 998, the peak of upstream migratory activity
appeared to be similar for the 1997 and 1998 field season. occurring during the

last two weeks of March and the first two weeks of April.

Rawn-tagged fish appeared to be more likely to move upstream as water
teTTIperature and stream flow increased. Leider (1985) suggested that the
pr Opensity to migrate upstream increases as the migration season progresses.
Later in the season, the increased temperature and flow effect was most
pr Onou nced in steelhead that migrated upstream of Custer in as little as 0.5 days,

a . . . .
n“ past the Bowman Bridge base station In as little as 5 days from release.

Per
can 1‘ Passage Through Base Station Reception Areas
RadiOwt .

Qgged fish moved quuckly upstream through the Custer and Bowman
“S‘Ntions. The short passage time through the base station reception areas
dwaited that the base stations were located in areas that did not impede

an
upstream passage and did not provide adequate spawning habitat for radio-

tagged steelhead and longnose suckers.

(__o rignose suckers were not as successful as steelhead at reaching the Custer
and Bowman base stations. It is possible that the spawning habitat for longnose
SUckers was downstream of the habitats used by steelhead, thereby accounting
for their absence at the Bowman Bridge receiver. In addition, at least 30% of the

(ongnose suckers implanted with radio tags did not survive long enough to

37

 

 

Migrate to Custer. The suckers may have suffered higher mortality or impairment
from the radio transmitters, which were larger, relative to the fish size than they
Were for steelhead. Winter (1996) suggests that fish should not be equipped with
transmitters that weigh more than 2% in air of the fish’s weight out of water. The
r adio tags used in my study weighed 1.8% of mean weight, and measured 1 7%
oi the mean length of the longnose suckers l implanted with radio tags. Even so,
the suckers that did reach Custer were moving almost as quickly as the

steelh ead. A smaller radio tag would probably reduce incidence of mortality from

the su rgical implant procedure.

M _ _ . . .
03’ of the steelhead that were radio-tagged during fall 1997 did not remain in
the P
Q re Marquette River. Whether this represents typical behavior for staging
elh
ste & Qd or a response to the transmitter implantation procedure cannot be

¢Q‘Q‘M\\ed. Winter (1976) found 89% of fall-run steelhead that were radio-

gged in Lake Superior streams left the streams following radio-tagging, and
{e

returned the following spring to spawn. However, the fact that fish tagged in
February to April rarely left the river suggests that steelhead in the river in

September to December are more prone to return to the lake.

Migrating steelhead in Pacific coast rivers are known to use holding or staging
areas that are also referred to as overwintering areas (Burger et al. 1983).
Hooton and Lirette (1986) observed immigrating steelhead that occupied a

heavily fished area within a western Canadian river for an extended period of

 

38

time 3 to 4 months prior to spawning. Little is known of the significance of the
Staging areas. it is possible that they may serve as a source of energy, such as
a location with an abundant food supply, or possibly a refuge from harsh winter

r lVer conditions such as icing.

Evaluation of Capture Methods
The combination of fyke nets and hook-and-line methodology were successful
and necessary in acquiring enough fish for the implantation of radio tags. Using
the m u ltiple net configuration to ensure successful steelhead capture, | blocked
as MUCh of the river as possible. i was careful to maintain complete blockage

f
mm ”W e surface to the bottom of the river in any location where the fyke net was

Placed -

E<&1\\§\gs were implanted in fish far downstream of the proposed barrier so that
ar‘ immediate responses of the fish to the surgery would be likely to abate by
the time they were ready to ascend the river. Because some radio-tagged fish
arrived at the Custer base station relatively quick (0.2 to 2 days), it is unlikely that

the surgical implant procedure adversely affected the steelhead’s ability to

mtg rate upstream.
Future Considerations

The telemetry data that I collected from 1997 to 1998 demonstrated that

steelhead and longnose suckers moved quickly upstream in the Pere Marquette

39

 

Rive, when they began their final ascent of the river for spawning. Furthermore,

b0th species moved quickly past the site where the electric barrier was
Constructed. These data allowed for an evaluation of how long the barrier and
fish passage structure delay movement for both species (Snell 2001). Likewise,
the percent of fish traversing each section will be used as a measure of the

number of fish unable to traverse the barrier/fish ladder and “balking” or dropping

back out of the section, or being caught by anglers.

4O

 

CHAPTER 4
A Model of Steelhead Movement in Two Lake Michigan Tributaries in Relation to

Water Temperature

41

Abstract

i used movement data from two Lake Michigan tributaries to develop a new
approach for analyzing upstream adult steelhead migration. My data included 28
radio—tagged steelhead (Oncorhynchus mykiss) on the Pere Marquette River and
a larger (5,876 to 10,083 steelhead), multi-year (1993-1999) data set of camera
recorded steelhead passage through a fishway on the St. Joseph River. My
model used a temperature-based movement rule developed from the data to
quantitatively predict the probability of upstream movement. Exponential, logistic
and Power functions were evaluated as a means to express the probability of
movement. Of these, the power function resulted in the closest fit between

observed and predicted movement. Probability of movement increased over

increasing water temperatures above a movement-threshold water temperature-

Stream flow was incorporated into the temperature-based movement (TBM)

model, but did not add substantially to the model’s ability to describe the

migratory behavior of steelhead in the Pere Marquette and St. Joesph Rivers.

The TBM modeling approach is broadly applicable and transferable to Other

Great Lakes tributaries, and may work well for describing the migratory behavior
of other species whose migration is also dependent on water temperatures
Furthermore, the TBM model could be used to predict percent passage by a fixed
location, forecast run size early in the migration season, and aid in the timing of

operation of sea lamprey (Petromyzon man'nus) control structures such as

electric migration barriers.

42

Introduction

Several approaches have commonly been used to characterize fish movements,
One approach is to collect sequential descriptions of the location of individual fish
while taking note of environmental conditions such as water temperature, pH,
dissolved oxygen, stream flow and others (Doerzbacher 1980; Schulz and Berg
1992; Workman 1994). Descriptive statistics such as the mean and range are
then used to characterize observed movement patterns of fish. The information
gained from this approach tends to be fish specific and largely descriptive in
nature. Another approach is to determine the timing of movement relative to
environmental thresholds (Geen et al. 1966; Bailey 1969; Power 1981; Jonsson
1991). Environmental thresholds can be thought of as cues (e.g., water
temperature, dissolved oxygen level, etc.), above which fish movement is
observed. Typically, little or no movement is associated with levels below a
minimum threshold point and movements are observed at the threshold point and
increase until a maximum movement rate is reached. Other approaches make
use of regression analyses, analysis of variance (ANOVA), and multivariate
analyses to identify environmental cues and their relation to fish movement
(Clapp et al. 1990; Trepanier et al. 1996; Giorgi et al. 1997; White and Knights
1 997). The information gained from these studies partially describes fish
mOVement by linking exogenous (environmental) cues to movement, or
identifying a range of conditions where many fish are likely to move. However,

these ana'Yses generally provide a static picture of fish movement, and do not

43

 

provide a base for predictive models using the range of one or many

environmental cues.

Because of the important role migration plays in fish population dynamics,
improvements to our understanding are critical to the management of migratory
fishes. One aspect of fish migratory behavior that is of particular interest is the
timing of migration and its relation to environmental cues. Understanding
migratory behavior thus may help fishery biologists better manage these species

by limiting exposure to mortality sources.

My objective is to develop an alternative approach for describing fish migratory
behavior that is based on a dynamic model that treats migratory behavior in a
probabilistic manner. I illustrate the application of this modeling approach using
a case study of steelhead (Oncorhynchus mykiss) in two Lake Michigan
tributaries. In this model, we gain a better understanding of how water
temperature and stream flow affect the probability of upstream movement of
steelhead in these streams. I also evaluate how the model performs for two
different types of data collection; radio telemetry passage data on the Pere

Marquette River and fishway passage observational data on the St. Joseph

River.

Steelhead were chosen because of their high value to anglers and resultant need

for management, and their well known migratory behavior through previous

44

studies. The majority of previous studies have focused on the energetic cost of
migration (Hinch and Rand 1998), migration timing between separate runs of the
same species in the same river (Burger et al. 1984), factors that influence the
migration ofjuveniles (Northcote 1962; Muir et al. 1994; Zabel et al. 1998), and
factors that influence upstream migrations of adults (Shepard 1972; Miller 1974;
Jensen et al. 1986; Trepanier et al. 1996). My study focuses on adult steelhead

movement behavior over a range of exogenous cues.

Many fish migrate up Great Lakes tributaries for the purpose of reproduction, and
initiate upstream migrations to spawning areas based upon exogenous and
endogenous (physiological) cues. While endogenous cues are difficult to identify
and measure in the field, exogenous cues are more readily identified. Water
temperature and stream flow are the most frequently cited exogenous cues that
initiate upstream migration (Peters et al. 1973; Miller 1974; Power and McCleave
1 980; Power 1981; Jensen et al. 1986; Trepanier et al. 1996). In my model, I test

the influence of both of these factors.

45

Methods
Migration Data Sources
Methods of data collection differed on the Pere Marquette and St. Joseph Rivers,
On the Pere Marquette River I radio-tagged steelhead, whereas, a camera was
used to observe fish passage on the St. Joseph River. F ifty-four steelhead were
implanted with radio transmitters from 15 October 1997 to 11 April 1998 in the
lower Pere Marquette River. Fyke nets and angling were used to capture the
fish. Steelhead larger than one kilogram were anesthetized in a tank filled with
60 mglL tricaine methanosulfate (MS-222) dissolved in river water, and a LOTEK
Engineering, Inc. CFRT-7A digitally encoded radio transmitter was surgically
implanted into the peritoneal cavity (Winter 1996). The gills of the fish were
irrigated with river water during the surgical procedure. The radio tags measured
16.0 mm in diameter by 83.0 mm long, and weighed 29 gm in air and 12.8 gm in
water. The radio tags were rated to last for 282 days at a 5-second burst rate.
The incisions were closed with 000 gut suture, and the radio-tagged fish was
transferred to another tub with fresh river water where the fish remained until it
maintained an upright posture within the tub. The radio-tagged fish was released
as close to the point of capture as possible. Fish captured in the fyke nets were

released approximately 200 m upstream of the nets.
Radio tagged fish were continually monitored for passage at a fixed location at

the CUSter public access site approximately 29 km upstream of Pere Marquette

lake, the dOwnstream terminus of the Pere Marquette River (Figure 2-1). The

46

 

Custer public access site is the location of an electric sea lamprey (Petra/by?”
man'nus) barrier that is intended to block the spawning migrations of sea

lamprey, and maintain the integrity of salmonine spawning migrations.

Twenty-eight of the 54 radio-tagged steelhead were recorded moving upstream
at the Custer monitoring site and were subsequently used in this study. Of the
26 fish not recorded at Custer, seven were recorded moving out of the Pere
Marquette River, and the status of the remainder was unknown. Stream
temperature was monitored at 90-minute intervals using an Onset HOBO® Temp
Logger at the Custer site for the purpose of movement analyses. Stream flow
was monitored hourly at a United States Geological Survey flow recording station

located in the river in Scottville, Michigan (Figure 2-1).

The Michigan Department of Natural Resources provided me with seven years
(1 993 to 1999) of steelhead passage data through the fish ladder at Berrien
Springs, on the St. Joseph River, Michigan (Figure 2-1). Fish passage was
continually recorded over 24 hours using a camera mounted in a viewing window
Of the fishway, typically beginning on 1 March and continuing through April in
most years except in 1999, when fish passage was monitored for a one-week

periOd from 12 through 19 February and resumed again on 1 March, and 1998
80d 1997 when monitoring began on 23 and 18 February, respectively. Because
Ofthe gap in data collection during the month of February 1999, I only used data

that had been collected beginning on 1 March 1999 for purposes of model

47

 

simulation. Water temperatures were recorded once daily at 8:00 am, 1-2 kn;
below the Berrien Springs fish ladder. Annual fish migrations in the St. Joseph
River data ranged from 5,876 steelhead during 1994 to 10,083 steelhead in

1997.

Model Development
Upstream migration begins when exogenous and endogenous conditions are
appropriate to stimulate the movement of a few or a group of fish from a source
location such as a lake or ocean, into and upstream. The model portrays the
passage of fish upstream beyond a detection site to spawning locations based
upon changes in water temperature and stream flow, by drawing daily from a
population of fish from the source. The source population of fish (Nt) includes
individuals that will migrate upstream to the spawning area, but does not include
the portion that will remain in the lake and not participate in spawning migrations
during the current year. Once upstream migration begins, the source population
continually declines until there are no more fish left to move upstream. The
n umber of fish available to move upstream each successive day (NM) is
determined from the number of fish available to move the previous day (Ni)
miWS the number of fish passing the detection site. The number of fish passing
the detection site varies daily, and is expressed as the number of fish available to
move (Ni) times a probability of movement (PM,). For the purposes of this model,
the probability of movement is based upon a function of water temperature alone,

or water te mperature and stream flow.

48

 

l
The initial model development included water temperature as the exogeno Us 1‘ ‘
factor upon which movement was based. A time step of one day, which Was in
accordance with the data collected from the Pere Marquette River study, was
used for the simulation. The differential equation representing movement was
solved using a fourth order Runga-Kutta integration (Press et al. 1992). Three
functions were evaluated as a means to express the relationship between
temperature and the probability of upstream movement (PM). These functions
represented the general concept that as water temperature increases, fish are
more likely to migrate upstream. An assumption of my model is that steelhead
will migrate before stream temperatures are warm enough to inhibit upstream
spawning migrations. Therefore, the functions I used to depict the probability of
movement do not account for a decreasing probability of movement as water

temperature increases. The functions explored were:

Exponential function: PM = ae””"")
Power function where PM = a(T - 11),)

1
[1 + e(—b(T — h) + a)]

Logistic function where PM =

 

a = rate variable
b = rate variable
T = mean daily water temperature

h = Minimum water temperature threshold for steelhead movement.

49

 

The Solver feature in Excel was used to minimize the sum of sq [Jared differences
between the observed and predicted number moving (RSS: res idua/ sum of
squares) by varying the natural log (Loge) values of h, and the a and 1)

parameters in the migration functions. By varying the natural log values of my
model parameters, the model parameters were constrained to positive values.
Best-fitting parameter values were determined by performing a non-linear search
to minimize RSS. The non-linear searCh was initiated with several different
values to ensure that the search had achieved a global minimum. The functions
tested were compared for goodness of fit by evaluating the sum of squared

differences between the observed and predicted number of steelhead moving.

After selecting the best fitting function, I examined how the effect of watei
temperature varied among years for the St. Joseph river data using an F: 433‘
based on extra sum of squares (Neter and Wasserman 1974). A TBM modal
with a set of parameters representing each year (full model) from 1993 to 1999
was compared with a model where one set of parameters was used to represent

the data among all years (reduced mode|),

Approximate Standard errors for the model parameters were estimated using a
likelihood approach (Ratkowski 1983). The standard errors were approximate

because I aSSUmed that my data were normally distributed. The concentrated

50

 

 

g7:

 

likelihood for each parameter was calculated according to meth ods described ,-,,

Seber and Wild (1989):

 

R
-Concentrated [Loge Likelihood] = 231n[2”USS]+%

u = n — k degrees of freedom

n = sample size
k = number of model parameters
RSS = Residual sum of squares.

Standard errors for parameters we re calculated by perturbing the “best fit”
Parameter estimates by 1 percent, and determining the change in the
concentrated likelihood. \ estimated the standard error of each parameter
estimate (i.e., W) from the variance-covariance matrix for the parameter

estimates (Ratkowsky 1983):

S: cov(h,a) cov(h,b)
Variance-Covariance matrix = cov(h,a) sf, cov(a,b)

cov(h, b) cov(a, b) 5,?
Where,

h = Minimum water temperature threshold for steelhead movement

a = rate variable

b = r ate variable.

51

 

 

 

."~

l derived the variance—covariance matrix from the inverse of the information

matrix (Ratkowsky 1983):

 

 

62L 62L 3:5
62h ahaa ahab
. . 62L 62L 62 L
Information matrix = 2 “’7;
ahaa a a aaza
62L 62L 9};
2b

 

ahéb 6616!) a

l estimated the second derivatives numerically for the h, a and b parameters from

the following equation:

6L
62L _ :3;
(32x 0.5(x‘ — X)

x = best fitting value for parameter estimate h, a, or b from the TBM model

x1: The value of each parameter (h, a, or b) when it is perturbed one percent

above the best fitting estimate

a_L=Ll—L

6x xl —- x
L1 = concentrated likelihood value when one model parameter (h, a, or b) is

 

perturbed one percent above the best fitting estimate

L = concentrated likelihood value for best fitting parameter values.

52

 

 

 

The second derivatives for the co-varying parameters were derived from the

general equation:

62L _ Lx+y — L

63va y (xl — x) + (yr - y)

 

x = best fitting value for one parameter BSllm ate h, a, or b from the TBM model
x1 = the value of one parameter (h, a, 0r b) When it is perturbed one percent
above the best fitting estimate (x)
y = best fitting value for one of the tWO remaining parameter estimates h, a, or b
from the TBM model that were not used to determine the (x) value in this
equation

Y1 = the value of one of the two remaining parameters (h, a, or b from the T BM
model) perturbed one percent above the best fitting estimate (y) that was “0‘
used determine the (x) value in this equation.

Lx+y= the concentrated likelihood value when two parameters (h, a, or b from the
TBM model) are perturbed one percent above the best fitting values.

L = concentrated likelihood value for best fitting parameter values.

The effects of Stream flow on migratory behavior were incorporated into the
temperature-based movement model (hereafter referred to as TBM model),
following initial evaluations of the effect of water temperature alone. In order to
determine the qualitative relationship between flow and movement, I examined

plots of the residuals from the TBM versus flow. For these models, three

53

 

 

 

 

of

str

ll0'

flo

MI

I .
functions were evaluated to describe the probability of upstream movement (pM)

as a combination of stream flow and water temperature. These functions were:

Linear - Power function where P M = (60’) + d)(a(T —- ’0")

Power - Power function where PM = (C(F)d)(a(T ‘ 10b )

Logistic ~ Power function where PM = (

WWW - ’0”)
c = rate variable

d = rate variable

F = stream flow

a = rate variable

b = rate variable

T = mean daily stream temperature

h = Minimum water temperature threshold for steelhead moveme ht.

Because stream flow is affected by the size of the watershed, stream channel
width, streambed gradient and other factors, the effect of flow on the pr0bability
of movement is likely to vary among watersheds. Therefore, I represented
stream flow in the TBM model using the relative deviation from the mean daily
flow(X — E/ )T , where X represents flow and J? represents the mean seasonal
flow. Stream flow data were collected from a United States Geologic Survey
(USGS) monitoring station located approximately 8 km downstream of the Custer

site, near the Scottville public access site on the Pere Marquette River, and were

54

 

A

 

recorded at a USGS gauging station located approximately 71- 6 km upstream of
Berrien Springs in Niles, Michigan on the St. Joseph River (Figure 2-1). The

Solver feature in Excel was used again to determine the best fitting parameter

 

values, and was determined by the smallest sum of squared differences between

the observed and predicted movement.

55

 

 

 

 

Results

The number of fish passing by the Custer monitoring site each day showeda
general correspondence to stream flow and water temperature (Figure 4-1).

Incidence of higher numbers of fish passing generally occurred when stream flow
and water temperature were increasing. However, it is difficult to quantitatively
describe movement behavior based on stream flow and water temperature using
this figure alone. in the St. Joseph River. few tiSh migrated upstream early in the
migration period when water temperatures were low (Figure 4-2). As water
temperatures increased, there we re many tiSh available to migrate and many did.
Later in the migrating season whe n the water temperatures continued to increase
and there was presumably a stronger propensity to migrate, fewer fish were

moving upstream because there were fewer fish available to move.

Water Temperature
Three functions were compared for their ability to represent the relation ship
between water temperature and the probability of movement. The pOWer
function provided the best fit for the Pere Marquette River telemetry data, and for
the St. Joseph River count data for 6 of 7 years, except for 1995 where the
exponential function provided the best fit (Table 4-1). The average sum of
squared residuals for the power function was substantially smaller than for the
exponential and logistic functions for the St. Joseph River data. Using the 1996
St. Joseph River data as a typical example, all three functions predicted trends of

fish movement that generally corresponded to periods when steelhead were

56

 

enema umum; COOS >=m

 

 

 

 

 

 

 

c'v
,_

 

 

a.
r.

3 e.

.95 5589 $330 9.: on ooEooo. 3:: 28 .02 Same (combo EoEoEoo ochmmoE new 0:3:on S E85238
3:3 26: Emobm mow: 9.: so 3c: 85% inc: BefrootonEocv 003me on comma 595:2 coax ozoaoem—z
Ema 05 co mew c2382 56:0 215 $ch 9.6m 9:on 839-231 ._.-v 2:9...
as

wvwmmmmnmmmmmmmmr m
mammzbmnpmwmmunpw w
W%%%%%%%%%%%%%%%% m.

w

. t slut/u so w

. _ m

m _. .. n
...m m

. m

e

a .. a. w .7.

my . N
e . m
d.

w o m
H w.
m 3?. seasoneotam; .w s.
\w' 30-E Emmhwwl I I I fi m
I... u.
6 osmmmm 9&5QO 3952' o_. ”a.
. .. p

d

0.0. ‘ «W
S

m-

.D

O

n

w

m

 

 

m“

59.6.2 .52
.1 common .6 m5 5 academies poems

 

 

GEM: UOHCDOO
9 532 s 82 mesa 8% $828 a... assesses 32.8.... a page 2: .3 can.
as 2323.28. .325
mos/crate e e a a e N o
. . - . r . o
o o o o a 141i
o o o o o o o o 2:
s s w . o N
Q Q .oom n
. a
o m
” . foom m. mm
0 m.
o a.
fiOOV w.
o B
p
0
38 m
u
w.
.08
i
o .02
o _v

 

. cow

 

Observed to move (Figure 4-3) - The observed steelhead move the,” was Var/ab/e
f mm day to day, and all three functions did not consistently pred/C, to the extreme
ma,rirnum and minimum numbers associated with the observed movement The
poWer function was the best predictor of movement early in the migration period
When there was less week-to-week variation in movement, and it consistently did
the best job of apprOXimating the observed data. Based on these results, l used
the power function for the remaining analyses.
Table 4-1: The sum of squared residual (RSS) values for the power, logistic and
exponential form of the temperatu re-based movement model for the Pere

Marquette (1998) and St. Joseph (1993 to 1999) Rivers, MiChigan. A low value
indicates the best fit for that form of the model and is indicated by an asterisk for

 

 

 

 

 

 

 

 

 

each year.

D S r P RSS 4

ata ource Year ower LO istic EX nential
Pere Marquette River 1 998 21* xi 23 pg/fi.
St. Joseph River 1 993 232,41 1* 262,736 290,192

1 994 323,337* 339,036 340,647
1 995 480,621 470,424 470,114*
1 996 369,816* 482,955 518,423

1997 1,163,361* 1,284,841 1 ,326,333

1998 396,91 8* 607,810 574,139

1999 532.58? 694,935 694,935
W WWW

 

The minimum temperature threshold for movement in the St. Joseph River varied
from 00 °C in 1995 to 5.1 °C in 1996. Across years, the minimum temperature
threshold for movement averaged 3.2 °C (Table 4-2). The lowest temperature
t lwreshold Value (00 °C) occurred in 1995 where the logistic (3.4 °C) and the

exponential (39 °C) functions produced a smaller sum of squares than the power

59

 

 

700 i
600 ‘
500 ‘

400'
300'
200

100‘

—— Power
-—Observed

 

 

 

 

600‘

 

 

 

 

 

500
400
300
200
100

N umber of Steelhead Counted

600
500 '
400 '
300 ‘
200 ‘
100 ‘

 

 

—. Exponential
”Observed

‘u

 

--.-‘~

 

-- Logistic
"' Observed

—L

 

--‘

 

 

 

3/1/96
3/8/96

I

l

l

l
<0 “3 CD <0 co co co
0) Q O) O) a) o: 0')
\ \ \ \ \ \
LO N 03 Q N a) co
: Q Q v S 2 fi
<0 0'3 00 sr sr <1-

Date

Figure 4-3. plot of Power, Exponential and Logistic models of temperature-based
movement and the observed steelhead movement data in the St, Joseph River,

60

 

 

‘2-

Table 42. Parameter estimates (i 1 standard error) for the power funCfiOn 9‘
Version of the Temperature-Based Movement model (TBM) for the Pere Z
Marquette River 1998 and St. Joseph River, Michigan 1993 to 1999. The "

parameter values and approximate standard error estimates for the reduced
model (combined 1993 to 1999 St. Joseph River data) that was used in the F tegt
to compare among years for the St. Joseph River data are also included.

¥

 

Model Parameters Estimates and Standard EECE‘
\

 

\TBM Model Year TempThreshold (°C) a b
P... N. River 1998 3.2 i 0.23 0.016 i 0.0028 0.79 i 0.0?
St- Joseph River 1993 1.6 i 0.17 0.001 i 0.0001 2.43 i 0.031
1994 4.4 i 0.07 0.028 i 0.0029 0.67 :t 0.069
1995 0.0 i 0.00 0.001 i 0.0001 2.35 i 0.044
1996 6.1 i 0.03 0.053 i 0.0045 0.34 i 0.068
1997 3.2 i 0.24 0.009 i 0.0012 1.28 i- 0.090
1998 5.1 i 0.15 0.018 i 0.0016 0.98 i 0.040
1999 6.1 i 0.01 0.086 i 0.0067 0.40 i 0.037
Red uced Model 3.7 i 0.08 0.011 i 0.0004 1.25 i 0.025

3.6359 (1993-99)

“1th i Qn-based TBM model (Table 4-1). The intercept value for the power
99‘“: i on (a) varied substantially ranging from 0.001 in 1995 to 0.086 in 1996
(Tat) I G 4-2). The exponent value for the power function (b) also varied
subgtantially from 0.34 in 1996 to 2.43 in 1993. High values of b were generally
associated with low temperature thresholds and low intercept values (Table 4-2).
Despite the variation in parameter estimates, the predicted probability of
movement as a function of water temperature for the St. Joseph River was

similar most years except 1993 when the probability of movement rose sharply

with water temperatures in excess of 10 °C (Figure 4-4).

61

 

 

L$8 m mew: 832,9 26

cm

 

AOL 239.359... been;

 

m: m: 3 we or m o o
. . _ . l . . Ens
HO...

0 .
lllllllllflflI-Iimfiwn .9 .Po
... IIIW«*W.”.

I
n... 6« + .3
.“.««. X“ .7
<< x +
d XX + .
x + Imo
x +
xx + m8: .
+ mam—.I .vo
3m:
+ 82: .
+ 88x .mo
+ some.
33+
+ .3.
+
+
:0
5d

 

.89 0. mm? memos» E9.— omsom=oo memo So: awning—2 cozm common em 65 hoe cozoca
UoE :o 889 mot—6 EoEo>oE so >.___omno.a 839939889 65:06.1 .Tv 9:9“.

weansdn tuawerw ,io Miiqeqmd

62

Acompaf'lson of the RSS for the full TBM model (RSS = 3,549,146) and the

reduced TBM model (RSS = 4,336,051) indicated that the parameter estimates

for the power function TBM model varied in the St. Joseph River annually (F 18.41 s
= 1 .57, F Care: 5.12, P < 0.001). The minimum temperature threshold was 3.7 °Q

for the reduced model from the F test years 1993-99 (Table 4-2).

We fer Temperature and Stream Flow
in both the Pere Marquette and St. Joseph Rivers there was generally no
r GI ati onship between residual errors and flow from the TBM model. This result is
\\\u$t rated by the results for 1998, which was a typical year (Figure 4-5). The
Per r * power, logistic * power, and linear * power stream flow-water
temp erature models had the same sum of squared residual values for the Pere
Mar q u ette River data and did not result in a substantial improvement in model fit
«om D ower function TBM model (Tables 4—1 and 4-3). The average sum of
squa red residuals for the three stream flow-water temperature functions indicate
(halt h e logistic * power function was the best fitting function and was only a
S‘\\Q)\\|y better fit than the power * power and logistic * power functions for the St.
Joseph River data (Table 4—3). The parameter estimates of the best fitting
stream flow-water temperature function (logistic * power) varied from year to
year, indicating no consistent effect of flow on movement. Based on these
results, additional analyses on the effects of stream flow and water temperature

on steelhead migrations were not further considered.

63

 

 

 

 

 

 

2.0
o
c
1.5 ' A
o
o
E 1.0 ‘
g 0 °
g . ‘ O .0
$3 0.5 ' ° 0 ° ’ o .
U .
a:
a.) 0'” -------- W7».- -‘- —————————————————
E o ”a. o
5%; ’ ’3‘. ’ o $
“:3 -0.54 ° 0 . ’ ° ,
3 o
-2
cu: 1 0‘ . '
é: ' .
‘C,
=
g -1.5 ‘ A ‘ 7
ed: 300 ' 1.4 2.8 4.2 5.6 7.0
8
a o
o 200 4 o B
E 0
Q: .100 ' . o . ’
m 9 9. 0
g . .9 °. . O. ..
w 0' _________ .‘_'_,.:,,._.._ _____ m. __ooaoo.-—¢.-___
)3 0 .°. 9 o o .
9 ° . ° .
to. .9 o
9 -100 .
c6 0
:3
‘U 0
E ‘
m -200 . ’
-300 ' ' ' ' '
70.8 85.0 99.1 113.3 127.4

Stream Flow (m3ls)

Figure 4-5. The residual plots of the difference between observed and predicted
movement using a temperature-based movement model and stream flow for data
from the Pere Marquette (A) and St. Joseph Rivers (B), Michigan, 1998.

64

 

 

Table 4—3. The sum of squared residual (RSS) values for the power, logistic and
linear function flow-temperature-based movement models, using relative
deviation from mean daily flow ((X — XV}? ) to represent flow for the Pere
Marquette (1998) and St. Joseph (1993 to 1999) Rivers, Michigan. A power
function was used to describe the temperature portion of the stream flow * water
tern perature-based function. Power, logistic and linear refer to the flow portion or
function used to describe steelhead movement. A low value indicates the best
fit for that form of the model and is indicated by an asterisk for each year.

¥

 

 

 

RSS
¥ Data Source Year Power Logistic Linear
Pe re Marquette River 1998 21* 21* 21*
St. .Joseph River 1993 232,411* 232,411* 232,411*
1994 297,682 284,663* 323,337
1995 457,889* 459,091 479,475
1996 396,238 369,816* 369,816*
1997 1,163,361 1,129,662* 1,134,143
1998 396,198* 396,853 396,918
St 1999 582,682 581 ,180* 582,682
Mseph River Mean 503,780 493,382 502,683

 

65

 

Discussion
Water Temperature
i found the power function-based TBM model was the best fitting model for
representing the effect of water temperature on adult steelhead upstream
mig ration in the Pere Marquette and for six of the seven years of the St. Joseph
River data. As such, the TBM model serves as a quantitative representation of
the migratory behavior of steelhead in relation to water temperature. My model
ind icated increasing probabilities of upstream movement over a range of water

tem peratures above the minimum temperature threshold for movement.

The effect of water temperature on fish migration is well studied and varies
”770 '19 systems and species (Banks 1969; Miller 1974; Leggett 1977; Jensen et
al. 1 986; Jonsson 1991; Lucas and Batley 1996; White and Knights 1997).
See I head movements have been linked to water temperature during all phases
0”“ e ir life cycle, including upstream spawning migration of adults, downstream
WOVQ ments of juveniles, and distribution of steelhead during their residence in
waGreat Lakes (Northcote 1962; Banks 1969; Shepard 1972; Haynes et al.
1986; Giorgi et al. 1997). While the upstream movement of adults has been
linked to water temperature in some systems, movement has not typically been
Viewed as occurring over a range of temperatures (Shepard 1972; Miller 1974;
Leggett 1977; Jonsson 1991). However, Power (1981) found adult Atlantic
salmon (Salmo salar) to migrate within a range of temperatures, and also

identified a minimum temperature threshold of movement for smolts.

66

 

 

_‘fi

 

The relation between movement and water temperatures was similar betWeen
the Pere Marquette River data and the larger data set from the St. Joseph River-
There was an increasing probability of movement associated with increasing
Water temperatures for steelhead in both systems. The TBM model predicted
general trends of movement during periods of observed movement for both
rivers. Despite the different data types (radio telemetry and camera count data),
and the physical separation (200 km of Lake Michigan shoreline) of the two
rive rs, the application of the TBM modeling approach to the Pere Marquette and
St. J oseph Rivers suggests the possible transferability of this modeling approach

to other river systems in the Great Lakes for spring-run steelhead.

W“ i ' e the TBM model predicts the temperature-based migration of steelhead,
the: we are several factors that were not incorporated into the model that may
“mu 6 nce steelhead migrations. The TBM model did not generally depict the
\3'9 Q day-to-day variability observed in movement for both rivers. Group
‘fievior, where several or many steelhead migrate together upstream at one
time may account for the day-to-day variability. Other factors such as
neurological and physiological interactions may also account for daily variation in
migrations (Leggett 1977; Jonsson 1991). Lorz and Northcote (1965) found that
onshore winds and light intensity stimulated river entry by kokanee salmon
(Oncorhynchus nerka). The winds helped disperse creek odor along the shore,

congregate greater numbers of fish in onshore areas. Spending more time in the

67

 

. . r1 l .
onshore areas increased the p robablllty of the kokanee salmo °Cating the ”2,6,5

f - - .
Other studies suggest high flow or the maturity of the fish may aCIIItate rlver

entry (Miller 1974', Smith et al. ’I 994).

- St.

. 3 to 1 999 In the
erature varied amon rs 199
The effect of water temp g yea the ye ar—tO—

a . y tnfluence
Joseph River. Other factors such 8 ChUrnal period the . .
emead migration.

3
year Variability Of the effect 01 Wat er temperature on 5‘

5 and exogenous faCtOrs

There a re many possible combinations of endog 800

d. The TBM
that may stimulate the upstream migration of ste elhea model

. - 'nflu
demonstrates that we can simplify the complex rt)! Of ' Slices 0n Upstream
migration by viewing the effect of one exogenOl—I 5 factOf (Wat
their migratory behavior expressed as an incfeas ' ”9 p,

b b
. . C II/
migration with increasing water temperatures on e Wate Olaf Upstream
,

above the temperature threshold for movement Prev/bu mpe’atures increase

s 8t
he exis ent based 0 ”die _ . - d
t tence of a threshold for movem I) ate” 8 have , d entifie
salmonines (Menzies 1939; Power 1981: Jonsson 199,. "'perature so ,

Je
s
The water temperature threshold serves as a marker betw en et al. 1 a 86).

- -. Qn l'ttl

migration and an increasing probability of steelhead to migh ‘ e 0' “<3

. Qte upstream _

increasing Water temperatures. lt is pOSSible that the temp erth

. . "ature threshol .

greater barrier to rhigration earlier in the migratory Season d Is a
en water

temperatures to be colder for longer periods of “m
are like” . than late,- i

eason whe erature may dr0p to or below t the
s n the water temp he threshOId f
0r

68

 

 

relatively short durations. My TBM model is unique from other, °Vement SIM/£95

- - M
by quantitatively predicting migration behavior over a range of ater

I . ‘ ' 'o n t r
temperatures and combining the prediction With an identificatl of a we 3

- - m ovement
temperature movement threshold, rather than the qualitat‘Ve

description -

or of other fish that

. _ aha“
The TBM model may work to desar’be the migratory b ts on
their moveme“ water

move int 'vers for re roductive PUfposes and ti r119 . _
O r l p migrate into Great

rev
temperature and other factors. For example, see lamp

ratuie as a mi r
Lakes tributaries for spawning, and use water te rn e 9 atory Cue

. red
Woman et al. 1980). Sea lamprey have Sigmfic antly uced salmo
resul . .
ted In i

"tensiVe control

nine game
5935\38 in the Great Lakes and their impact has

efiorts to reduce their abundance (Smith and Tibbies 7980)
- lfa 7' '
BM mode‘ ‘5

developed for sea lampre)’. it 00““ serve to identify the I

. - [071};
percent passage of sea lamprey into rivers to aid the o 9 ofmigration and

p6,-

structures such as the electric lamprey barriers located 0
’7 t

River and Jordan Rivers, Michigan (SWink 1 999). e Pere Maq uette

Water Temperature and Stream Flow

The combination of stream flow and water temperature diq
t\ot substal'itiQ”y

improve the fit of t TBM modei. WhiCh W35 CODSistent With .
he an examina tic/7
of
eam flow. 0
991 su mi ration may be a response to hose
(1 ) Qgests Ki stream 9 ‘5‘ Combinatio n

0f

the residual distribution of the TBM model compared with Str

69

 

 

 

“‘-

stream flow and water temperatu re fluctuations. Stream flow and water
temperature are important in upstream migration of Atlantic salmon (POW/er 1 981 ;
Jensen et al. 1986). Stream flow has been linked to migratory behavior in other
salmonid populations (Ellis 1962; Alabaster 1970; Shepard 1972; Power and
McCleave 1980; Smith et al. 1994). Both rivers typically do not exhibit drastic
fluctuations in stream flow during steelhead migration that are likely to affect their
ability to successfully migrate and subsequently affect their migratory behavior.
Therefore, the range inflow and flow fluctuation may have been insufficient to
detect how migration would vary in response to flow in streams exhibiting greater

Va riation in flow.

St"earn flow is characterized by many factors including, watershed size,
‘57P .
§ ambed morphometry, surrounding topography, weather patterns,
§Qundwater input and others (Bras 1990). Hellawell et al. (1974) found that
stream flow and water temperature were secondary to the effect of time-of-year
or season on migratory behavior. Shepard (1972) found that stream flow and

water temperature are important in steelhead migration, but their influence varies

from system to system. Trepanier et al. (1996) found stream flow to be important

 

and water temperature to have little effect on the migratory movement of Atlantic
salmon. Stream flow is likely to be a greater factor in steelhead migrations in
watersheds that experience drastic changes in flow that may promote movement
by creating an avenue of passage over barriers that are typically impassable at

low water periods, or inhibit movement during high flows that are strong enough

70

 

 

s“

to prohibit upstream movement. In rivers with small fluctuations in water levels,
water temperature is important in stimulating upstream migration (Jonsson 1 991) ‘
The response of steelhead migratory behavior to stream flow is likely to be

different among watersheds and difficult to consistently represent in models

among different streams or steelhead populations with different life histories.
Jonsson (1991) suggests different behavior in different rivers may occur because
there is a hierarchy of environmental factors initiating migrations, or the fish

ad apt the timing of migration to different factors in different rivers.

In conclusion, I developed a new approach to analyzing upstream adult
Steelhead migrations using a temperature-based movement model that
q u a ntitatively predicts the migratory behavior of steelhead over a range of water
xerh peratures. The TBM model demonstrates an increasing probability of
mbvement over increasing stream temperatures for temperatures greater than
the threshold of movement temperature. The TBM model predicts migratory
behavior in two Lake Michigan tributaries that differ in size, location and other
factors. The transferability of the model to the two different tributaries suggests
that the modeling approach is broadly applicable and may work well in other
Great Lake tributaries and possibly other rivers that receive spawning migrations
of steelhead that rely on water temperature as a cue for movement. The TBM
model works well for steelhead and can be a useful management tool by

predicting of the migratory behavior of nuisance species such as sea lamprey

71

 

“|

and by synthesizing fish passage data to provide predictive migratory behavioral

information on other species of fish.

72

 

CHAPTER 5
The spawning habitat selection of steelhead in the Pere Marquette River,

Michigan

73

 

w I

Abstract

l evaluated features (groundwater, substrate particle size, etc...) associated with
the selection of steelhead (Oncorhynchus mykiss) spawning habitat in the Pere
Marquette River, Michigan 1997 to 1998. Steelhead redds were evaluated for
the presence of groundwater using GIS-based groundwater prediction model and
another method based on intragravel temperature. In addition to groundwater,

m icrohabitat-scale features associated with the redds (i.e., substrate particle
Size, redd depth, stream velocity above redds) were evaluated among redd and
reference (non-redd locations) data to determine their importance to steelhead

S Dawning habitat selection. The GIS model and probe-based evaluation were
in Conclusive as a means to identify groundwater associated with steelhead redds
d“ ri ng the spring. Steelhead preferred to construct redds in a substrate
(:0 '3 sisting of small gravel, large gravel, and small cobble particle sizes
§\'Sproportionately to clay, silt, sand, and large cobble. Steelhead redds were
located in areas where the stream velocity was significantly higher (F = 97.77, P
< 0.0001) than velocities that were recorded at reference sites, and redds were
located in water that was significantly shallower than what was typically found in
the study reaches (F = 113.84, P < 0.0001). Stream temperature did not appear
to influence the selection of redd locations. Although the groundwater data were
inconclusive, the microhabitat data can be used to identify specific areas in the

Pere Marquette River that may serve as steelhead spawning habitat as part of

habitat improvement projects that commonly occur throughout the watershed.

74

 

‘_ .

Introduction

spawning habitat characteristics such as substrate particle size, spawning site

water depth and stream velocity of steelhead (Oncorhynchus mykiss) spawning

sites and other Pacific Coast salmonids such as sockeye (O. nerka) and pink

salmon (O. gorbuscha), are well known (Cooper 1965; Kogl 1965; Smith 1973;

F u kushima and Smoker 1998; Groves and Chandler 1999). However, spawning
habitat information about Great Lakes steelhead is scarce. In addition, little is
known about the influence of groundwater on the selection of spawning habitat

by steelhead in the Great Lakes and Pacific Coast regions.

 

3 r0 undwater is important to the reproductive success of the brook trout
[Se lvelinus fontinalis) and chum salmon (O. keta) by protecting eggs from ice
eh. ‘QI infiltrating surface water (Kogl 1965; Fraser 1985; Curry and Noakes 1995;

erry et al. 1995), and may be important to reproductive success of steelhead by
providing more stable flows and a thermal refuge to developing alevin from warm
water in the summer (Shepherd et al 1986). Locations where groundwater is
abundant in a river may be preferred by steelhead as spawning areas.
Identifying groundwater locations within a river system where steelhead
reproduce could provide insight to the importance of groundwater and its role in

the selection of steelhead spawning sites.

There are several approaches that are typically used for detecting groundwater

within river systems. Piezometers are used to detect intragravel water

75

 

movements and the presence of groundwater (Lee and Cherry 1978; Sowden
and Power 1985; Curry and Noakes 1995; Curry et al. 1995). Because the
installation and monitoring of piezometers is a time-consuming process, most
studies where piezometers are used tend to evaluate small-scale (site specific)
groundwater flows rather than flows within a watershed or portions of a

watershed ONhite 1990; Curry et al. 1995; Curry and Noakes 1995).

l ntragravel water temperatures can be monitored and compared to stream
tern peratures to determine the presence or absence of groundwater (White et al.
1 987; Silliman and Booth 1993). The thermal regimes of intragravel and stream
”Va ter can be influenced by groundwater, infiltrating surface water, and solar
h 3 Eating effects. The degree of interaction between the stream and intragravel
“IQ ter influences the water temperature in both environments and may depend on
‘fige flow rate, turbulence, gradient, and other factors such as substrate
composition (Shepherd et al 1986). To accurately assess the presence of
groundwater, water temperature is usually monitored over an extended period
(e.g., one year) via data loggers or periodic visits to sample locations (Silliman
and Booth 1993). Detecting groundwater by monitoring water temperature can
provide information about larger areas of river than piezometer-based studies,

because the data collection process tends to be less labor intensive.

Another approach is to evaluate groundwater areas based on predictions from a

model. A Geographical Information System (GIS) groundwater model was used

76

 

to predict areas of groundwater accumulation within the lands Cape base 6
0”

et al. 2000). The groundwater model PrOVlded a coarse-scale (100 m2)
landscape prediction of groundwater accumulation within watersheds. The GIS-
based groundwater data are readily available without collecting additional field
measurements, making it POSSib'e to link the groundwater information with

steelhead spawning data.

The primary goal of my study was to gain a better understanding of steelhead

spawning habitat use in a tributary 0f Lake MiChigan- ldentifying groundwater
areas in relation to steelhead spawning locations may he”) determine if
groundwater is an important feature 0f steelhead spawning habitat selection.
Because I evaluated steelhead Spawning habitat throughoUt a Watershe d, I used
water temperature monitoring and a GIS groundwater mOde! as surrogates of
piezometers to Provide evidence of groundwater presence. One Objective of my
study was to determine if steelhead selected spawning habitat where

groundwater was abundant Within a stream. I also evaluated the use of Water

temperature and a GIS model as tools to examine the presence of grou ndWater.

Groundwater is not the only factor that may influence the selection 0f Spawning

habitat by steelhead. Groundwater is low in oxygen and can inhibit the
development of embryos (Cooper 1965; Fraser 1985). Spawning site (redd)

formations can improve convective patterns and the delivery of oxygenated Water

77

 

 

to the redd within groundwater areas, and promote convective flowparterns/h
areas where intragravel flows via groundwater supplies and advect/l/e flows are

limited or not available (Cooper 1965'. Curry and Noakes 1995).

The substrate ParttCte comPOSitton 0f redds Vary widely among steelhead and
other salmonids (Olsen 1968; Neilson and Banford 1983)- However. the
substrate particles must be large enough and adequately sorted to accommodate
alevin emergence; the substrate mUSt consist Of a movable porous material Such
as gravel so that a female can excavate the r edd; and the redd must be located

or shaped in a way that promotes the movement of oxygenated water through

the redd (Cooper 1965; Fraser 1 985; SOWden and Power 1985).

Because water movement (stream flow and intragravel) and SUbstrate
composition influence the reprOd uctive success 0f steelhead, they may influence
the selection of Spawning habitat by PaCific Coast steelhead and other salmonids
(Smith 197 3; Fukushima and Smoker 1998', Groves and Chandler 1999), and
may influence the selection of spawning habitat by Great Lakes Steethead.
Another objective of this study was to evaluate microhabitat-scale features
associated with the redds (i.e., substrata particle sizes, redd depth, stream
velocity above redds). Specifically, I evaluated the use of microhabitat-scale
features in relation to their availability within selected reaches of river. Identifying
habitat features that are important to steelhead spawning site selection in a Great

Lakes tribUtary can be used to identify other areas within the river that may be

78

 

d b ning steelhead, or possibly could be modified to Scream/traded?
use y spaw

steelhead.

79

 

 

Methods
Sample Locations
The Pere Marquette River is located in west-central Michigan (Figure 2- 1).
Sample reaches (1-2 km reach es) where the bottom was readily observed and
the stream was accessible (public access, boat launches, and access permission
from riparian landowners) were identified Within the Pere Marquette River (Figure
5-1). Five study reaches were randomly selected and sampled in 1997 and
1998, and nine were randomly selecmd and sampled in 1999. I attempted to
collect data from each site between 10300 and 14:00 hours when the daylight
was the greatest intensity to optimize the Visual detection of redds. Redds were
identified by areas where there was evidence of excavated SUbstrate free of
debris, typically Shaped with a conical depression in the center, and occasionally
by the presence of spawning steelhead. Redds with fish tending them were
recorded as active and redds without steelhead present were considered as
suspected steelhead redds. The location of the redds With r espect to the river
channel (river-left, mid-channel, and river-right) were noted, and with respect to

run, riffle, and pool habitat types (Hicks and Watson 1985).

Stream temperatUre, intragravel temperatures, mean column stream VSIOCity,
depth of water above each redd. and substrate particle size composition data
were collected at each redd. water temperature was measured using a Yellow
Springs 'nstrument Company (YSI), Inc.TM meter at most redds. For reaches with

high redd density. | sampled every fifth redd. Mean column stream velocity, and

80

 

 

 

  

Km. 3 m.‘ 4 comm: ho Eon 58:8: 05 ”902

98. c . ago .
b _ 9592 mom? 5 .5 $2 .8 59:25. .521 9383.2 “Mauwgcs s flames some been 33.65%
£ 5 £52.95 comm: Emoem .Tm. 9:9“.

        

.5;

 

.521
oxen—:22 Sum KO
552m each 2.: V\

81

 

 

0 /
w/V . 52.x
. I . .— 3338: Bon—
k»\ + a 555 5.5m 9m
33m <
383922 Son. 2235 23:2 .3 21* Q
EBEmm . Biz
a 2.6.3 98
5.5. c E a
i|\I|II\., .

3. m D /

 

 

 

water depth at the measured redds were recorded using a Pr 3 Ce AA mete d
ran

wading rod. Stream discharge data were obtained from the L13, Geo/09,03,

Survey gauge station. located i“ the river at Scottville (Figure 2-1). Inna-gravel

temperature was measured in 1 998 and 1999 using a YSl series 400. 95-cm long

stainless steel temperature PTODe and tttermister. lntragravel measurements

were taken at depths of 5. 10. 20 . 30’ 40’ and 50 cm into the substrate at eaCh
redd for 1998. All redd measurements were taken in an area that was assumed

to be the egg depOSitiona‘ zone, or apprOXimate’y One-third Of the diStance

between the deepest point of the redd dePTeSSiOn, and the downstream edge of

the redd.

Substrate particle size composition was visually estimated Within a 1 m area
centered on the egg dePOSlttona' zone for each sample Site Using a modified
Wentwprth @922) c\assification (Table 5-1). The large and Small cobble particle
size classes were combined during the 1997 sampling and were differentiated

during the 1998 and 1999 sampling.

Table 5-1. Particle size classes that were used to classify substrate Particle
sizes of sample Sites in the Pere Mar uette River, Michigan 1997-1999.

 

 

 

 

 

Partich Particle size \
Gray - 000024-0004 mm\
Silt 0.04-0.062 mm
Sand 0.065-1 mm
Small gravel 1-8 mm
Large gravel 8-64 mm
Small cobble 64-128 mm
w 128-512 mm

 

82

 

. “3‘ (

Reference data from non-redd locations (the Same informatio '7 that was coflected
from redds) was collected within each site for the 1 998 and 1 999 data, and there
were approximately 10 reference locations per kilometer of study reach.

Reference locations were equally Spaced from each other and alternated from
river left (RL) to mid-channel (CH) to river right (RR) locations in a downstream

direction (Figure 5-2).

RL

Stream Flow

CH

 

i R

CH

 

 

lRL i

Figure 5_2_ Reference sample site configuration where RL indicates river.left
CH indicates channel, and RR indlcated river-right. ’

83

 

Groundwater Evaluation
Spawning areas were evaluated based upon the number of redds/m2
corresponding to the relative area of categorized groundwater flows at 25, 50,
and 1 OO-m distance-trom-the—river intervals and the groundwater flow values
immediately adjacent to the river to determine if the distance from the river
influenced the groundwater predictions among stream reaches. The
groundwater flow predictions were referenced to distance intervals using Arc
View version 3.2 geographical information System software. A correlation
analysis of the flow values for eaCh distance We rval was performed using
Statistical Analyses Software (SAS VB-O) to determine if there were significant

correlations among reported flows at each distance interval.

GlS-based predictions

To determine if steelhead redds were linked to areas of groundwater input in the
Pere Marquette River, redd data were linked to groundwater accumulation
predictions from a model based on surficial geology, digital elevation and Darcy’s
Law of groundwater flow (Baker et al. 2000). Groundwater predictions Varied
within each StUdy reach. The Darcy groundwater predictions were categoriZed

as high (1 ,5 to 3 standard deviations from the mean groundwater potential),
medium (0 to 1.5 standard deviations from the mean groundwater potential), and
low flows (0 to --1 standard deviations from the mean groundwater potential) from
the model predictions to unify the prediction within each stream reach for

comparative purposes.

84

 

 

i“

 

A Mixed_Genera| Linear Model (Littell et al. 1996) was also used ,0 gm,“ dd
ate re

density among study reaches and groundwater predictions. RGda’ dens ity was

evaluated according to the following model:

y=#+ ar+l31+ l’i.l+ ’7k+ 9

Where,

y = redd density by year and dary flow Prediction

,u = model intercept

a,= year (fixed effect)
51 = darcy flow prediction (low, medium, 0' high 1’low). fixed model effect
71.]: interaction parameter

77k = Study reach (random effect With mean zero and 0': )

e = model error.

Prolgpasedm

A Mixed-General Linear Model (GI-M) was used to evaluate water temperatures
in the stream and intragravel portions of redd and reference site among study
reaches. Least square means (Littell et al. 1996) were used to determine Where
significant differences occurred. The following model was used to 6Va|uate Water
temperature in the intragravel eriVii‘Oi‘iriifi‘nt and stream:

y=p+ar + flj-r- g,+ l’i.j"' 1,3,+ Kj,r"' 77k+e

Where,

85

 

 

y ____ water temperature (intragravel and stream) by year, temperature pro be

depth, and sample location
,u = model intercept

(1i: year (fixed effect)

,6;- = depth of temperature sample (fixed effect)

4 = sample location (fixed effect)

m = interaction between year and depth of temperature sample
r,;/= interaction between year and sample location
Kp= interaction between depth of temperature Sample and Sample tocation

77k = Study reach (random effect with mean zero and 03)

e = model error.

According to Silliman and 800th (1993), a stream reach is Considered to be
neutral with respect to gaining groundwater or losing Stream water to the
sediments in stream reaches where the Stream temperatUre is Significa
warmer than the intragravel temperature , and the intragravet tempera nt/l’

’Ure
fluctuates daily. A stream reach where the intragravel temperature is
s

. tab/e
cooler than the stream temperature IS considered to gain groundwater and

and
. a
stream reach where the lntragraVe' and stream temperature fluctuate 0v
er a .
. . d
CYCie and are not significantly different are consrdered to lose stream w t 3"y
a e

r i”to
the sediments.

86

 

 

 

 

 

 

at .
7 hr ’fl/EWa/s

over 25 days in their study. I monitored intragravel and street—“.7 temp
e’t'itu/es'

within each stream reaCl‘ approximately 4 hours, or the time i 1 tOO/r to Sam

p/e ”7%
redd and reference sites within each reach. Groundwater presence in each
stream reach was determined according to the mean difference between the
intragravel temperature at each depth and stream temper atUre. Stream r eaches
where the mean difference between the intragravel and Stream temperature Was
negative were considered as high groundwater UPWemng areas. Stream r eaches
where the mean difference between the intragravel and stream temperature was
positive were considered as low groundwater upwelling areas, and stream

reaches where the mean difference was near zero were considered as medium

groundwater upwelling areas.

- ' nd
A Mixed GLM was used to investigate the relationship between redd denSitY a
i ntragravel prObe-based groundwater predictions among study reaches. Redd

density Was eValuated according to the following model:

Y‘ =p+ai+fll+nj+77k+e

Where:
y § redd density by year and probe-based predictions (low, medium, or high

i\o\lv), at 5. 10, 20, 30, 40’ and 50 cm

# = model intercept

a: = year (fixed effect)

_—_- probe-based prediction (fixed model effect)

flj

87

 

 

 

 

if k = interaction parameter

77k = study reach (random effect With mean zero and 02)
r]

e = model error.

Redd Characteristics

Substrate particle size distributions were Compare d for year—to-year (1997, 199 8 ,
and 1999), and reference and redd Site differences (1998 and 1999) using the
Kolmogrov-Smirnov two—sample test statistic (Berry and Lindgren 1995).
Vanderploeg and Scavia’s eleCtiVitY index (Er) was used to compare steelhead

redd substrate use with reference site substrate data (Van derploeg and 803 via

 

1979):
\W —-K\—3\ /
Er: ‘ "\ Where, W13???)-
\W.’r K—\\ ’ ’
Y1
Where,

ri = the proportion of observations of particle size I in the re dd

SUbStrate
composition
Pl = the proportion of observations of particle size I in the referen
composition strate
n = the number of particle size categories.
The index ranges from +1 to -1. A positive value of 5 indicates a larger

propOrfiOn of the substrate partiCIe size in the redd than is found in the str
earn

reaches A negative Value of E indicates a particle size was Used less th
an was

88

 

 

 

available in the stream reach es. A value of 0 indicates a particle size Was u d
se

in proportion to availability within the stream reache
s.

A Mixed GLM was performed on stream velocity, water depth. and water
temperature, to examine diffe rences among Years, channel locations, study
reaches, and between samP|e ”cations (mod and reference sites), Least square
means (Littell at al. 1995) were used to determine where significant differences
occurred. The following model was Used to SValuate stream velocity, water

depth, or water temperature:
yzfl+ai+,flj+ g-f-fi/ + Iii/4. [(1, + 77k+e

Where,
Y = si‘oam velocity. Water depth. or water temperature by
and samo\e \ocat‘ion Qar, Channei ‘ocafion’
r1 = mode\ \ntercept
a.- = year (fixed efiect)
,8,- = channel location (fixed effect)
(P sample location (fixed effect)
m: interaction between year and channel location
In: intefaction between year and sample location

K131: interaction between channel location and sample location

m = Study reach (random effect with mean zero and 02)
)7

9 = modelerror.

89

 

 

 

The Tukey-Kramer Dam/Vise Comparison Was used to determine where the

significant differences occurred.

y=fl+m+e

y = discharge recorded at the USGS QaUging Station in Scottville. Michigan

,u = model intercept

a,- = year

8 = model error.

All tests were conducted using or = 0.05.

90

 

 

 

 

 

 

 

 

   

Resuits

I recorded steelhead redd and r*"v‘f'i-irence data from 15 April to 24 April 1997, 15

April to 23 April 1998, and 28 Mar Ch ‘0 17 April 1999, l observed the greatest
redd density (53 redds/ha) in section C (Figure 5-1) during 1999. and the lowest
redd density (0 redds/ha) in $961k)" E during 1 999 (Table 5—2). Redd densities
did not vary significantly among years (F = 0.1 9, P = 0.84).

Table 5—2. Redd density by stream f each for 1 997, 1998, and 1999 in the Pere
Marquette River, Michigan-

 

 

 

 

 

 

 

/_'_______—/ ,— ‘//——-—
Streaf‘f“ Da Fl Redd density Numifil‘ 0f
Year Rey re): ow (numbW
1997 A High 530 37
0 Medium 26.0 57
J High 0.3 4
L Medium 5.0 100
”i998 A High 35-0
B Low 7.0 23
C Medium 35_ 0 59
tooe B Low 9. 76
(3 Medium 68‘ O
E Low 0. Q 72
F Low 30.0 749
H High 4.0 72
TN
Groundwater

GIS-_based predictions
I performed a correlation analysis to evaluate the groundwater flOW (as in d‘
Icated

by the Darcy flow model) at the 0, 25, 50, and 100-m distance-from the

91

 

 

river intervals. Correlations between distance intervals were higmy Significant (p
< 0001) indicating that any Of the intervals would be adequate to use in further
analyses- Because “‘3 °°”e'a“°" coefficients varied little (Table 5-3) and all
gniflcantly correlated . I used the Darcy area estimates from the 0 m

were at

category for the additional an alysis.

Table 5-3. Correlation coofilcwfl‘tspfor 0 t0 100 m categories of Darcy flow area
estimates in study areas from the ere Marquette River, Michigan.

 

 

 

 

 

/’/1
Cate o n=13 0m 25m soy; 100m
0 m 1.000 0.986 0.986 0,911
25 m 0.986 1.000 0.997 0 ~991
50 m 0.986 0.997 1.000 0 .993
.971 0.991 0.993
100 m )7 N

 

 

 

#1

, tatives in each D
Because of the \ack of mump‘e represen Q"W flow Ciass 00%
medium, anohgh ’i\ow) in the 1997 and 1998 data, the tegt f0“, t
. n eraction
between \leat and Darcy flow effects was not possible- Redd d
ensity
_ _ was not
significantw di‘tieren't among years (F — 0‘10’ P _ 0'91) and Darcy fl
0W3 (F ..

0.14, P = 0.88).

Prooe-based redictionS

There was no consistent relationship between water temperature, probe (1

redd and reference sites, and years. There were no significant differencesepth.
between intragraVel temperatures of redd and reference sites (F = 0.04, p =
0.84). However, intragfaVe‘ water temperatures wer e Significantly different

among probe depths (F = 58.88, P < 0.0001) and years (F = 520 53 P < 0 0001
. , _ )'

92

There were significant interactions between redd and reference sites and years
(F = 539, p .—. 0.02), and water temperature at all probe depths (F = 4,53, p<

0.0001), and between years a ”d water temperature at all probe depths (F =

10.42, p < 0.0001).

Because Of the limited availability Of temperature probe data at depths greater
than 10 cm, the following analyses evaluated intragravel water temperatures at 5
and 10 cm. The mean difference bt'3‘tWeen intragravel temperatures measured at

5 and 10 cm into the substrate and Stream temperatures, ranged from -1,6 to 1 .0

°C among depths and stream reaches (Table 5‘4) sum)! reaches with a mean

difference between the intragravel and stream temp eratui- Q (A d from

T) that range
' ' d stud reac
— —0.7 °C were te orlzed as high, an y -
1.7m Ca 9 es wrth a mean
ditierence that ranged from 0-29 t0 1-1 °C were categorrzgq
as low with reSpecl
to groundwater presence thrOUQhout the reach. Study re
Chas with a
mean

difference between the intragravel and stream temperature that
range
(1

0.69 to 0.2 °C were categorized as medium. from _

All probe-base d predictions were high among study reaches in 1998

. and we
”Ct CODsistent among probe depths m 1999' P r Che‘based predictions ’9
Wer-

o e n
consrstem among study reaches (e.g., study reach B) and Years and W 0t
' e

r e not
consistent with the GIS-based groundwater Predictions (lable 5-4) Beca
I U

. . 88 0f
unbalanced data, the evaluation 0t redd denSItY and probe predictions we
s

not
possib 1e.

93

 

 

 

 

Table 5-4. The mean difference (AT 1' standard error) between int ragrave/
tern eratures measured at 5 and 10 cm into the substrate and stream
tures, and the prObe-based and Darcy groundwater flow predictions in

 

 

 

 

 

tempera River M ichi
the Pere Marquette . 9a“ sturdy reaches 1998 and 1999.
//i \f/
*/ Mewbe prediction
Study Darcy 5 I
Year Reach Flew ‘_ 1 1 :7; 1 o 1 0 cm 5cm 10cm
1998 A High 0'8 : 0'1 -1 .2 i 0.15 High High
B Low — - j; 0° 3 -1.0 i 018 High ngh
C Medium '0’8 1 0' 1 3 -1 .0 i 0.18 High High
J High ‘1-3 ’ '40 -1.5 i 0.12 High High
1999 B Low 1 -0 i 0-24 0.8 :t 025 Low Low
C Medium -0-7 i 0-14 -1 - O ,+__ 0.17 High ngl'l
E Low -12 i 0.11 -1 - 5 3: 0.14 “‘9“ High
F LOW -O.6 i 0.27 -1 -6 i 0.35 Medium High
G High -O.4:l:0.10 —O-5 i 0.12 Medium Medium
9- —0.6i0.10 -0-9 i0.09 LOW .
J H- h 0010.19 —0.1 :020 edlum (rum
'9 - +005 —0.1 :0.08 Medium M9 ‘
Medium W

 

 

Substrate Particle Size

Steelhead did not appear to construct redds in areas with larger
6 $1295. I

did not observe the small—boulder particle size (256-512 mm) and I
arger

particle
o
f redds Sampled
in

1993 and 1999 (n = 219). Although large particle sizes Were more f
requen
ti

sizes (> 512 mm) in redds. I observed large cobble in 6%

sampled (10% of the sampleS) in the 1998 and 1999 reference sites (
n = 249)

larger particles were not common within the study reaches.

Steelhead also avoided smaller particles such as silt and clay °Ccurring in
. . redds.
I did not observe Silt and clay m any Of the active and suspected redd sam I
p es,

and Sllt was observed at an average of 3% (1999) and 0.8% (1998) in the

94

reference samples. I observed clay at an average of 0.7% and 0% in the

reference samples.

l evaluated the substrate particle size distributions among active and suspected
reddS. reference sites, and sample years using the Kolmogorov-Smirnov test.
The Particle size distributions of active and suspected redd locations were not
significantly different (P > 0.27, for each year) for 1997 to 1999. Thus, I
combined the active and suspected redd data from each sample year for further
pa’TiCle size analyses. The particle size distributions of redd (active and
sL’S'Fbected) and reference sites were significantly different in 1998 (P < 0.001)
an d 1 999 (P < 0.001). In 1998 and 1999, steelhead redds were associated with

\QTQer particle sizes than reference sites within the study reaches (Figure 5-3),

GDd did not select locations with silt or clay.

The 1998 and 1999 redd substrate data (large cobble was differentiated from
small cobble) were not significantly different (P = 0.186), and neither were the
1998 and 1999 reference data (P > 0.27). The combined 1998 and 1999 redd
particle size distributions were not significantly different between run and riffle
habitats (P > 0.27). I did not observe redds in pool habitat (Table 5-5). I
Observed pool habitat in 15 reference samples over two years (Table 5-5). The
combined 1998 and 1999 reference site particle size distributions were not

Significantly different among habitat types (P = 0.22 for run vs. riffle, P > 0.27 for

all other Combinations).

95

.82. new .83 .53 .2 59:25. cozy. 26:05.2 meme on. E woem 8553.. use
now. 5:23 momma? mum 333 36 new Emv Em .6; 0:58 ones .83 228 :95. .Amv name .3: .055 deem.
.Ammv .990 =me ho @9898 88.8 .92 8 8:39.66 5:033: E083 9:3:an of. .m-m 9:9“.

cozatomou on? o_o_tmn_

 

o_ om 9 mm ncmm
- . p . p 0 My
. w
m
._..o m...
x m
\ IN.O w
x m
. . W
\ m o n...
mmmelll \ .m
wmmelll \ .vd n
321T \ m
momell \ . m
82! l \ .m o o
\ u.
\ .ed m
x m.
\ .50 d
a
\ w
\ rwe W
\\ o
x m
-®.o E
III |||| a“
\ \ilJH‘lIIlil“ F a

 

96

Table 5'5 - Redd and reference habitat characteristics in the Pere Marquette
Wigan 1997, 1998 and 1999.

Sample Year
1998 1999
Redds

0.39 (s = 0.40) 0.37 (s = 0.47)

Site Metric 1997

 

Mean water depth above redd (m) 0.43 (s = 0.45)
Mean water temp above redd (°C) 8.5 (s = 1.2)

9.4 (s = 0.2) 10.3 (s = 2.5)
Mean velocity above redd (cm/s) 73 (s = 50) 73 (s = 44) 64 (s = 48)
Mean nuh‘lber redds/ha 30 26 17
Percent of run samples 15 (n = 27) 36 (n = 27) 7 (n = 10)
Percent of riffle samples 85 (n = 151) 64 (n = 48) 93 (n = 140)
Percent of pool samples 0 0 0
Reference Sites
Mean Water depth above ref site (m) n/a 0.64 (s = 0.71) 0.60(s = 0.77)
Mean water temp above ref site (°C) n/a 10.3 (s = 1.4) 10.0 (s = 2.3)
ean velocity above ref site (cm/s) n/a 60 (s = 73) 46 (s = 64)
P er Cent of run samples n/a 83 (n = 82) 76 (n = 113)
er Gem of riffle samples n/a 11 (n = 11) 18 (n = 27)
e—er\Cent of pool samples n/a 3 (n = 6) 6 (n = 9)

 

‘QV aluated the combined 1998 and 1999 data for steelhead electivity of
Substrate particle size. Steelhead preferred small gravel, large gravel and small
CObble disproportionately to their availability within the stream reaches (Figure 5-
4) - Although sand was frequently observed in the reference samples (Figure 5-

3) . sand, large cobble, silt, and clay were not preferred within redd substrata

(Figure 5-4).

The 1998 data were not significantly different from the 1997 data (P = 0.23), and
the 1999 data were not significantly different from the 1997 data (P > 0.27). All

substrate particle size data (1997 to 1999) were combined for further analyses.

The Cornbined 1997 to 1999 redd particle size distributions were not significantly
different between run and riffle habitats (P = 0.12). Redds were not observed in

pool habitat. The combined 1997 to 1999 redd particle size distributions and the

97

8:358 82 use 82 5822.2
cozm 30:05.2 Eon. 05 c_ wcozatomou cum 0.2th 05:33 @595QO noo£c$m .8 mo:_m> b_>=om_m_ .vrm 059...

 
 
 

 

I O.—...
cozatomoo cam 23th Satan—aw
. we-
I $.01
. . - a
e o m
m.
M.
. No- .m.
W
p
. ed m
A
W
. Ne m
S
m
. so (
2.300 093 ofinoo __mEm .390 omen; .920 =me 9:3 Em >30 . ed
3

0.?

98

location Of the redds with respect to the river channel (RR, CH, and RL) were not

significantty different (P > 0.27) for all possible combinations (RR vs. RL, RR vs.
CH, and CH vs. RL).

Velocity, Water Temperature, Water Depth

Stream velocity was an important component of steelhead spawning site
selection. As illustrated using the 1998 data in Figure 55 reference velocities
Were significantly lower than redd velocities (F = 97.77, P < 0.0001) and redd
I’e'CJCities were significantly different among years (F = 15.01, P < 0.0001).
There was no significant interaction between redd and reference site velocities
a“?! years (F = 0.23, P = 0.63). Redd velocities varied less than velocities
measured at reference sites (Table 5-5). The 1999 redd velocities were
Significantly lower than the 1997 (t = 4.57, P < 0.0001) and the 1998 (t = 3.18, P
= O- 016) redd velocities (Figure 5-6). The 1997 and 1998 mean redd velocities
we re 73 cm/s (Table 5-5) and were not significantly different (t = 0.10, P = 0.22).
Like the redd data, the mean daily discharge recorded at Scottville, Michigan for
the month of April was significantly lower in 1999 than 1997 (t = 3.01, P = 0.003)
and 1998 (t =4.87, P < 0.0001). The 1997 and 1998 discharge data were not

significa ntly different (1‘ = -1.80, P = 0.08) for the month of April.

Stream Velocities for redds in the center of the channel were significantly greater

than redds located in the river-left (t = 0.008, P = 0.02) and river-right (t = 0.002,

99

.cmmEoS ..0>_m 020290.). 0.0;
05 5 meme ocean 00:0 00:000. cam 80.. “3005006 “0 “0980. b_oo_0> €00.50 eo 5:030: E099; .m-m 059“.

323 E02;
2 a 5 me mm 5

 

 

 

 

. l . , l .
. t. - .. a . . l. .. 3 L
«n _ _m .H . .. w ,. _ .fi
_ _ a _h _ .. . . . a ..
.. at e . .. ... . . e r ~ 3 a. .
.e. _ . _. .. e . . . . n ..l.
_x n . m . ... n.
_. r u _ l. . . a
v .. n. s .. o v . vl
. . .. a . _. . .. a . c.
t Y . .I s. . ‘ o 0.. V .. v
e r . f a r. ... . e.
w . N. «V w . m m s.
.. .. Inn u H .. : ..
. _ "a. .. _. .. “
. . . w .. . .7
. .. ... ... e n I m
.
. . . .r A n ..
.. . .. . . ..
. v . z o
._..
. .
. .. ..,
a. .... a.
1m w.
...~
T
u
. m
1...!

92
llauanbald wasted

20m meme I
0oc0e0e0m meme H

.mm

 

100

7/
33?”? W

rm. ‘v. . ""'1"W"*" “(0.“.

,’.'. - r,
., .. ._

7

I 1997 Redd
a 1998 Redd
R 1999 Redd

 

 

30 1}

zsl

N

AOUGI‘IDGJ} tuaolad

101

0 LO 0 t0 0
1- ‘—

co
0)

73

C)
v

Velocity (cm/s)

24

Figure 5-6. Percent frequency of stream velocity recorded at steelhead redds during 1997, 1998, and 1999 in the

Pere Marquette River, Michigan.

P = 0.05) Channel locations. However, velocities of reference samples were also

higher (P = 11.88, P < 0.001) in the center of the channel.

Within S‘te deviations of stream temperature was not an important component of
steelhead selection of spawning habitat in the Pere Marquette River. Water
temperature above the redd was not significantly different between redd and
reference sites (F = 1.16, P = 0.28), but was significantly different among years
1997 (t = 11.78, P < 0.0001), 1998 (t= 12.75, P < 0.0001), and 1999 (t= 15.37,

p < 0.0001).

3\%elhead redds were located in significantly shallower water (F = 113.84, P <
0.0001) than reference sites (Figure 5-7). Steelhead constructed redds in
approximately 0.20 m shallower than was normally found within the study
reaches (Table 5-5). The sample year did not appear to influence the depth of
Water above redds or reference sites (F = 1.92, P = 0.15). Redd depth was not
significantly different according to the location of the redd within the river channel
(F = 2.65, P = 0.07). The water depth among reference sites was only
Significantly deeper (I = 2.29, P = 0.02) in channel locations when compared with

river-left locations.

102

.cmmEoS £02m 0:02:05. 0.01 05 E meme 2 meme Eo: 002m

00:990. 80.: ocm dome 9 32. So: m0ew 20.. 0005005 .0 “.0280. £QOU e903 eo 5:030: E0801 5m 05?.
as :38 pose;

we 5 ma 3 .3 3 ed we we no me No ed

J . . It'd . are r ...,._..,.._ I... re . o

a h
m .-
.. .H
e ...
.... r

 

 

 

is :23. a..-
.-
,. . . . . .. ~
‘1 .
Luau!“ runs :- r?

.n 1 u

n<.ruusv

 

.ON

’1'! '. iffl‘NH‘U .

y .

.; .n...‘.. .. ..,., a .l.
,_ . . .r. A..‘...L...s.A...::.‘.-s.. . .a- .-. ... .
., .. . ,. .
5' . \ l‘ -
-nvnrruowmc-rinuv -.- r. r. v . v v ., v v v\ v‘ :

.mN

.om

eeeeo seem I
2600 00:0.0e0m e . mm

Aouanbald JUGGJOd

.ov

 

me

om

103

 

Discussion
Groundwater
Steelhead redds were “0‘ d isProPomonate'y 'Ocated in areas of low. medium, or
high groundwater flow Within the Per e Ma rquette River using the GIS—based
groundwater predictions. Because the G '8 Model Was designed for coarse-sea le
predictions (100 m2) it may not provide enough detail to accurately assess
steelhead spawning locations in relation to groundwater areas within the Pere

Marquette River (Baker et al. 2000).

In addition to the GIS-based groundwater prediCtionSr the eva|uaflon Of
. _ . _ 'n
intragravel water temperature was Inconcluswe as a means of ldentli‘l‘ 9

. . . , . \ed
groundwater input areas in the river. Silliman and Booth (1993) ldent‘f
' ' ' s over
groundwater within a stream reach by monitoring lntrag rave) tempe rature
25 days within a fixed location. I monitored portions of a stream reach for
' cessive obsen, . ,
approximately 4 hours by collecting suc atlons . n a downstream

. . h was ablg t .
. _ onltormg approac Q detect dIFf
directlon. Although my m in fences
among stream water and intragravel temperatures "705‘ cases, I

- fth intra as Unable to
establish the status (stable or fluctuating) 0 e gravel temperat

r 63.

Finally, groundwater may be Plentifu' mmughom mUCh Of the Stee'head
Spawning

th

respee't to groundwater unimportant. Curry and Noakes (1995) found that

habitat within the Pere M arquefle River. making the location of a redd Wi

groundwater may have drawn brook trout to generalized spawning areas but d'
’ l

104

 

 

 

 

 

not influence the selection of redd locations in areas where groundwater was

abundant within the baseflow of the stream.

Subs/rate Particle Size
an
Steelhead exhibited a preference for Substrate particles that were larger th
53nd yet smaller than large cobble. Sowden and Power (1985) indicated

ove it.
spawni n 9 substrate must be coarse yet small enough that the fish can m

Nthough I visually estimated substrate particle size, I consistently detected at
preterence for specific particle sizes by steelhead. Another substrate sampling
method like the pebble count pebble (Kondolf and Li 1992) where the particles
are randomly chosen and physically measured produce data that can be
evaluated parametrically. However, because of the large sample areas and the
abundance of redds, the pebble count method was too time consuming and “at a

feasible method given these constraints.

l’e/ocit‘y, Water Temperature, Water Depth
Stream velocity appeared to influence the location where Steelhe
- ' - . Co"Structeq
FQ dds. Steelhead constructed redds In locations wrth higher str e
. . VelOCi 1‘
average within each study reach. The stream velocmes that I met), d ty ham
6 at the

V% dds in the Pere Marquette River were consistent with steelhead r

' d th valocities
as determined by Smith (1973). Smith foun e average mean‘COlumn

105

 

 

velocities at redds to range from 63 to 70 cm/s, as compared to 64 to 73 cm/s in

my Study (Table 5-4).

The faster moving water may assist embryo devel0pment by providing
oxygenated water (Stuart 1953; and Sowden and Power 1935), The delivery 0f
oxygenated water to the redd is 3'80 affected by water depth. The velocity of
water m Ust be fast enough to penetrate the interstices of the redd and deliver
oxygenated water, and the water must remain deep enough to cover the redd

until the fry emerge from the redd (Semko 1954, and Smith 1973).

Although steelhead consistently preferred to construct redds in ioeations with
stream velocities that were higher than average, the Stream velocity at the redds
and river discharge recorded at Scottville varied annually. The an nua\ Variatlo“
may be attributed to a large-scale effect that influences Stream flow (e.g., annual

differences in precipitation)-

Water temperature was not a deciding factor in the ChOice of steeu...
. , ead redd
’0 Qations- Because steelhead spawn during the spring in Michiga

. he" Water
t3 rnperatures are cooler, water temperature may "0t '“fluence the

fQ .— redd construction.

Water depth influenced the location of steelhead redds. As indicated b
Ythe

disproportionate number of redds located 5" riffles (Table 5‘5): Steelhead

106

 

 

Preferred to construct redds in water that was shallower than the average depth
within the study reaches. The steelhead in my study constructed redds in water

depths that were consistent with a 1973 Pacific Coast steelhead spawning StUd Y

(Smith 1973)_

Although my study did not provide additional information regarding the
relations hip between steelhead spawning and groundwater. l have identified
subStrate particle composition, stream velocity, and water depth features that are
specific to steelhead redds within the Pere Marquette River. Should Spawning
habitat be found to be limiting within the Pere Marquette River, these features
could be used as a measure to evaluate other locations Within the river where

habitat may be suitable for spawning.

Identifying these spawning habitat features and redd denSitleS Within eaCh Study
r each serves as a quantitatiVe description of steelhead Spawning habitat
preference within a Great Lakes tributary, and may help to guide fut
ure
anitoring of steelhead numbers within the river. Stream habitat:
p d . h' h h d mPfOVement
rQJ-ects are regularly conducte wut Int e waters e (persona, c

' - . mUni .
D§re Marquette Watershed Councnl). The habitat ImprOVement Drcy cat/on’
1e

. . . CtS a
st ream-bank stabilization efforts to minimize the amount of fine Seq- re often
Im

. . ° ents
Shiering the river. These habitat improvement prOJects may create
OP 0

p

awnin
\0 cations and redd densities prior to the improvement Project and eval 9
Uatin

107

 

 

 

 

Changes to spawning habitat following the work. My reSUltS could be used as

baseline data to evaluate the improvement project in terms of impr0Ving 0’

creating additional steelhead spawning habitat.

 

CHAPTER 6

SUMMARY

109

 

 

 

SUMMARY

Steelhead and longnose spawning migrations were described in the Pere
Marquette River using radio telemetry as part of an evaluation of an electrical S ea

lamprey barrier. Steelhead and longnose suckers moved upstream quickly (all
r adio-tag ged fish moved upstream Within an average of 32 minutes) through the

593 lam prey barrier vicinity, prior to barrier operation. The upstream movements

of r adt0~tagged fish after the barrier was in Operation were not evaluated in this

study. Rad io-tagged steelhead and longnose suckers showed a general

- . . nd
correspondence of upstream movements Wlth Increasing water temperature a

stream flow.

The telemetry data I collected from 1997 to 1 998 prOVid ed base-line migration
information for a future study on the effect of the Operation of the electrica‘ barrier

on migratory steelhead and longnose suckers, SF’eCifiCally the dat from my
' a

StUdy were used to evaluate migratory delay caused by -
barrier op .
e ration, the

Speed of passage through each study section, and the percent

8
eSch section. Sage through

A model was developed to evaluate the upstream movements of
Ste

r‘ . . e/head .
Q |atlon to water temperature and stream flow In the Pere M 8qu tt In
e
e an
J Oseph Rivers, Michigan. My data for the model inCtUded 28 rad- d the St.
'°~ta

. gge
Steelhead (Oncorhynchus mykiss) on the Pere Marquett e River a d
”d a

tar
(5 ’876 t° 10’083 Stee'mad): mum-Year (1993-1999) data set of ca 9e,
de

110

_._

_—

 

Steelhead passage through a fishway on the St. Joseph RiVer. My model
Pred'Cted the probability of movement as increasing over increasing water

temperatures above a movement—threshold water temperature. Stream flOW was

incorporated into the model, but did not add substantially to the model’s ability to

describe the migratory behavior 0f Steelhead in the Pere Marquette and St'

Joesph Rivers.

The TBM model may work to predict the migratory behavior of other fish such as
sea \amprey that move into rivers for reproductive purposes and time their
movements on water temperature and other factors. In addition. my modeling
approach could be used to predict the apprOpriate sampling periods based on
the

migration timing, or predict future run size from percent passage data trom

model.

Finally I described the spawning habitat use of steelh
I e .
ad I” the Pere Marquette
RiVer. A GIS-groundwater model and an intragravel temperature
D"obe
e\zaluated as tools to identify areas of groundwater within the rive were
re
W§re inconclusive as a means of identifying groundwater. Groun he methads

. a .
a D pear to be related to redd density in the study reaches. te’ dld not

Steelhead preferred to construct redds in a substrate consisting f
0 s

. . mall
\a rge gravel, and small cobble particle sizes disproportionately“) gravel,
Cla

Y. silt
and large cobble. The selection of coarse particle Sizes Small en . sand,
0Ug

h to mOVe

111

‘4
_—
_

 

 

and large enough to allow water to flow through the redd were consistent pacific
0038i salmonids. In addition, similar to Pacific Coast steelhead, Pere Marquette
River Steelhead exhibited a preference for stream velocities faster than normal,
and 3 Preference for water shallower than normal (99” riffles) within the study
reaches- Although stream temperature influenced the upstream migration 0f
steelhead, it did not appear to inflUence their selection of redd locations.

identifyi ng these spawning habitat features and redd densities within eaCh study

reach served as a description of steelhead spawning habitat preference within a
Great Lakes tributary, and may help to guide future monitoring of steelhead
ence

numbers within the river. My study results C0uld be used to monitor the influ
of habitat improvement projects that regularly occur in the watershed on

steelhead reproduction.

112

 

 

REFERENCES

113

REFERENCES

Alabaster, J. S. 1970. River flow and upstream movement and catch of
migratory salmonids. Journal of Fish Biology 2:1-13.

Bailey, M. M. 1969. Age, growth, and maturity of the longnose sucker,
(Catostomus catostomus), of western Lake Superior. Journal of the
Fisheries Research Board of Canada 26:1289-1299.

Baker, M. E., M. J. Wiley, and P. w. Seelbach. 2000. Spatially-exis"Clt modes
of groundwater loading in glaciated landscapes: considerations and
development in Lower Michigan. Michigan Department of Natural
Resources Research Report, Ann Arbor, MI.

Banks, J. w. 1969. A review of literature on the upstream migration 0‘ adu't
salmonids. Journal of Fish Biology 1 :85-136.

Behnke, R. J. 1992. Native trout of western North America. American Fisheries
Society Monograph 6.

d
Berry, D. A., and B. W. Lindgren. 1996. Statistics theory and (“eti'EOdS 2"
edition. Wadsworth Publishing Company, Belmont. California.

Beschta, R.L., R.E. Bilby, G.W. Brown, L.B Holtby, T.D. Hofstra. 1987. Chapter
6. Stream temperature and aquatic habitat. In: Salo, EC, and T.W.
Cundy, editors. Streamside Management: Forestry and Fishery
Interactions. University of Washington, Institute of Forest Resources.
Contribution No. 57. P. 191-232.

Burger, Carl V., M. Scott, M. Small, and W. Potterville. 1983. Overwintering and
spawning areas of steelhead trout (Salmo gairdnen) in tributaries of the
Upper Copper River, Alaska. Final Report. Alaska Department of Fish
and Game. Glennallen, Alaska.

Burger, C. V., R. L. Wilmot, and D. B. Wangaard. 1984. Comparison of
spawning areas and times for two runs of Chinook salmon (Oncorhynchus
tshawytscha) in the Kenai River, Alaska. Canadian Journal of Fisheries
and Aquatic Science 42:693-700.

Bras, R. L. 1990. Hydrology. Addison-Wesley PUb'iShiW. "IQ. Reading,
Massachusetts.

Brown, c. J. o. 1944. Michigan stream - their lengths. distribution and drainage
areas. Misceiianeous Publication #1, Institute for Fisheries Research,
Fisheries Division, Micl'ilgan Department Of Natural Resources, Ann Arbor.

114

 

 

Clapp, D. F., R. D. Clark, and J. S. Diana. 1990. Range, activity and home
range of large, free-ranging brown trout in a Michigan stream.
Transactions of the American Fisheries Society 119:1022-1034.

Cooper, AC 1965. The effect of transported stream sediment on the survival of
sockeye and pink salmon eggs and alevin. International Pacific Salmon
Fisheries Commission, Bulletin 18.

Curry, RA, and D.L.G. Noakes. 1995. Groundwater and the seleotion 0f
spawning sites by brook trout (Salve/inus fontinalis). Canadian Journal of
Fisheries and Aquatic Sciences 52:1733-1740.

Curry, RA, and D.L.G. Noakes, G.E. Morgan. 1995. Groundwateroand
the incubation and emergence of brook trout (Salve/inus fontinalis).
Canadian Journal of Fisheries and Aquatic Sciences 52:1741-1749.

Dexter, J. L., Jr., and N. D. Ledet. 1997. Estimates of fish passage .at St.
Joseph River in 1993 using time-lapse video recording. Michigan
Department of Natural Resources, Fisheries Technical Report 95-4. Ann
Arbor, Michigan.

 

Doerzbacher, J. F. 1980. Movement and home range of Iargemouth bass
(Micropterus salmoides) in relation to water qgeilty 0f the Atchafalaya
River Basin, Louisiana. Master's thesis. LOUISIana State University.

Baton Rouge.

Eddy. 3., and T. Surber. 1947. Northern fishes. University of Michigan Press.
Ann Arbor, MI

Ellis, D. V. 1962. Preliminary studies on the visible migrations of adult salmon.
Journal of Fisheries Research Board of Canada 19(1):137- 143,

Fraser, J.M. 1985. Shoal spawning of brook trout, Salve/inus fontina/is in a
precambrian shield lake. Le Naturalste Canadien 112:163-174.

Fukushima, M., and W.W. Smoker. 1998. Spawning-habitat segregation of
sympatric sockeye and pink salmon. Transactions of the American

Fisheries Society 127:253-260.

Geen, H.G., T. G. Northcote, G- F- Hartman, and C- C- LiNdSey. 1966. Life
histories of two species of catostomid fishes in Sixteenmile Lake, British
Columbia, with particular reference to inlet stream spawning. The Journal
of Fisheries Research Board of Canada. 23(11): 1761-1788.

115

 

 

 

Giorgi, A. E., T. W. Hillman, J. R. Stevenson, S. G. Hays, and C. M. Peven.
1997. Factors that influence the downstream migration rates of juvenile
salmon and steelhead through the hydroelectric system in the Mid-
Columbian River Basin. North American Journal of Fisheries
Management 1 72268-282.

Groves, PA, and J.A. Chandler. 1999. Spawning habitat used by fall Chinook
salmon in the Snake River. North American Journal of Fisheries
Management 19:912—922.

Harbeck, J. R. 1999. Relative contribution and comparative life history .
characteristics of hatchery and wild steelhead trout in the BetSie River,
Michigan. Master’s thesis, Michigan State University, East Lansmg.
Michigan.

H«Ell/hes, J. M., D. C. Nettles, D. M. Parnell,- M. P. Voiland, R. A. Olson, and J. D.
Winter. 1986. Movements of rainbow steelhead trout (Salmo gairdnen) In
Lake Ontario and a hypothesis for the influence of spring thermal
structure. Journal of Great Lakes Research 12(4):304—313.

Hellawell, J. M. H. Leatham, and G. I. Williams. 1974. The upstream migratory
behaviour of salmonids in the River F rome, Dorset. Journal of Fish
Biology 6:729-744.

Hicks, B. J., and R. R. N. Watson. 1985. Seasonal changes in abundance 01
brown trout (Salmo trutta) and rainbow trout (S. Gairdnen'i) assessed by f
drift diving in the Rangitikei River, New Zealand. New Zealand Journal 0
Marine and Freshwater Research 19 :1-10.

Hinch, S. G., and P. S. Rand. 1998. Swim speeds and energy use of upriver-
migrating sockeye salmon (Oncorhynchus nerka): role of local
environment and fish characteristics- Canadian Journal of Fisheries and
Aquatic Science 55:1821-1831.

Hooton, RS. and MG. Lirette. 1 986. Telemetric studies of winter steelhead,
Gold River, 1982-1983. Fisheries Management Report No. 86. Ministry of
Environment, British Columbia.

Jensen, A. J., T. G. Heggeberget and B' 0' Johnsen. 1.986“ Upstream migration
of adult Atlantic salmon. Salmo salar L., In the River Vefrsna, northern
Nonivay. Journal of Fish Biology 29:459-465.

Jonsson, N. 1991. Influence 0f water flow, water temperature and light On fish
migrations in rivers. Nordlc Journal of Freshwater Research 66:20-35.

116

 

Kogl. DR. 1965. Springs and ground-water as factors affecting survival of chum
salmon in a sub-arctic stream. Masters thesis, University of Alaska,

College, Alaska.

Kondolf, MG. 1988. Salmonid spawning gravels: a geopmorphic perspective
on their size distribution, modification by spawning fish and criteria for
gravel quality. PhD. thesis, Johns Hopkins University, Baltimore.

Maryland.

Kondolf, G. M., and S. Li. 1992. The pebble count technique for aging???

surface bed material size in instream flow studies. Rivers 3 2

er use and harvest in

Kruger, K.M., W.W. Ta lor and JR. R . 85. An I . .
Y yokman 19 gMichigan Ac ademICIan

the Pere Marquette River near Baldwin, Michigan.
17(3): 317-330.

on groundwater flow using

Lee, DR, and JA. Cherry. 1978. A field exercise . .
rnal of Geological Education

seepage meters and mini-piezometers. Jou
27: 6-10.

Leggett, W. C. 1977. The ecology of fish migrations. Annual Review of Ecology
and Systematics 8:285—308.

 

tions by repeat steelhead

Leider. 1985. Precise timing of upstream migra
Fisheries Society 1 14:906-

spawners. Transactions of the American

908.

Littell, R.C.. G.A. Milliken, W.W. Stroup, and RD. Wolfinger, 1996, SAS system
for mixed models, SAS Institute Inc., Cary, North Carolina

Lorz, H. W.. and T. G. Northcote. 1965. Factors influencing stream location, and
timing and intensity of entry by spawning kokanee (Oncorhynchus nerka)
into an inlet of Nicola Lake, British Columbia. Journal of the Fisheries

Research Board of Canada 22:665—687.

Lucas, M. C., and E. Batley. 1996. Seasonal movements and behaviour of adult
barbel Barbus barbus, a riverine cyprinid fish: Implications for river
management. Journal of Applied ECO'OQY 33313454353-

MDNR/IFR (Michigan Department of Natural Resources Institute for Fisheries
Research). 1998. Streambed gradient data for the mainstem Pere

Marquette River.

Menzies, W. J. M. 1939. Conference on Salmon Problems. Pages 100-101 in
F, R. Moulton, editor. American Assomation for the Advancement of

Science.

117

 

Miller, B. 1974. The spawning migration and age and growth studies of
steelhead in the Huron River, Baraga County, Michigan 1966-1970.
Michigan Department of Natural Resources Fisheries Division Technical
Report 74-10, September 1974.

Morman, R. H., D. W. Cuddy, and P. C. Rugen. 1980. Factors influencing the
distribution of sea lamprey (Petromyzon man'nus) in the Great Lakes.
Canadian Journal of Fisheries and Aquatic Science 3721811 -1826.

Muir, W. D., W. S. Zaugg, A. E. Giorgi, and S, McCutcheon. 1994. Accelerating
smolt development and downstream movement in yearling chIrIOOk
salmon with advanced photoperiod and increased temperature.
Aquaculture 123:387-399.

Neilson, JD.. and C. E. Banford. 1983. Chinook salmon (OncoanChUS
tshawytcha) spawner characteristics in relation to redd phySIcal features.
Canadian Journal of Zoology. 61 :97-1 10.

Neter, J., and W. Wasserman. 1974. Applied Linear Statistical Models, Richard

D. Irwin, lnc., Homewood, Illinois.

Northcote, T. G. 1962. Migratory behaviour of juvenile steelhead trout, Salmo
gairdnen', in outlet and inlet streams of Loon Lake, British Columbia.
Journal of the Fisheries Research Board of Canada 19:201-270.

Olsen, JG. 1968. Physical environment and egg devel0pment in mainland
beach area and island beach area of llimana Lake. In Further Studies 01
Alaska sockeye salmon, ed. R.L. Bu rgner. University of Washington
Publication of Fisheries News. Series 3,

Peters, J. C., H. R. Farmer, and P. J. Redford. 1973. A simulation model of the
upstream movement of anadromous salmonid fish. Reading Water

Resources Board 21.

Power, G. 1981. Stock characteristics and catches of Atlantic salmon (Salmo
salar) in Quebec, and Newfoundland and Labrador in relation to
environmental variables. Canadian Journal of Fisheries and Aquatic

Science 38:1601-1611.

Power, J. H., and J. o. McCleave. 1980. Riverine movements of hatchery-
reared Atlantic salmon (Salmo salar) upon return as adults

Environmental Biolon of Fishes 5(1): 3-13.

Press, W.H., S. A. Teukolsky, W. T. Vetterling. and B. P. Flannery. 1992.
Numerical recipes in C: the art of soientific compUting, second edition.
Cambridge University Press, New York.

118

 

Ratkowsky, D. A. 1983. Nonlinear regression modeling, a unified practical
approach. Marcel Dekker, Inc. New York and Basel.

Raymond, H.L. 1979. Effects of dams and impoundments on migrations of
juvenile chinook salmon and steelhead from the Snake River 1966 to
1975. Transactions of the American Fisheries Society 108: 505-529.

Rozich, T.J. 1989. Evaluation of the Pere Marquette River electrical lamprey
barrier. Unpublished report, Michigan Department of Natural Resources —
Lansing. Michigan.

gged brown trout

Schulz, U., and R. B . 19 2. Mov 'call ta
erg 9 ements of ultrasoni y logy 40:90991 7.

(Salmo trutta L.) in Lake Constance. Journal of Fish Bio
Seber, G. A.. and C. J. Wild. 1989. Nonlinear regression, Wiley. New York.

Semko, R. S. 1954. The stocks of west Kamchatka salmon and their
commercial utilization. lzv. Tikhookean. Nauchno- lssled. Inst. Rbn.
Khoz. Okeanogr. 41:3-(Translated from Russian; Fisheries Research
Board of Canada Translations Series 288).

Shepard. M. F. 1972. Timing of adult steelhead migrations as influenced br’y flOW
and temperature in four representative Washington streams. Maste S
thesis, University of Washington, Seattle, Washington.

Shepherd, B.G., G.F. Hartman, and W.J. Wilson. 1986. Relationships
between stream and intragravel tem peratures in coastal drainages, and
some implications for fisheries workers. Canadian Journal of
Fisheries and Aquatic Sciences 43:1 818-1822.

Silliman, SE, and DP Booth. 1993. Analysis of time-series measurements of
sediment temperature for identification of gaining vs. losing portions of
Juday Creek, Indiana. Journal of Hydrology 146:131-148.

Smith, A. K. 1973. Development and application of spawning velocity and depth
criteriea for Oregon salmonids. Transactions of the American Fisheries

Society 102:312-316.

Smith, 8. R., and J. J. Tibbles. 1980. Sea lamprey (Petromyzon marinas) in
Lakes Huron, Michigan, and Superior: history of invasion and control
19354973. Canadian Journal of Fisheries and Aquatic Science 37:1730.

1936-1979. 1801.
Smith, G. w., i. F. Smith. and S. M- Armstrong. 1994. The relationship between

river flow and entry to the Aberdeenshire Dee by returning adult Atlantic
salmon. Journal of Fish Biology 45:953-960.

119

 

 

 

Snell, A. J. 2001. Migratory behavior and angling of steelhead in the Pere
Marquette River, Michigan. Master’s thesis. Michigan State University,

East Lansing, Michigan.

Sowden, T. K., and G. Power. 1985. Prediction of rainbow trout embryo survival
in relation to groundwater seepage and particle size of spawning
substrates. Transactions of the American Fisheries Society 114: 804-31 2.

Stuart, T. A. 1953. Water currents through permeable gravels and their
Significance to spawning salmonids, etc. Nature 172: 407-408.

SWink. W. D. 1999. Effectiveness of an electrical barrier in blocking sea lamprey
migration on the Jordan River, Michigan. North American Journal 0
Fisheries Management 19:397-405.

Trepanier, 3.. M. A. Rodriguez, and P. Magnan. 1996. Spawning migrationsdin
landlocked Atlantic salmon: time series modeling of river discharge an
water temperature effects. Journal of Fish Biology 48:925-936.

Vanderploeg. H. A., and D. Scavia. 1979. Two electivity indicies for feeding with
special reference to zooplankton grazing. Journal of the Fisheries
Research Board of Canada 36:363-365.

Wentworth, C. K. 1922. A scale of grade and class terms for clastic sediments-

Journal of Geology 30: 377-392.

White. 0.8. 1990. Biological relationships to convective flow patterns within
stream beds. Hydrobiologia 196:149-158.

White, D.S., C.H. Elzinga, and SP. Hendricks. 1987. Temperature patterns
within the hyporheic zone of a northern Michigan river. Journal of the

North American Benthological Society 6(2):85-91.

White, E. M., and B. Knights. 1997. Environmental factors affecting migration of
the European eel in the Rivers Severn and Avon, England. Journal of
Fish Biology50:1104-1116.

Winter, J. D. 1976. Movements and behaviour of largemouth bass (Micropterus
salmoides) and steelhead (Salmo galrdn'en) determined by radio
telemetry. Ph. D. thesiS. University Of Mlnnesota, Minneapolis, MN 195 p.

Winter, J. 1996. Advances in Underwater Biotelemetry. Pages 570-599 in B. R.

Murphy and D. W. Willis, editors. Fisheries techniques, 2nd edition.
American Fisheries Society. Bethesda, Maryland.

120

 

Workman, R. D. 1994. The spawning and movement of largemouth bass
(Micropterus salmoides) in Reelfoot Lake, Tennessee. Master’s thesis.

Murray State University, Murray, Kentucky.

Zabel, R. W., J. J. Anderson, and P. A. Shaw. 1998. A multiple-reach model
describing the migratory behavior of Snake River yearling chinook salmon
(Oncorhynchus tshawytscha). Canadian Journal of Fisheries and Aquatic

Science 55:658-667.

Zar, J. H. 1984. Biostatistical Analysis, second edition. Prentice Hall.
Englewood Cliffs, New Jersey.

121