, f'a‘

at” z

-. L
’_——Al_

.1. ‘-
T44

.‘j :l LS"- "I“
‘3"

u
out!“ V

 

Siﬁ

5193"

a

 

 

9:1

.\

, 3,;
mm
1‘,

. 3.3":1;}"
_, 3AA

-u‘w
Q ~

' at"
i};

1'". 1“ 111‘? ‘ 3‘
mm

 

THES'S

‘&

UNIVERSITY LIBRARIES

IIIIIIIIIIIIIIIIIIIIII II IIIIII I

31293 01712

 

 

 

 

 

 

 

 

 

 

 

This is to certify that the

dissertation entitled

Juvenile Steelhead Production in the
Betsie River Watershed

presented by

Tammy Joann Newcomb

has been accepted towards fulﬁllment
of the requirements for

Ph.D. degreein Fish. & Wildl.

 

 

Major Efessor

Date April 27. 1998

MS U is an Afﬁrmative Action/Equal Opportunity Institution 0- 12771

n

 

LIBRARY

, MIChIQan State
Unlverslty

 

 

 

PLACE IN RETURN BOX
to remove this checkout from your record.
TO AVOID FINES return on or before date due.

 

MTE DUE DATE DUE MTE DUE

 

{31' 2 6 3031

 

 

 

{Vol I b Yum

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1/98 chIFICJDubDuopGS—p.“

PRODUCTIVE CAPACITY OF THE BETSIE RIVER WATERSHED
FOR STEELHEAD SMOLTS

3)!

Tammy Joann Newcomb

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

DOCTOR OF PHILOSOPHY

Fisheries and Wildlife Department

1 998

ABSTRACT

PRODUCTIVE CAPACITY OF THE BETSIE RIVER WATERSHED
FOR STEELHEAD SMOLTS

BY

Tammy Joann Newcomb

Steelhead (Oncorhynchus mykiss). a Paciﬁc salmonid and popular sportﬁsh,
were introduced into Michigan rivers in 1864. Since their initial stocking, they have
become naturalized in many rivers and produce wild offspring. Knowledge of the
abundance of wild juvenile steelhead and the factors inﬂuencing their mortality is
important when considering the management options for harvest and sustainability of
the ﬁshery. Temperature can be an important inﬂuence on the distribution. growth, and
mortality of juvenile steelhead. The Betsie River in northwestern Michigan is a marginal
trout stream and temperatures regularly exceed the growth and lethal limits for juvenile
steelhead. Objectives for this study included: estimating the number of juvenile
steelhead leaving the Betsie River and comparing the results of three monitoring
techniques. describing the distribution. density, and mortality of juvenile steelhead in
the Betsie River and its tributaries, describing the thermal regime in the watershed, and
developing temperature models to evaluate the effects of management alternatives for
lowering instream temperatures in the summer. I used mark-recapture, visual
observation, and time-lapse videography to monitor and estimate the number of

juvenile steelhead emigrating from the Betsie River each spring, 1993-1996. I used

stratiﬁed random sampling (18 sites) to determine the distribution and abundance of
juvenile steelhead in the Betsie River and its tributaries. In July and October, 1993-
1996, I used electroﬁshing and multiple pass depletion methods to estimate the number
of juvenile steelhead. Continuous recording data loggers monitored hourty stream
temperatures at 7 sites in the watershed. Mean daily temperatures were calculated to
evaluate the thermal regime and empirical and physical process models were
developed to predict instream temperature in the Betsie River and evaluate dam
removal or hypolimnetic siphon options. The number of juvenile steelhead leaving the
Betsie River ranged from 1,143 :t 266 in 1996 to 2,151 :I: 286 in 1993. Mark-recapture
over-estimated the number of ﬁsh leaving the river while visual observations and time-
lapse videography gave similar results. Juvenile steelhead were found in the greatest
densities in the tributaries and lower main channel Betsie River. In terms of total
abundance. approximately 50% of the juvenile ﬁsh were found in the tributaries which
comprised only 11% of the total channel area considered. The density of juvenile
steelhead was negatively correlated with daily mean maximum summer temperature
and mortality was positively correlated with daily mean minimum winter temperature.
The number of juveniles emigrating in spring was negatively correlated with prior winter
severity. Removal of a dam in the Betsie River will result in lower instream
temperatures, but they would still exceed the optimal growth and preferred
temperatures for juvenile steelhead. A hypolimnetic siphon would reduce instream

temperatures to within the preferred and optimal growth limits for the main channel.

ACKNOWLEDGMENTS

This project received funding and support from the Department of Fisheries and
Wildlife at Michigan State University, Michigan Department of Natural Resources
Fisheries Division. Michigan River Guides Association, Trout Unlimited. Kalamazoo
Valley Chapter of Trout Unlimited, and West Michigan Chapter of Trout Unlimited. The
following chapters of the Michigan Salmon and Steelhead Fishermen’s Association also
contributed generously to the housing fund: Detroit Area, Battle Creek. Thumb Area.
Saginaw Valley, Midland Area, Mid-Michigan. Holland Area, Southwestern Michigan,
Benzie Area, Grand Rapids, Flint River Valley. and Metro-West Steelheaders. I also
thank the Challenge Chapter of Trout Unlimited for the donation of temperature
monitoring devices that were integral for several facets of my research. Fred and Carin
Stunn. proprietors of Homestead Resort in Benzonia, provided reasonable cost housing
and tolerated our odd working hours during smoltwatch.

I am appreciative of the advice and guidance proffered by my committee
members, Dr. Richard Menitt, Dr. William Taylor, Dr. Graham Larsen, Dr. Paul
Seelbach, and my advisor Dr. Thomas Coon. Dr. Scott VWnterstein and Dr. Jim Bence
graciously provided statistical insight.

I thank John Westley and Jim Vandennaas for their advice and guidance on the
liver and lake regarding sampling methods for steelhead. May I live to one day outﬁsh
them both. Insight and assistance in the early years of this project was graciously

bestowed upon me by Dr. Howard Tanner.

iv

I am grateful to several wonderful volunteers who braved the elements of winter
to collect staff gage. Thank you Dick Gillespie, Dick Mills, Lew Tipper, Dr. Chuck
Nelson, and Chuck Barnes.

Every year, my ambitious ﬁeld schedule was accomplished mainly due to the
dedicated efforts of my summer crews who toiled tirelessly to get the work done.

Those who know that it really is the darkest and coldest before the dawn include: Mike
Wilson. Dan Weisenreder. James Ladonski, Jon Holmes. Scott Laprad, Joe Kozﬂtay,
and Troy Weaver. Also, thanks to Danielle Fletcher, Aaron Snell, Vince Videan, and
many others who volunteered with electroﬁshing.

Finally. my heartfelt appreciation goes to my husband John, for his support and

patience through the last 5 years and for cheerfully tolerating a “weekend wife”.

TABLE OF CONTENTS

LIST OF TABLES ..................................................................................................... viii
LIST OF FIGURES .................................................................................................... x
INTRODUCTION ....................................................................................................... 1
Steelhead in Michigan ...................................................................................... 1
Life History of Steelhead ................................................................................. 3
Temperature Modeling ................................................................................. 8
Betsie River Watershed .................................................................................. 10
Dissertation Overview ...................................................................................... 14
CHAPTER 1
EVALUATION OF ALTERNATE METHODS FOR ESTIMATING NUMBERS OF
OUTMIGRATING STEELHEAD SMOLTS ................................................................. . 16
Abstract .......................................................................................................... 16
Introduction ...................................................................................................... 17
Study Site ......................................................................................................... 20
Methods ............................................................................................................ 23
Results ............................................................................................................. 32
Discussion ....................................................................................................... 38
Conclusions and Management Implications .................................................... 48
CHAPTER 2
DISTRIBUTION. DENSITY, AND SURVIVAL OF STEELHEAD PARR
IN A THERMALLY DIVERSE WATERSHED ............................................................... 49
Abstract ............................................................................................................ 49
Introduction ...................................................................................................... 50
Study Site ........................................................................................................ 52
Methods ........................................................................................................... 54
Results ............................................................................................................. 58
Reach Characteristics ........................................................................... 58
Water Temperature and Environmental Variability ............................... 59
Abundance and Distribution of Parr Throughout the Watershed .......... 62
Number of Outmigrating Smclts ........................................................... 72
Mortality Rates ...................................................................................... 73
Steelhead Density, Mortality Rates, and Temperature Variables ......... 75
Observed Differences in Fish Lengths ................................................. 78
Discussion ........................................................................................................ 85
Conclusions ..................................................................................................... 94

vi

CHAPTER 3
ASSESSMENT OF MANAGEMENT ALTERNATIVES FOR ALTERING THE

THERMAL REGIME THROUGHOUT THE BETSIE RIVER WATERSHED ................. 95
Abstract ........................................................................................................... 95
Introduction ...................................................................................................... 97
Methods .......................................................................................................... 103
Results ........................................................................................................... 1 1 1

Stream Network Temperature Model Results ..................................... 120
Stream Segment Water Temperature Model Results ......................... 127
Discussion and Conclusions .......................................................................... 137

SUMMARY ................................................................................................................ 142

APPENDIX A ............................................................................................................. 150

APPENDIX B ............................................................................................................. 153

BIBLIOGRAPHY ........................................................................................................ 156

vii

Table 1.

Table 2.

Table 3.

Table 4.

Table 5.

Table 6.

Table 7.

Table 8.

Table 9.

Table 10.

Table 11.

Table 12.

LIST OF TABLES

Duration of observations and smolt movement at Homestead Dam,
Betsie River, Michigan, 1993-1996 ....................................................... 32

Peterson mark-recapture estimates of the number of steelhead in the
Betsie River smolt run ........................................................................... 37

Estimate of the number of smolts from 20 minute direct observation
sampling and continuous time-lapse videography at the Homestead
Lamprey Weir, Betsie River. Michigan .................................................. 38

Hydrological and morphological characteristics of reaches sampled
in the Betsie River watershed ................................................................ 58

Summer and winter air temperature from the Cadillac, Michigan
weather station ...................................................................................... 59

Mean summer and winter water temperature in the Betsie River at
Homestead Weir and Little Betsie River, 1993-1996 ............................ 61

Summer and winter water temperatures in reaches sampled in
the Betsie River watershed ................................................................... 62

Density estimates and 95% conﬁdence intervals for age-0 and

age-1 steelhead parr at sites sampled by electroﬁshing and

depletion methods throughout the Betsie River watershed,

1993-1996 ............................................................................................. 65

Density (pan/ha) of age-O and age-1 steelhead parr in the upper
(1-4) and lower (5-8) reaches of the Betsie River, October
1994-1995 ............................................................................................. 66

Watershed abundance estimates and 95% conﬁdence intervals

for age-0, age-1, and age-2 steelhead parr calculated from

stratiﬁed reach estimates throughout the Betsie River watershed,

July, 1993-1996 .................................................................................... 72

Age structure and watershed yield of wild steelhead smolts
emigrating from the Betsie River watershed, May-June. 1993-1996 ..... 73

Annual, winter, and summer instantaneous mortality rates (Z) for
steelhead parr cohorts observed in the Betsie River watershed,
1993-1996. The annual mortality rate includes an adjustment

for the cohort’s emigrating smolts ......................................................... 74

viii

Table 13.

Table 14.

Table 15.
Table 16.

Table 17.

Table 18.

Table 19.

Table 20.

Table 21.

Table 22.

LIST OF TABLES (continued)

Annual instantaneous mortality rates (Z) throughout the Betsie River
watershed without adjustment for outmigrating smolts. 1993-1996 ...... 75

Mean lengths and 95% conﬁdence intervals for age-0 and age-1
steelhead parr sampled from four reach categories throughout
the Betsie River watershed, 1993-1996 ................................................ 84

Comparison of steelhead parr densities in select streams .................... 88

Sites and reaches used for monitoring flow and temperature
in the Betsie River watershed .............................................................. 104

Input ﬁles and associated parameters required for the stream
network model ..................................................................................... 108

Mean water temperature by season and year measured at sites
in the Betsie River and its tributaries ................................................... 112

Probability of exceeding, meeting, or remaining below preferred

and optimal growth temperatures for juvenile steelhead at sites
throughout the Betsie River watershed. Calculated from mean

daily temperatures recorded in water years 1993-1996 ...................... 114

Mean error at 3 validation nodes in the Betsie River and Little
Betsie River resulting from 5% increases and decreases
in a single parameter ........................................................................... 120

Mean, probable, and maximum error for the stream network
model predictions at 3 validation sites in the Betsie River
watershed ........................................................................................... 123

Mean daily water temperature in June, July, and August for

the river at the downstream section of Segments A and B, in

the Betsie River under current conditions and under simulated
conditions of removing Grass Lake Dam and releases of

hypolimnetic water from Green Lake at Grass Lake Dam. Data

from sites at M-115, Homestead, and the Little Betsie River

are also included for comparison ........................................................ 135

Figure 1.

Figure 2.

Figure 3.

Figure 4.

Figure 5.

Figure 6.

Figure 7.

Figure 8.

Figure 9.

Figure 10.

Figure 11.

LIST OF FIGURES

Fall and spring adult steelhead returns to the Little Manistee River
weir, Michigan, 1968.1997 (Data obtained from the Michigan
Department of Natural Resources, Fisheries Division, Lansing .............. 2

Location and map of the Betsie River watershed in northwestern
Michigan ................................................................................................ 1 1

Location and map of the Betsie River watershed in northwestern
Michigan ................................................................................................ 22

Diagram of equipment used at the Homestead Lamprey Weir to
monitor emigrating smolts in the Betsie River, Michigan ....................... 24

Constriction weir and block net design for capture of emigrating
smolts in the Betsie River, Michigan ...................................................... 30

Number of steelhead smolts stocked. estimated number that
remained as residuals in the river. and resulting number of potential
migratory steelhead smolts in the Betsie River, Michigan, 19934996.. 31

Temporal distribution of outmigrating steelhead smolts in the Betsie
River. Michigan. Hourly values represent the mean for observations
during that time period .......................................................................... 34

Species composition of the juvenile salmonids captured at
Homestead Weir, Betsie River, Michigan, 1993-1996 ........................... 35

Comparison of mark-recapture estimates and 95% conﬁdence
intervals of steelhead smolt numbers with and without adjustment
for residual steelhead in the Betsie River, Michigan, 1993-1996 .......... 36

Comparison of the number of outrnigrating smolts and 95% conﬁdence
intervals calculated from variable sampling efforts ranging from 100%
to 20% of the days during the smolt run ................................................ 39

A comparison of the standard error of number of outrnigrating
steelhead smolts as affected by sampling effort. Fitted line is a
logarithmic relationship with r‘ = 0.7697 ................................................ 40

Figure 12.

Figure 13.

Figure 14.

Figure 15.

Figure 16.

Figure 17.

Figure 18.

Figure 19.

Figure 20.

Figure 21.

Figure 22.

Figure 23.

LIST OF FIGURES (continued)

Comparison of the steelhead smolt numbers determined from
mark-recapture (MR), 20 minute observations (20 Min) and

continuous recording (Video) in the Betsie River, Michigan,

1993-1996. Error bars represent 95% conﬁdence intervals ................. 42

Sampling sites for electroﬁshing in the Betsie River watershed,
Michigan, 1993-1996 ............................................................................. 53

Mean discharge measured in July at sampling reaches in the
Betsie River watershed, 1993—1996 ...................................................... 60

Length-frequency distribution of steelhead parr captured in July
in the Betsie River watershed, 1993-1996 ............................................ 63

Length-frequency distribution of steelhead parr captured
throughout the Betsie River watershed in October, 1994-1996 ............. 64

Comparison of age-0, age-1, and age-2 steelhead parr densities
between the Betsie River and its tributaries, July, 1993-1996 ............... 67

Comparison of the densities of age-0 and age-1 steelhead parr
in the upper and lower reaches of the Betsie River, 1993-1996 ........... 68

Estimate of abundance of steelhead part in 4 reaches throughout
the Betsie River watershed, July 19934996 ......................................... 70

Comparison of abundance of steelhead parr in the Betsie River
and four tributaries that were sampled in the watershed,
July 1993—1994 ..................................................................................... 71

Correlation between the density of age-0 steelhead parr and the
maximum summer water temperature at sites throughout the

Betsie River watershed, 1993-1996 (adjusted r2 =0.4981,

p = 0.00002) .......................................................................................... 76

Correlation between the density of age-1 steelhead pan and the
maximum summer water temperature at sites throughout the

Betsie River watershed, 1993-1996 (adjusted r2 = 0.6225,

p = 0.000007) ........................................................................................ 77

Correlation between the density of age-1 steelhead parr and the
minimum water temperature at sites throughout the Betsie River
watershed, 1993-1996 (adjusted r2: 0.339‘, p = 0.0275) ...................... 79

xi

LIST OF FIGURES (continued)

Figure 24. Correlation between mean winter temperature and the annual
mortality rate for steelhead parr (age-0 to age-1) at sites
throughout the Betsie River watershed, 1993-1996
(adjusted r’ = 0.389, p = 0.0581) ........................................................... 30

Figure 25. Correlation between the winter severity index (number of days
when air temperature is less than -12°C) and the estimate of
smolts leaving the Betsie River watershed, 1993-1996
(adjusted r’=0.946,p = 0.019) ................................................................ 91

Figure 26. Correlation between the abundance of age-0 steelhead in July
and the mean May and June discharge in the Betsie River
watershed, 1993-1996 (adjusted r’ = 0.771, p = 0.079) ........................ 82

Figure 27. Comparison of mean monthly water temperature among 6 sites
in the Betsie River watershed. The gray shaded area represents
the optimal growth range for juvenile steelhead, the heavy
dashed line is the upper zero growth level (23°C), the light
dashed line is the lower lethal level (0°C) .............................................. 86

Figure 28. Location of the Betsie River watershed and temperature
monitoring locations (T) and ﬂow monitoring locations (H) .................... 98

Figure 29. Skeleton framework and nodes for the Betsie River Watershed
network temperature model. Nodes H = headwater,
C = change in shading or stream width, T = terminus of
tributary, J = junction below tributary, B = branch above
tributary, V = validation node, E = end of study reach ......................... 107

Figure 30. Temperature exceedence curves for sites in the Betsie River
watershed, water years 1993-1996. The large rectangle
indicates temperatures preferred by juvenile steelhead and the
smaller rectangle indicates the temperature range of optimum
growth of juvenile steelhead (Hokanson et al. 1977, Wismer
and Christie 1987) ............................................................................... 113

Figure 31. Spring water temperatures during steelhead smolt emigration
in the Betsie River watershed, 1994-1996. Shaded area
denotes smoltiﬁcation temperatures above which smolts may
revert to part ........................................................................................ 115

Figure 32. Predicted water temperature (symbols) as a function of air

temperature and mean daily water temperature (line) at sites in
the Betsie River, 1993—1996 ................................................................ 117

xii

Figure 32.

Figure 33.

Figure 34.

Figure 35

Figure 36.

Figure 37.

Figure 38.

Figure 39.

Figure 40.

LIST OF FIGURES (continued)

(continued). Predicted water temperature (symbols) as a function
of air temperature and mean daily water temperature (line) at
sites in the Betsie River, 1993-1996 .................................................... 118

Predicted water temperature (symbols) as a function of air
temperature and mean daily water temperature (line) at tributary
sites in the Betsie River watershed, 1993-1996 .................................. 119

Comparison of mean weekly observed water temperature
and network model predicted temperature at validation sites
in the Betsie River and Little Betsie River ........................................... 122

Longitudinal distribution of mean weekly water temperature

as predicted for the Betsie River during the fall and the start

of the water year, October-November in a normal flow year.

The shaded box is the preferred temperature range for juvenile
steelhead ............................................................................................ 124

Longitudinal distribution of mean weekly water temperature

as predicted for the Betsie River during winter, December-February,

in a normal-flow year. The shaded rectangle is the preferred
temperature range for juvenile steelhead ............................................ 125

Longitudinal distribution of mean weekly water temperature

as predicted for the Betsie River during spring, March-May,

in a normal ﬂow year. The shaded box is the preferred

temperature range forjuvenile steelhead ............................................ 126

Longitudinal distribution of mean weekly water temperature

as predicted for the Betsie River during summer, June-August,

in a normal-ﬂow year. The shaded rectangle indicates the

preferred temperature range for juvenile steelhead ............................ 128

Longitudinal distribution of mean weekly water temperature as

predicted for the Betsie River during September in a normal-ﬂow

year. The shaded rectangle indicates the preferred temperature

range for juvenile steelhead ................................................................ 129

Measured and predicted mean daily summer temperatures

from the stream reach temperature model for segment A, Green

Lake to Grass Lake, in the Betsie River during a typical-ﬂow (1994)

and low-ﬂow water year (1995). The large shaded area indicates

water temperatures preferred by juvenile steelhead and the inner
rectangle identiﬁes optimum growth temperatures for juvenile

steelhead (Hokanson et al. 1977, Wlsmer and Christie 1987) ............ 131

xiii

LIST OF FIGURES (continued)

Figure 41. Measured and predicted mean daily summer temperatures
from the stream reach temperature model for segment B, Grass
Lake to Thompsonville, in the Betsie River during a typical-ﬂow
(1994) and low-flow water year (1995). The large shaded area
indicates water temperatures preferred by juvenile steelhead
and the inner rectangle identiﬁes optimum growth temperatures
for juvenile steelhead (Hokanson et al. 1977, Wismer and
Christie 1987) ...................................................................................... 132

Figure 42. Comparison of mean daily summer water temperatures under
current conditions with Grass Lake Dam and simulated conditions
of dam removal in a typical-ﬂow (1994) and low-ﬂow (1995)
water year for Segment A. The large shaded area indicates
water temperatures preferred by juvenile steelhead and the inner
rectangle identiﬁes optimum growth temperatures for juvenile
steelhead (Hokanson et al. 1977, Wlsmer and Christie 1987) ........... 133

Figure 43. Comparison of mean daily summer water temperatures from
Grass Lake Dam to Thompsonville under current conditions
with Grass Lake Dam present and simulated conditions with
the dam removed for a typical-ﬂow (1994) and low-ﬂow (1995)
year. The large shaded area indicates water temperatures
preferred by juvenile steelhead and the inner rectangle identiﬁes
ptimum growth limits (Hokanson et al. 1977, Wismer and
Christie 1987) ...................................................................................... 134

Figure 44. Comparison of mean daily water temperatures simulated for the
dam removal and the addition of cold hypolimnion water in the
Grass Lake to Thompsonville reach of the Betsie River in a
typical-ﬂow (1994) and low-ﬂow (1995) water year. The large
shaded area is the water temperature range preferred by juvenile
steelhead and the smaller rectangle depicts optimum growth
ranges (Hokanson et al. 1977, Viﬁsmer and Christie 1987) ................. 136

xiv

INTRODUCTION

Steelhead In Michigan

Steelhead (Oncorhynchus mykiss) were introduced into Michigan rivers in the
late 1800's and anglers widely accepted them as a popular sportﬁsh. The ﬁrst
steelhead eggs were brought by private interests from California to Bay City, Michigan
in 1876 and the fry were stocked in the Au Sable River. The Michigan Fish
Commission received 2,000 eggs from the McCloud River in California in April 1880
that were incubated in the Pokagon Fish hatchery (Latta 1974). Steelhead have been
stocked annually throughout the Great Lakes since their introduction, but many
populations are naturalized and their progeny have the opportunity to contribute
signiﬁcantly to the sport ﬁshery (Latta 1974, Seelbach 1989, Rand et al. 1993, Peck
1992). In the Great Lakes, adult steelhead provide multiple and diverse ﬁshing
opportunities that include offshore ﬁshing in the summer, pier and surf ﬁshing in the
spring, summer, and fall, and instream ﬁshing for driftboat and wading anglers during
the spawning migrations in the spring and fall.

Naturalized steelhead production (the product of instream spawning) has
occurred in varying and largely unquantiﬁed levels since their introduction. Some
streams, such as the Little Manistee River are not stocked but they experience large
and highly variable returns of adults (Figure 1). ln contemporary ﬁsheries science in

the Great Lakes, managers and researchers are investigating these naturalized

*Fall SH-e--Spring SH

Number of Steelhead

 

 

5': , 2.
8000Ir 5 '. g ,-:
,9 2:? {Xe}: :.
'I a" : i g ' I i :
600°" ‘ 3 1 -‘ '. .’. 3 :
9' ' 5 ". ' '. a it " ;'
e ' ‘ . ‘,' ‘ . O
4000.. ,- : z: o ‘ ‘2:
: ' ' '
2000- :'
‘e'.
0. : 'r i : : i i
1968 1972 1976 1980 1984 1988 1992 1996
Year

Figure 1. Fall and spring adult steelhead returns to the Little Manistee
River weir, Michigan, 1968-1997 (Data obtained from the Michigan
Department of Natural Resources, Fisheries Division, Lansing).

populations and consider them a valuable resource for the maintenance and
sustainment of the steelhead ﬁshery. Furthermore, watershed capacities to produce
steelhead parr are increasing as a result of recent improvements in water quality,
efforts to decrease erosion and sand bedload, and instream habitat improvements
(Seelbach 1989, Rand et al. 1993). Production of juvenile steelhead can be very large
in streams of good quality trout habitat that contain cold thermal regimes, such as the
Little Manistee River (Seelbach 1993). Little is known however, about smolt production
from streams that contain habitat with limiting thermal regimes for salmonid species.
Many streams in southem Michigan that are classiﬁed as marginal host large runs of
adult steelhead, but very little is known about the quantities in the resulting smolt
cohort. Determination of the contributions of marginal streams to the lakewide
steelhead ﬁshery and knowledge of the conditions which may inhibit natural production
and successful recruitment of steelhead is necessary for the management of a

sustainable population of Great Lakes steelhead.

Life History of Steelhead

Adult steelhead begin immigrating into Great Lakes tributary streams in late
autumn (October-November) through the early spring (February-April) after 1 to 3 years
of residence in the lake (Biette et al. 1981, Seelbach 1993). Hypothetically, fall-run
adults that overwinter in the river may spawn earlier than the early-spring immigrating
adults. In Great Lakes tributary watersheds, the spaWning period lasts for 4 to 7.5
' weeks (Biette et al. 1981). Steelhead typically home and enter their natal stream to
spawn. In Great Lakes tributaries, studies indicate that rates of straying vary but can
be as low as 10% (Biette et al. 1981). Unlike coho (O. kisutch) and chinook salmon (O.

fschawchtya), post-spawning survival is probable and steelhead may return to the lake

and return to spawn in subsequent years (Seelbach 1993). Throughout the Great
Lakes, the proportion of repeat spawning ranges from 16% to 61% and the majority of
the repeat spawners are 2nd year spawners (Biette 1981, Seelbach 1989, Seelbach
and Miller 1993).

Adult female salmonids construct saucer shaped redds in gravel of 20-30 mm in
diameter, velocities greater than 15 cm/s, and depths between 5 and 50 cm. An
individual female may construct one or more redds and may deposit eggs more than
once per spawning season (Crisp and Carling 1989). Fecundity varies with the length
of the female and location of the ﬁsh stock. In general, a 57.1 cm ﬁsh will produce
between 2,424 and 4,964 eggs (Bulkley 1967). The eggs of rainbow trout require
approximately 50 days at 73°C or 25 days at 12°C to fully incubate and hatch (Bardach
et al. 1972). Water temperature can inﬂuence year-class strength throughout a
watershed as different hatching times may result from a difference in temperatures at
redd sites throughout the watershed (Bardach et al. 1972).

During stream residency, steelhead parr use a variety of microhabitat conditions
diumally, seasonally, and with increased size. Within the stream environment,
steelhead parr are subject to temperature extremes in both summer and winter and to
competition for space and food. Certain time periods during stream residency may act
as bottlenecks for their recruitment to the next year class. Water temperatures required
for parr growth range between 13 and 19°C with an optimum range of 17 to 19°C
(Wismer and Christie 1987). No growth occurs above 22°C or below 8°C and water
temperatures greater than 24°C are lethal (Hokanson et al. 1977).

Newly emerged steelhead fry inhabit shallow water, such as that found along
the stream margin or in shallow riffles, and they use crevices in logjams and boulders

as cover (Hartman 1965, Cambpell and Neuner 1984). In the fall, steelhead parr seek

shelter in log jams and boulders (Hartman 1965). Viﬁth increased size, steelhead parr
use deeper and faster water (Everest and Chapman 1972, Waite and Bamhart 1992,
Campbell and Neuner 1984). Age-0 parr prefer depths of less than 0.15 m and
velocities less than 0.15 m/s, whereas age-1 parr use depths from 0.60 to 0.75 m and
velocities 0.15 - 0.3 m/s (Everest and Chapman 1972).

Seelbach (1987) documented consistent production of age-1 steelhead parr in
the Little Manistee River, however the numbers of outrnigrating juvenile steelhead
varied from 10,000 to 72,000 ﬁsh per year. One hypothesis regarding the variability in
yield of smolts was in relation to environmental conditions, particularIy winter severity.
Overwinter mortality estimates were 13 to 90 % and a signiﬁcant relationship (r2 = 0.54)
existed between the severity of the presmolt winter and the number of returning adults.
Further evidence suggests that instream habitat for overwintering parr may be very
critical to the success of individual year classes (Bjomn et al. 1991). Critical winter
habitat for steelhead includes groundwater fed side channels and boulder or rubble
strewn stream margins with water less than 30 cm deep and very slow water velocities
(Everest and Sedell 1983, Campbell and Neuner 1984). Below 11°C, rainbow trout parr
tend to hide in the interstices under and between rocks and reduce their activity and
their selection of instream habitat may be crucial (Heam and Kynard 1986, Everest and
Sedell 1983). In Oregon streams, parr occupied low energy environments that were
less than 0.15 m deep and had water velocities that were less than 0.15 m/s. Parr
were more concentrated in the favorable habitats in winter than in summer (Everest
and Sedell 1983). In Michigan streams, where substrates are predominantly composed
of sand or small gravel, overwinter habitat in the form of cobbles and boulders may be
limited. Furthermore, low velocity areas that are not sufﬁciently buffered by

groundwater may experience ice and freezing conditions lethal to juvenile steelhead.

Steelhead parr spend 1 to 3 years in the river during which they grow to a size
suitable for smoltiﬁcation, typically greater than 180 mm (Seelbach 1993). During
smoltiﬁcation, juvenile steelhead metamorphose, becoming longer and thinner and
tuming a solid silver color as their parr marks diminish (Wedemeyer et al. 1980, Hoar
1988). Smolts begin to school and emigrate nocturnally from the river to the lake or
ocean (Brege et al. 1996). Temperature, photoperiod and discharge are the primary
environmental factors that inﬂuence the timing and rate of smoltiﬁcation and the
outbound movement of smolts from a watershed. The majority of smolt movement
occurs right after dusk. For example, from 6 years of observations of smolt movement
through the John Day Dam on the Columbia, Brege et al. (1996) found rates of
nocturnal movement to range from 66.2 to 87.4 % (mean 77.9%) of the total steelhead
smolts passing the dam. Similarly in Michigan, Stauffer (1968) found that over 93% of
migrating juvenile steelhead moved between 17:00 and 08:30 hours and peaks of
migration occur between May 21-June 30 in Lake Michigan tributaries (Stauffer 1968,
Biette et al. 1981). Steelhead smolts begin migrating when the water temperatures
approach 10°C (Stauffer 1968) and smoltiﬁcation behavior begins to decline when
water temperature warms to greater than 13°C (Zaugg et al. 1972, Zaugg and Wagner
1973, Zaugg 1981). Increases in stream discharge usually result in increased rates of
smolt movement and decreases in smolt travel time (Berggren and Filardo 1993,
Hvidsten and Johnsen 1993, Dempson and Stansbury 1991).

Size of steelhead smolts may also inﬂuence their survival and timing of
emigration. Steelhead smolts range in length from 120 mm to over 200 mm (Ward et
al. 1989, Seelbach 1993), and numerous studies report greater survival rates for larger
smolts (Seelbach et al. 1994, Ward et al. 1989). Survival rates for smolts greater than

180 mm are much higher (17- 33%) than 140 mm smolts (<5% )(Ward et al. 1989).

 

The majority of naturalized steelhead observed from rivers throughout the Great Lakes
smolted at the age of 2, and age 1 smolts were more prevalent than age 3 smolts
(Biette et al. 1981, Seelbach 1993).

Because smolt movement is predictable and concentrated through a short
window of time during the spring the annual event gives ﬁshery managers an
opportunity to examine the cohort before they disperse into the vast lake environment.
Estimation and observation of smolt emigrations provide an excellent opportunity to
obtain information on the future potential of the ﬁshery as well as the productive

capacity of the watershed.

Monitoring Techniques For Estimating Smolt Production

Several methods have been used with varied success to monitor and estimate
the abundance of emigrating smolts. Typical means of enumeration include: 1)
trapping either a portion or the entire run (Raymond 1979, Davis et al. 1980, Seelbach
et al. 1994), 2) videotaping through counting windows, 3) using a combination of
trapping and mark-recapture (Dempson and Stansbury 1991), or 4) use of resistivity
counters (Beach 1978, Reddin et al. 1992). Several of the trapping methods require
the use of fyke-nets, which can be problematic with debris loads, or construction of a
low head dam or similar apparatus which can be cosz to implement (Davis et al. 1980,
Seelbach et al. 1984). Fence counts which involve constraining ﬁsh movement through
an area where they can be counted are often used as reliable methods with adult
escapement estimates (Cousens et al. 1982). Success of the methods also varies with
discharge: typically as ﬂow increases, enumeration becomes less efﬁcient (Raymond

1979).

One method of trapping, the inclined screen smolt trap is a successful method
(Seelbach et al. 1984, Dubois et al. 1991). The method incorporates a dam or weir in
its design. An inclined screen is attached to the sill of the dam or weir and guides ﬁsh
into a ﬂoating barge trap. Drawbacks to this approach are that it requires an instream
structure, its efﬁciency is limited at higher ﬂows, and it requires handling of the
physiologically sensitive smolts.

Methods most desirable for enumerating smolts would feature minimum labor
requirements, cost effectiveness, reliable and accurate counts or estimates, and
would not require handling or other stresses to smolts. Time-lapse videography is a
technological approach that is effective for enumerating returning adult salmonids
(Dexter and Ledet 1994). As the cost for the required equipment decreases (Collins et
al. 1991), a time-lapse videographic approach may be valuable for estimating the
number of emigrating smolts. Advantages to the development of such a system
include: collection of a permanent record, minimizing stress to the smolts, and making

more efﬁcient use of labor than in trapping efforts.

Temperature Modeling

Temperature plays an integral part in the function of a stream biota and can
have many interacting inﬂuences on organisms throughout a stream. As water
temperature increases, the metabolic rates of instream organisms also increase
(Adams and Breck 1990). Properties of water such as viscosity, sediment load, and
concentrations of both nutrients and dissolved oxygen are inﬂuenced by temperature.

Factors such as groundwater inﬂow, geomorphology, and stream shading
inﬂuence the thermal regime found in streams and these inﬂuences can be predictably

modeled. Using a stream-reach temperature model, Pajak (1992) discovered that

 

stream side shading had the most inﬂuence on maximum temperature in Wisconsin
streams and estimated that 75% shading would maintain summer temperatures in a
stream suitable for brook trout. Temperature models can also be used in diagnosing
problem areas in a watershed. In the John Day River basin in Oregon, Li et al. (1994)
used a model to illustrate that stream temperatures were 5-11°C warmer due to
increased levels of insolation as a result of a loss of vegetation from cattle grazing.

Stream hydrology as well as land uses, shading, bank stability and severity of
the winter are all contributing factors to the overall physical suitability of a stream for
steelhead production. Furthermore, accumulative factors throughout the watershed
such as tributary inﬂuences may provide better indications for a watershed’s capacity to
produce steelhead.

Stream temperature models are of two types, empirical models and physical
process models. Empirical models are statistically based and result from tests for
relations between measured variables (such as air temperature) and instream water
temperatures, usually through simple or multiple regression (Barthalow 1989).
Empirical models are useful for inferring water temperatures when they are not
measured but tend to be limited in predicting responses to the system in the event of a
physical change, such as in stream width or changes in shading. Furthermore, in some
cases, when such physical changes occur, new data often must be collected for the
development of a new empirical model.

Physical process models incorporate the physical parameters that act to .
produce the instream water temperatures. These models usually incorporate measures

to predict the heat ﬂux within the modeled reach (T heurer et al. 1984). Conceptually,

the heat ﬂux model can be presented as:

NetHeatFqu=SR+AR+VTS+EV+CV+CD+F~WBR

where,
SR = solar radiation,
AR = atmospheric radiation,
VTS = vegetative and topographic shading,
EV = evaporation,
CV = convection,
CD = conduction,
F = friction, and
WBR = water back radiation.

Physical process models are powerful in that they can be used to predict likely changes
that would occur with changes in the river morphology or thermal dynamics. Models
such as this are often used to make management decisions in regulated systems with
hydropower dams. These models can be used to alter dam operations regarding ﬂow
and temperature to beneﬁt the ﬁshery. In unregulated systems, these models can be
used to evaluate management alternatives for the thermal environment such as the
effects of increased shading or changes in channel morphology. Once developed for a
stream or watershed, a physical process model can be used in a pro-active approach

for predicting changes in water temperature as a result of changes in land or water use.

Betsie River Watershed

The Betsie River is located in northwestern Michigan (86°N, 44°W) (Figure 2)
and ﬂows through Grand Traverse, Benzie, and Manistee Counties draining a
watershed of 67,126 hectares. The river originates as an unregulated surface outﬂow
from Green Lake near lnterlochen and ﬂows approximately 79 km from its source to

where it empties into Betsie Lake and ultimately into Lake Michigan at Frankfort.

10

 

 

 

 

 

Betsie

Lake
Green Lake
‘\ Dair Grass Lake
Homestead Creek Dam
Dam ‘
Little Betsie
River

Thompsonville
Dam

 

 

 

Figure 2. Location and map of the Betsie River watershed in northwestern
Michigan.

11

Land use in the Betsie River Watershed is hilly, with several glacial moraines
present and the watershed is ranked as one of the highest in the state in percent
composition of glacial outwash, a highly permeable material. The surﬁcial geology in
the Betsie River watershed consists of glacial outvvash (52%) comprised of thick
deposits ( 73.8 - 220.6 m) of stratiﬁed sand and gravel (Hendrickson and Doonan 1972,
Gooding 1995). Watershed soils are comprised of 55% sandy soil type, which is less
than the nearby Boardman and Manistee watersheds which contain more of the sandy
soil type (71% and 86% respectively) (Gooding 1995). Soils in the Betsie River
watershed are comprised of 20% loamy soil type, a mixture of sand, silt and clay which
gives it a larger capacity to hold water than the sandy soil type Gooding 1995).

Land use in the watershed consists of orchards, tree farms, abandoned farms,
forests and several small towns and villages. Between 1966 and 1994, total forested
land increased with the watershed presently containing 57% forest. In 1994, 52% of
the forested area was in deciduous forest and 5% of the forested land was in
coniferous forest (Gooding 1995, Letherberry 1994, Jakes 1982, Pfeifer and Spencer
1966). Approximately 32% of the land is under agricultural use and only 0.5% is
considered as urban (Gooding 1995). The watershed receives 78.7 cm of mean
annual precipitation, and a mean annual temperature of 7.2°C, mean July temperature
of 205°C (Gooding 1995).

The Betsie River is classiﬁed as a marginal trout stream by the Fisheries
Division of the Michigan Department of Natural Resources because of limiting high
temperatures that exist during the summer (Wicklund and Dean 1958). Historically,
there were 3 dams in the Betsie River that impounded the river ﬂow and altered water
quality by increasing water temperature. Two of these dams also precluded

potamodromous ﬁsh runs (such as steelhead), prevented access to the largest

12

coldwater tributary in the watershed and probably warmed the water of the river. Two
of the three dams no longer exist or are modiﬁed to allow ﬁsh passage and the river,
along with the largest coldwater tributary, is available to potamodromous ﬁsh.

The ﬁrst 7 km of river below Green Lake are impounded by Grass Lake Dam
which is 1.2 m high and ﬂoods nearly 465 hectares of land (Carbine 1945). This dam
was established in 1945 based on the location of a logging dam present in the eariy
1800's. Grass Lake Dam was designed to maintain a head of 1.2-1.8 m and to ﬂood
the Grass Lake area and several small spring-fed tributaries. The motive for
impounding this area was to increase duck hunting and northern pike (Esox Iucius)
ﬁshing opportunities in Grass Lake (Wicklund and Dean 1958).

Fourteen kilometers downstream and 0.5 km above the site of Thompsonville
Dam, the Little Betsie River joins the main channel of the Betsie River. Previous
reports (Viﬁcklund and Dean 1958, Bullen 1972) identiﬁed Thompsonville Dam as the
site that contributed most to the warming of the mainstem of the Betsie River and
proposed its removal. The dam maintained a head of 3.1 m and the impounded area
covered approximately 5.7 hectares. As early as 1958, Thompsonville Dam pond was
full of sand and silt that had eroded from unstable banks in the upstream channel.
Temperatures recorded at the outﬂow of the dam reached 27°C. Placement of this
dam also precluded potamodromous ﬁsh from entering the largest cold-water tributary
in the watershed, the Little Betsie River. Thompsonville Dam fell into disrepair and
collapsed in March, 1989, after a tremendous rainstorm. Remains of this dam still
stand along the stream margins downstream of the dam. Raw, eroding streambanks in
the former impoundment area appear to contribute large amounts of sand and silt to

downstream areas as the river cuts a new channel. Failure of the dam resulted in

13

heavy sedimentation in downstream spawning gravel and increased the number of
eroding banks. Presently a sand trap is in place downstream of Thompsonville Dam
but no efforts were made to stabilize the former impoundment area.

Homestead Darn, 35 km downstream of Thompsonville Darn, was converted to
a lamprey weir in 1972. Potamodromous ﬁsh can pass over the dam, but the design
prevents passage of other migratory ﬁsh such as sea lamprey (Petromyzon man'nus),
catostomids, and burbot (Lora Iota). A small, shallow backwater area still remains
upstream as a result of the weir. Temperature surveys do not indicate signiﬁcant
warming in this backwater.

Limited information exists on the Betsie River ﬁshery. However, both the
salmon and steelhead ﬁsheries are regarded highly by anglers. From 1985 - 1988, a
limited (2 months per year) creel study collected information on the retuming steelhead
ﬁshery. Estimates of harvest ranged from 2,600 1 1,401 steelhead in 1986 to a low of

1,129 1 1,119 steelhead in 1988 (Rakoczy and Rogers 1987, 1988, 1990).

Dissertation Overview

The Betsie River provides an opportunistic setting to obtain information on
production of naturalized steelhead smolts from a marginal watershed, effects of
temporal and spatial thermal variation on steelhead parr throughout a watershed, and
to evaluate management alternatives to thermal limitations. The general objectives of
my research were to:
1) Estimate the abundance of naturalized steelhead from the Betsie River watershed

by developing a cost effective, labor efﬁcient, time-lapse video system.

2) Investigate the production of pan throughout the watershed, identify and quantify
production areas, and identify the environmental limitations such as temperature
that inﬂuence population dynamics.

14

3) Determine the current thermal regime throughout the watershed and evaluate
management alternatives, such as a change in source water or removal of Grass
Lake Dam, and their potential effects on altering the thermal regime and making it
more conducive to steelhead trout production.

Each of these objectives is presented as an independent manuscript in chapters 1-3,

and the ﬁnal chapter is a summary, integration, and application of the results regarding

the steelhead population in the Betsie River.

15

Chapter 1

EVALUATION OF ALTERNATE METHODS FOR
ESTIMATING NUMBERS OF OUTMIGRATING STEELHEAD SMOLTS

Abstract
The annual emigration from streams tributary to the Great Lakes by migratory juvenile
salmonids provides an opportunity for ﬁshery managers to estimate the abundance of
the smolting cohort. This information is valuable for modeling recruitment potential to
the ﬁshery and for estimates of the returning broodstock. The purpose of this study
was to compare the results of several techniques for estimating the number of
steelhead Oncorhynchus mykiss smolts emigrating from the Betsie River in
northwestern Michigan. I also compared the production of steelhead smolts from this
river with other rivers in the state. In May and June, 1993-1996, I monitored smolts by
use of visual observations, time-lapse videography and mark-recapture to estimate the
number of steelhead smolts migrating past a lamprey weir located 18 km upstream of
the river mouth. In 1993 and 1994, two observers counted smolts passing over the
weir for 20 minutes out of each hour from 21:00 to 05:00 hours. In 1995 and 1996,
black and white video cameras and time lapse videocassette recorders continuously
monitored the passing smolts through the night. Three reviewers watched the
videotapes and counted the number of smolts passing over the weir from 21 :00 to
05:00 for each night, also collecting counts from hourly 20 minute samples similar to

the visual observations in 1993 and 1994. A subsample of videotapes were reviewed

16

a second time to measure the count variation between the reviewers. In all four years,
a constriction weir was constructed every 5th night during the smolt run to capture
emigrating ﬁsh and to quantify species composition and the origin of steelhead
(hatchery or wild). Steelhead smolts comprised 30-61% of all the emigrating salmonid
juveniles sampled. Other species present were brown trout Salmo tnrtta, coho salmon
Oncorhynchus kisutch, and chinook salmon Oncomynchus tshawytscha. Wild
steelhead comprised 12-52% of the steelhead smolts. Mark-recapture estimates and
95% conﬁdence intervals for steelhead smolts (hatchery and wild combined) ranged
from 13,837 (12,583-15,215) in 1985 to 56,661 (46,036-69,703) in 1993 and were 2-9
times greater than the estimates from visual observation and videography. However,
the results from time-lapse videography and visual observation were similar and gave
the most reliable estimates for steelhead smolts produced from the watershed.
Estimates of steelhead smolt numbers from observation methods ranged from 2,198 (1
512) in 1996 to 9,645 (1 1,111) in 1994. The Betsie River produced fewer wild

steelhead smolts (12-22/ha) than other streams studied in Michigan .

Introduction

Steelhead were introduced to Michigan streams in 1876 and since then, have
established naturalized p0pulations (Latta 1974, Biette et al. 1981). Great Lakes
steelhead retain the life history characteristics of their Paciﬁc counterparts and spend
2-3 years growing to sexual maturation in one of the Great Lakes. In late autumn and
early spring, adult steelhead return to their natal stream and spawn in the spring.
Juveniles reside in the stream for 1-3 years before smolting and emigrating back to the

lake (Biette et al. 1981, Seelbach 1993).

17

The smolting period is a sensitive and critical phase in steelhead life history.
Wild smolts reach a size of 160-200 mm before the necessary changes for
smoltiﬁcation, including the ability to hypoosmoregulate and migratory behavior, occur
(Wagner 1974, Seelbach 1987). Although Great Lakes steelhead do not encounter
saltwater, they follow the same smoltiﬁcation process as steelhead in Paciﬁc streams:
juvenile steelhead smolts lose their parr marks, become silvery, and emigrate in water
temperatures of 10-15°C during the months of April-June (Biette et al. 1981).

Estimation of steelhead smolt numbers is an important component in the
management of steelhead ﬁsheries. The abundance of the smolting cohort, coupled
with expected survival rates, can aid in forecasting recruitment to the ﬁshery and give
indices of expected returns for escapement and stock sustainment (Raymond 1988).
Furthermore, because steelhead parr are resident in the stream for more than one
year and require good water quality (high levels of dissolved oxygen, suitable water
temperatures for survival and growth, low sedimentation rates, an adequate forage
base, and instream habitat for refuge and cover), the number of smolts produced
annually may provide a relative index of the overall health of a watershed for cold water
spades.

Many methods have been used to monitor emigrating smolts, and each method
has produced various levels of success in providing accurate measures of emigrants.
Smolts are difﬁcult to monitor because they migrate at night. In addition, spring high
ﬂows from snowmelt and rain can be problematic for some methods, especially in larger
rivers. Methods to monitor and estimate smolt abundance include: 1)mark-recapture
for trapping efﬁciency (Seelbach 1987, Dempson and Stansbury 1991) or for total run
estimates (Macdonald and Smith 1980), 2) trapping a portion or all of the smolt run

with fyke nets and box traps (Davis et al. 1980, Seelbach and Miller 1993) or inclined

18

screen smolt traps (Wagner et al. 1963, Lister et al. 1969, Seelbach et al. 1984, DuBois
et al. 1991, Seelbach and Miller 1993), 4) electronic ﬁsh counters (Appleby and Tipping
1991), and 5)camera monitoring (Cousens et al. 1982). When a dam is present,
collection methods at turbine intake gatewells or bypass collection methods are often
used (Raymond 1979, Giorgi and Sims 1987, Peven and Hays 1989, Peven et al.
1994).

Problems inherent with various sampling methods for smolts include gear
function at varying ﬂows, debris loading of nets and screens, trap avoidance by smolts,
and mortality associated with handling stress and gear design. Modiﬁcations to the
inclined-screen smolt trap make this a promising method for a variety of applications.
However, trapping efﬁciency can be highly variable or very low, introducing large
amounts of error to the estimates and a low head dam must be in place to use inclined
screen traps (Dubois et al. 1991, Seelbach 1993). Fyke nets quickly become clogged
with debris (Davis et al. 1980). Electronic counters often require that ﬁsh pass through
a very constricted area, and are most accurate when ﬁsh are of a uniform length
(Appleby and Tipping 1991). Although uniform length and monitoring in a controlled
and conﬁned area are applicable to hatchery settings, both of these conditions are not
usually attainable for monitoring smolts in a ﬁeld setting.

Time-lapse videography has been used successfully to record escapement of
adult salmonids (Dexter and Ledet 1994, Hatch et al. 1994). The advantages of this
non-intrusive methodology include establishment of a permanent record, low labor
requirements, ability to review and automate counts, and fairly low cost of equipment as
videographic technology advances (Collins et al. 1991). In the Dexter and Ledet (1994)
study, ﬁsh were viewed through windows in a ﬁsh ladder which allowed for identiﬁcation

of species, but also involved an additional construction cost. Irvine et al. (1991)

19

 

developed a computerized video-camera system to monitor coho smolts by use of
cameras trained on tunnels through which smolts passed. Although this early prototype
of the method was promising and provided the additional beneﬁt of smolt length
measurements, it was a highly controlled situation and not generally applicable to
typical ﬁeld settings.

Steelhead have been naturalized in the Great Lakes Region for over 125 years,
yet few estimates of emigrating smolts exist. Yearly production of smolts is known to
vary widely in high quality trout streams in Michigan (Seelbach 1987, Seelbach and
Miller 1993), but little is known about their production from watersheds with marginal
trout habitat. In southern Michigan, for example, rivers tributary to Lake Michigan which
are stocked with steelhead receive large retums of adults, a fraction of which are wild.
Although some natural reproduction is known to occur, it has yet to be quantiﬁed
(Seelbach 1987). The tendency of rivers in this region to warm signiﬁcantly during the
summer makes them marginal or unsuitable for salmonid production (Seelbach et al
1994).

The objectives of this study were to:

1) develop a time-lapse, surface-view videographic system for monitoring
emigrating smolts

2) estimate the number of steelhead smolts emigrating from the Betsie River
by use of direct observation, time-lapse videography, and mark-recapture
and compare results among the three methods, and

3) compare the number of emigrating smolts in the Betsie River watershed
with other rivers in Michigan.

Study Site
The Betsie River watershed is tributary to northern Lake Michigan. The main

channel of the Betsie River is 79 kilometers long and the watershed drains an area of

20

——____ ,__ _

67,100 hectares (Figure 3). Land use in the watershed consists primarily of tree
taming, some fmit farms and a portion is owned by the State of Michigan and
managed as a State Forest. Approximately 15-35% of the watershed is in agricultural
use, 4-14% is covered by coniferous forest, and 45-60% of the watershed is deciduous
forest (Gooding 1995). The river also ﬂows through two small municipalities, with little
development or industry along the river. Sandy soil types dominate a large portion of
the watershed (35-66%) (Gooding 1995) and glacial moraines are dominant landscape
features throughout the watershed. Several permanent ﬂowing springs and seeps
supply groundwater directly into the river, especially in the mid-reaches of the
watershed. The two largest tributaries that empty into the Betsie River are Dair Creek
and Little Betsie River, and both of these are spring-fed.

The Betsie River has no permanent US. Geological Survey gaging stations. To
determine discharge, I measured river stage weekly by use of temporary staff gages
from 1993-1996. Staff gages were installed and rated according to standard
hydrological methods (Gordon et al. 1992, Appendix A). The mean May-June
measured discharge ranged from 4.3 to 11.3 cubic meters per second (m3/s) during this
period. Stream gradient is generally low, and averages 0.002% (range 0.001 -
0.005%). Historical water temperature data showed that the Betsie River warms
signiﬁcantly in the summer months (Wicklund and Dean 1958) and for this reason, the
river is considered a marginal trout stream. The ﬁsh community of the Betsie River.
includes rockbass Ambloplites mpestn's, creek chub Semotilus atromaculatus, common
shiner Notropis comutus, homyhead chub Nocomis biguttafus, blacknose dace
Rhinrbhthys atratulus, mottled sculpin Cottus bairdi, brown trout Salmo trutta, and

juvenile steelhead. Dair Creek and Little Betsie River ﬁsh communities consist of cold

21

     
        
   
 

Green Lake

Grass Lake

Homestead Da

Darn

Little Betsie
River

Figure 3. Location and map of the Betsie River watershed in
northwestern Michigan.

22

water species, including brown trout, slimy sculpin Coffus cognatus, mottled sculpin,
steelhead, and coho salmon parr. Although the mainstem Betsie River is classiﬁed as
marginal for trout, it supports a popular ﬁshery for returning salmon and steelhead
(Rakoczy and Rogers 1987, 1988, 1990).

Since 1991, angling interest organizations in cooperation with the Michigan
Department of Natural Resources, Fisheries Division, have raised juvenile steelhead in
a hatchery adjacent to the Betsie River and linked to the river by a small ﬂowing spring.
Yearling steelhead were released volitionally, beginning in April. Most of the smolts
from this hatchery were clipped with either a right ventral (1993-1995) or right pectoral
clip (1996). The number of steelhead released from this facility has ranged from
20,000 (1991) - 55,000 (1996).

l monitored emigrating smolts at a lamprey weir located in the lower river reach,
17.7 km upstream from Lake Michigan. Formerly the site of Homestead Darn, a low-
head hydroelectric dam, the site now consists of a lamprey weir which is divided into
two sills (10 m wide each) that are staggered, with one sill approximately 5 in further
downstream than the other (Figure 4). The weir has a head of approximately 2 m
which is stepped into 1 m jumps on the most upstream sill and there are no steps on

the downstream sill.

Methods

Beginning the ﬁrst week in May and ending in late June, 1993-1996, I monitored
outrnigrating smolts by use of three methods: visual observation, time-lapse
videography, and mark-recapture. In 1993 and 1994, two observers counted smolts

passing over the downstream sill of the weir for 20 minutes out of every hour from

23

Water Flow

 

 

 

 

 

 

 

 

 

 

Light
Recorders
and
Equipment
—t \ \ ‘\ \ \
\ \

Camera 3

 

Camera A

Figure 4. Diagram of equipment used at the Homestead Lamprey
Weir to monitor emigrating smolts in the Betsie River, Michigan.

24

21:00 to 05:00 hours, approximately dusk to dawn . To check for dayﬁme emigration, I
randomly sampled during daytime hours approximately every other day during the
smolt run. Visual observations in 1993 and 1994 indicated that the majority (80%) of
the smolt movement occurred over the most downstream sill. Therefore, the visual and
video observations were concentrated on this sill and estimates were adjusted
accordingly. During night sampling, each viewer used a hand-held light to illuminate the
area and monitored half of the 10 m wide sill to count the smolts, which were readily
visible passing over the weir. In 1995 and 1996, visual observations were corroborated
with camera observations and similar 20 minute observations were obtained from the
videotapes. For each night of observation, a mean 20-minute sample was calculated
and multiplied by 24 to calculate a nightly estimate for 8 hours of smolt movement.
Because a few nights were not sampled due to equipment malfunction, weather

interference, and ﬂooding, a weekly estimate for the smolt run was determined from:

Ct = (N/n) ° ( 2 C. + EF) (Equation 1)
and,
Var (0.) = N(N-n)1/(N-1)° 2: (CI' CM) (Equation 2)
where,
C. = total number of smolts for the week,
N = total number of nights of smolt movement,
n = number of nights sampled by electroﬁshing or observation,
C. = the total estimate for a single night,
EF = the number of smolts captured by electroﬁshing, and
C... = the average of all night estimates for the week.

The estimate of total smolts was adjusted by the proportion of steelhead captured by

trapping in that week to obtain an estimate of steelhead smolts. All C. and Var(C.)

25

estimates were summed for an annual estimate of smolt numbers and variance of
annual smolt numbers, respectively.

In addition to estimating smolt numbers with 20 minute observations, l evaluated
estimates and the error rates associated with different frequencies of this type of
sampling. I used the original estimate, derived from all days of sampling, as the
estimate for 100% of observations through the duration of the smolt migration.
Estimates and standard errors were then derived for each of the four years by using the
values obtained from periodic sampling schedules including: every-other day (50%),
every 3rd day (30%), and by obtaining a random sample of 80% and 20% of the nights
during the smolt run in 1993-1996. Estimates and variances were calculated according
to Equations 1 and 2.

In 1995 and 1996, I used surface-view, time-lapse videography to continuously
monitor and record emigrating smolts. Two videocassette recorders (Sony SVT-100)
stationed in housing next to the river and connected to two solid state black and white
cameras (Vicon 12.5 mm F 1.3 (A) and 25 mm F 1.3 (B), housed in Environmental
Enclosures (Pelco EH4500)) each monitored and recorded smolts passing over half of
the sill (Figure 4). Five frames per second were recorded, resulting in 12 hours of
actual smolt movement condensed to 1 hour of videotape. The sill of the dam was
continuously illuminated with a 250 watt halogen light. Due to the remote location of
the monitoring site, gasoline powered generators were used to provide electricity.
Twice each night, recording was interrupted for less than 2 minutes to refuel the
generators. Cameras were set up each night before dark and removed in the morning
to prevent vandalism and theft. A small video monitor located in the housing next to

the stream assisted with camera placement and focus. A ﬁshing bobber, anchored at

26

the middle of the sill aided in positioning each camera’s view over half the sill and
avoided overIap and duplicate recording of smolts.

Video recordings of smolts passing over the wild were watched by 3 reviewers
using a professional editing videocassette recorder (Panasonic AG1980P). This
equipment provided for frame by frame analysis and reverse playback of the
videotapes. All reviewers worked together initially to develop a consensus on smolt
images and counting methods. Fish were distinguished from ﬂotsam on the videotapes
by their fusiform shape and thickness, and subsurface presentation. In addition, smolt
behavior while passing over the weir typically consisted of smolts backing downstream
tail ﬁrst and approaching the sill. Smolts also exhibited a slight bonding to form a C-
shape, movement that was slower and often lateral to the water current. Smolts
usually were present in two or more frames. All tapes were reviewed from the
beginning of the evening, 21:00 hours, to the end of the nightly smolt movement at
05:00 hours.

Camera A and camera B counts were summed to obtain an estimate of smolts
for the night. Similar to the 20 minute estimates, a weekly estimate of smolts from the
videotape counts was calculated by the equation:

C1(AB) = NW lnm ° 2 CI(A) + Na, lnm * 2‘, Care) + EF (Equation 3)

where A and B designate the cameras of observation. Each weekly estimate was then
adjusted by the proportion of steelhead in the captured smolts and weeks were
summed to obtain an annual estimate for 1995 and 1996.

In consideration of the viewer subjective interpretation that was required to
review videotapes and enumerate smolts, I conducted an error analysis from a

stratiﬁed, random sampling of the videotapes. Because rates of smolt movement

27

varied hourly, daily, and weekly through May and June, I randomly selected one night
from each week during the smolt run for recounting. The selected night was divided into
three equal time blocks, and I randomly selected one 30 minute period per time block to
recount by a second reviewer. A paired t-test was used to test for differences between
2 viewers and variances of the daily estimates based on viewer interpretation were
calculated according to the following equation:

Var (0.) = (k 0.)2 (Equation 4)
where,

C. = total number of smolts observed for night i, and
K = the coefﬁcient of variation (CV) among viewers.

The CV was calculated by:
cv. = (sqrto; «XI - xtmj)’ )/ n. - 1)/ x.,,,,..,,,) (Equation 5)
where,
X; = the sum of the 2 reviewers counts for a night,

Xx“, = the nightly average from the two reviews, and
n. = number of reviews of night (i) = 2.

A mean CV was calculated for each camera from the nightly variance among viewers.
The CV was assumed to be constant over time and a variance of the estimate of the

numbers of smolts based on viewer interpretation was calculated by:
Var (C,) = N(N-n)(1/N-1) * 2 (C. - QM)? + N/n * Z Var (Cj) (Equation 6)

Separate estimates of smolt numbers and variances were determined for each camera
and summed to determine a yearly smolt estimate and variance.

To determine species composition of the outmigrants and also to identify the
origin of the steelhead smolts, l constructed a temporary constriction weir upstream of

the lamprey weir approximately every 5th night during the outmigration period. The

28

weir consisted of construction netting angled from both streambanks to a 3 m
downstream opening (Figure 5). A large block not was placed in the opening to collect
the smolts. Beginning at 21 :00 hours, I used a 250 volt DC barge electroﬁshing unit
within the block not every two hours until 05:00 hours to capture smolts. All ﬁsh were
identiﬁed to species, measured for total length, weighed, and released below
Homestead Dam. In addition, I checked all steelhead for either a right pelvic (1993-
1996) or right pectoral (1996) ﬁn clip and removed scale samples from a subsample of
hatchery (ﬁn-clipped) steelhead and all steelhead without a ﬁn clip. I analyzed scale
samples by inspecting the circuli surrounding the ﬁrst annulus (Seelbach and Whelan
1988) to determine if the smolt was hatchery or wild.

I calculated a Peterson mark-recapture estimate of the steelhead smolts leaving
the Betsie River watershed by assuming that the number of stocked hatchery ﬁsh each
year comprised the marked population, and that clipped ﬁsh were equally likely to be
recaptured in the constriction weir as the unmarked wild portion of the population
(Ricker 1975).

Summer electroﬁshing surveys (Chapter 2) revealed that a large number of
hatchery steelhead did not leave the watershed (Figure 6). I estimated the number of
stocked steelhead that remained in the river (residualized ﬁsh) in July to determine the
actual number of marked migratory steelhead that migrated downstream for the smolt
mark-recapture estimates. l stratiﬁed the watershed into reaches based on hydrology,
gradient, and thermal regime and randomly sampled equally in the reaches for a total
of 21 sites. I used electroﬁshing and multiple-pass depletion methods to determine the
number of residual steelhead (Zippen 1958). An estimate of residual steelhead
abundance was calculated from the mean density of hatchery ﬁsh for the sample reach

and the channel area in the reach. Because residual hatchery ﬁsh are known to remain

29

 

Construction
netting

Block net

 

Homestead
Lamprey Weir

 

Figure 5. Constriction weir and block net design for capture of emigrating
in the Betsie River, Michigan.

30

Number of steelhead smolts

60000

50000

40000

30000

20000

1 0000

 

I Stocked SH

 

E] Residual SH I Migratory SH

 

 

 

/ .
,, .;
’)
/ f
I .
,
, I
,.
1' / /
z ’1 / ‘3
z I ,1
z, , ‘ '
, . .
‘ .3 , '1'"
z x , , ,-‘
/ //’,‘ . ,/
1 / ‘. , "
1 ' / .,
1,1. //' ..
a/xfsV/r’f 4 , /
, . /. /,
I'd/"z”. ,{z'x’if
:1” 77* /,
. 3,7 ,
/,..'-,. / ‘I
‘/ , ,. , /
,, , ,,., /' ,,-,/
. / /' .7
‘ , . ' i/
/, / . ,/
. 7,;
[If ’r
'I ‘7
//,,'.r.3""1
./:~’ I
/

 

Year

Figure 6. Number of steelhead smolts stocked, estimated number that
remained as residuals in the river, and resulting number of potential
migratory steelhead smolts in the Betsie River, Michigan, 1993-1996.

31

within the area where they were stocked (Seelbach 1987), I estimated the density of
residual ﬁsh from a site sampled near the hatchery to determine the abundance of
hatchery ﬁsh within 2 miles upstream and downstream of the stocking location. All
reach estimates were then summed to obtain a total watershed estimate for residual
hatchery steelhead. The estimate of residual steelhead was subtracted from the
number of ﬁsh stocked from the hatchery to determine the number of marked,
migratory ﬁsh that was assumed to be randomly mixed with the wild migrating smolts
(Figure 6). I used standard methods to calculate the Peterson estimate and variance

(Ricker 1975).

Results
Observations of smolt movement commenced at the beginning of May each
year and continued until mid- to late June, with the exception of 1993 when sampling
was discontinued early due to logistical complications. Steelhead smolts ceased
migratory activity in the last two weeks of June and the duration of the smolt run ranged

from 39 to 51 days (Table 1). From 1993-1996, I captured 1,755 steelhead smolts of

Table 1. Duration of observations and smolt movement at Homestead Dam, Betsie
River, Michigan, 1993-1996.

 

 

Date of First First Smolt Last Smolt Date of Last Duration of

Observation Observed Observed Observation Smolt Movement
5-8-93 5-10-93 6-16-93 6-17-93 39 days
5-5-94 5- 7-94 6-22-94 6-23-94 47 days
5-1-95 5- 2-95 6-19-95 6-22-95 48 days
5-1-96 5- 7-96 6-27-96 6-28-96 51 days

 

32

which 365 were determined to be wild on the basis of scale analysis. Annual mean
length of wild steelhead smolts in 1993-1996 ranged from 193.6 to 195.0 mm. Over
half of the wild emigrating smolts were age-1 (SS-72%) each year, less than half were
age-2 ( 26-45%), and fewer than 3% of the emigrating steelhead smolts were age-3.

Smolts migrated past the monitoring site in a consistent nightly pattern. The
greatest number of smolts passed over the Homestead Weir between dusk and
midnight. Over 80% of the smolt passage occurred before 02:00 hours (Figure 7) on
most nights.

Four species of ﬁsh were captured in the constriction weir above Homestead
Dam during the hours of 18:00 through 05:00. Recently released hatchery brown trout
(ﬁn erosion indicated hatchery origin), wild coho salmon smolts, wild and hatchery
steelhead smolts, and wild chinook salmon smolts were present in the catch. The
proportions of each species captured varied between years; steelhead comprised 30%
to 61% of the spring smolt run (Figure 8).

Each year the number of marked ﬁsh stocked into the Betsie River increased.
The number of residual ﬁsh also was very large in some years, which accounted for
between-year variance in the number of marked migratory steelhead (Figure 6).
Peterson mark-recapture estimates that were not adjusted for residual ﬁsh resulted in
estimates 2-5 times greater than those adjusted for residual ﬁsh (Figure 9). Total
estimates for hatchery and wild steelhead passing the weir ranged from 13,837
(12583-15215) in 1995 to 56,661 (46,036-69,703) in 1993 (Table 2). Wild steelhead
smolts comprised 30%, 12%, 24%, and 52% of the captured steelhead smolts in 1993-

1996, respectively.

33

 

 

 

45 .. “F 90

_ 40 -- -- 30

5 35 -- -- 70
5‘

.1: 3° "' -- 60
.9

g 25 -- -- 50

*5 20 -- -- 40
0

§ 15 " -- 30
o.

10 -- 'r 20

5 -- .......... -- 10

o ‘ ..... p o

22:0023:0012:001:00 2:00 3:00 4:00 5:00 6:00

Time of night

Figure 7. Temporal distribution of outrnigrating steelhead smolts in
the Betsie River, Michigan. Hourly values represent the mean for
observations during that time period.

34

Cumulative percent of nightly total

Brown Trout
3°/o

    

33%

%

Figure 8. Species composition of the juvenile salmonids captured at
Homestead Weir, Betsie River, Michigan, 1993-1996.

35

1 O 'I D Peterson estimate

 

| Peterson estimate '
1400001 adjusted for residual
hatchery smolts
120000--
100000 .-
80000 1'-

_[_

Number of smolts

 

 

  

 

 

 

 

 

 

   

1993 1994 1995 1996
Year

   

Figure 9. Comparison of mark-recapture estimates and 95% conﬁdence
intervals of steelhead smolt numbers with and without adjustment for
residual steelhead in the Betsie River, Michigan, 1993-1996.

36

Table 2. Peterson mark-recapture estimates of the number of steelhead in the Betsie

 

 

River smolt run.
# Marked Total # of # of Hatchery Estimate and Estimate and
Migratory Steelhead Smolts 95% C. l. of 95% C. I. of
Year Steelhead Captured Recaptured All Steelhead Wild Steelhead
1993 39,888 124 87 56,661 16,998
46,036-69,703 15,057-18,939
1994 23,335 977 860 26,507 3,181
27,795 - 28,337 3,107 - 3,255
1995 10,463 561 424 13,837 3,321
12,583 - 15,215 3,164 - 3,478
1996 9,832 13 54 20,381 10,598
15,687 - 26,448 8,581- 12,615

 

The mean number of steelhead outmigrants was even lower based on the 20

minute visual sampling and videographic methods. The number of wild and hatchery

steelhead passing the weir ranged from 2,328 (1 1,249) in 1996 to 9,645 (1 1,111) in

1994. The number of wild steelhead smolts was determined from the overall estimate

of steelhead smolts and varied only slightly among years from 1,211 (1 649) in 1996 to

2,151 (1 286) in 1993 (Table 3). I used videographic methods in the last two years of

the study, and smolt numbers were similar to the estimates derived from direct

observation, 20 minute sampling (T able 3).

37.

Table 3. Estimate of the number of smolts from 20 minute direct observation sampling
and continuous time-lapse videography at the Homestead Lamprey Weir, Betsie River,
Michigan.

 

 

Estimate and 95% Conﬁdence Estimate and 95% Conﬁdence
Year Limit for all Steelhead Smolts Limit for Wild Steelhead Smolts
20 Minute Observations
1993 7,169 1 954 2,151 1 286
1994 9,64511,111 1,1571133
1995 7,120 1 1,282 1,709 1 308
1996 2,198 1 512 1,143 1 266
Time Lapse Videography
1995 5,259 1 3,328 1,262 1 799
1996 2,328 1 1,249 1,211 1 649

 

With a few exceptions, estimates of smolt numbers from subsamples of the
nights of observation were similar to those from sampling the entire smolt run (Figure
10). Large differences occurred however, in the 95% conﬁdence intervals and
standard error (Figures 10 and 11). Standard error increased with decreasing levels of
effort Effort involving every other day (50%) and an 80% random sample had similar
error rates, and sampling efforts of every 3rd day (30%) and a random 20% were much
larger in error (Figure 11).

Discussion

Steelhead smolts in the Betsie River followed similar migratory patterns to those
described from other Michigan streams and elsewhere in their timing, and nocturnal
movement (Stauffer 1968, Seelbach 1993). Because I observed no signiﬁcant smolt
movement during the hours between 05:00 and 21 :00, our efforts were concentrated
on nocturnal movement. Brege et al. (1996) monitored salmonid smolt movement on

the Columbia River over 6 years and nocturnal movement of steelhead smolts

38

Number of outmigrating smolts

20000 .. 1993

15000 ‘-

100001

 

 

a

1994 § 1995 1996

I"

ﬂ

 

1

 

étlll“

 

10080 50 3020 1008050 302010080 50302010080 50 3020

Percent Effort

Figure 10. Comparison of the number of outmigrating smolts and 95%
conﬁdence intervals calculated from variable sampling efforts ranging from
100% to 20% of the days during the smolt run.

39

Standard Error

4000 --

3500 1
3000 1*

2500 -

2000 -

1500 .

1000 a

500‘

 

 

 

0 i i i i i
0 20 40 60 80 100
Percent Sampling Effort

Figure 11. A comparison of the standard error of number of outmigrating
steelhead smolts as affected by sampling effort. Fitted line is a logarithmic
relationship with r2 = 0.7697.

40

accounted for 66.2% to 87.4% of the daily observations, with a mean of 77.9% (Brege
et al. 1996). Using this information, our estimates could be increased by 22% to
account for any daytime passage that may have occurred. However, from periodic
checks of smolt activity during the day, only once, in 1995 was any movement
observed between the hours of 07:00 and 21:00.

The mark-recapture estimates of outmigrating smolt numbers were 2-9 times
greater than those determined from the videography and direct observation methods
(Figure 12). Assumptions of mark-recapture methodology include: 1) animals do not
lose marks, 2) sampled ﬁsh are classiﬁed correctly, 3) either the population is closed, or
there is neither recruitment or immigration, and 4) sampling is random with regard to
marks, so that every ﬁsh has an equal chance of capture (Cormack 1968, Pollock
1981). Because ﬁn clips were used to identify hatchery ﬁsh, marks were not lost.
However, ﬁsh marking was not consistent (up to 35% of the smolts were not clipped in
one year). To avoid an error in the assumption of correct classiﬁcation, I used scale
analysis to verify classiﬁcation of smolts as either hatchery or wild. I also assumed that
the migratory population of smolts was closed. This was a reasonable assumption as
this was a watershed assessment and smolts were not able to move upstream past the
Homestead lamprey weir.

Unexpectedly large numbers of marked ﬁsh did residualize and I adjusted the
estimate of the marked migratory population accordingly. It is possible that l
underestimated the number of residual smolts: e.g., they may have concentrated in
large pools below the hatchery which were not amenable to sampling with our
electroﬁshing procedures. In a different approach to estimating residual steelhead, I
calculated the number of hatchery steelhead that were likely to residualize based on

lengths observed in the hatchery in April. From measures of 100-200 individual ﬁsh in

41

70,000 '

 

 

 

 

 

 

 

 

 

1p UM " R
I20 Min
60,000 .. ElVideo
I- ..
£ 50,000 i-
o
E ..
I
'2 40,000 ..
at
#1 30,000 "
3 F
5 20,000 " ‘I'
:i
2
10,000 ‘
0 . _iil_,
1993 1994 1995 1996
Year

Figure 12. Comparison of the steelhead smolt numbers determined
from mark-recapture (MR), 20 minute observations (20 Min) and
continuous recording (Video) in the Betsie River, Michigan, 1993-
1996. Error bars represent 95% conﬁdence intervals.

42

hatchery in 1994-1996, I determined that 97%, 81%, and 88% (respectively) of the
steelhead in the hatchery were less than 180 mm, the length at or above which
steelhead smoltiﬁcation is likely to occur (Seelbach 1987). Using this method, the
estimates of residual steelhead were 47,142, 40,500, and 48,400 in 1994-1996. In this
approach, the estimates for 1995 and 1996 are similar to the estimates of residual ﬁsh
determined from electroﬁshing throughout the watershed in July (Figure 6). However,
the 1994 estimate from the hatchery approach was neariy twice as large as that
determined from the watershed estimates. This may indicate that l underestimated the
number of residual ﬁsh for that year with electroﬁshing, or that other factors, such as
release dates or hatchery crowding may have inﬂuenced the emigration of these ﬁsh
(Wedemeyer et al. 1980).

A failure to recapture a sufﬁcient number of marked ﬁsh for a precise estimate
would have been indicated by large conﬁdence intervals (Dempson and Stansbury
1991). The conﬁdence intervals determined in this study ranged from 10-24% of the
estimates, which may be considered as fairly reliable estimates (Roff 1973, Cousens et
al. 1982)

The most likely reason for the seemingly disproportionate mark-recapture
estimates were due to violations of the assumption of equal chance of capture (Pollock
1981). Inherent in this assumption is that marked and unmarked ﬁsh mix randomly and
that the probability of capture is equal for either classiﬁcation. Juvenile steelhead were
released volitionally from the hatchery, which was located approximately 29 km
upstream of the Homestead lamprey weir. Volitional releases resulted in the hatchery
ﬁsh being released at different rates between years. In some years, smolts left the

hatchery one to two weeks after a peak in the number of outmigrating wild smolts and

43

as the river was warming above the upper limit (13°C: Zaugg and McClain 1973) for
steelhead smolting behavior.

In a similar mark-recapture application to estimate the number of outmigrating
smolts (Peven and Hays 1989), the estimates of smolt numbers were corroborated with
a life-history method (estimating smolt production based on spawner escapement
values, fecundity rates, and survival to age information) to determine the number of
smolts. Peven and Hays (1989) found similar results between the two methods, with
mark-recapture of smolts giving a likely and reasonable estimate. The difference in this
application from the Betsie River is that Peven and Hays (1989) veriﬁed that marked
smolts were migrating when they were marked, but I could not verify that the marked
ﬁsh were migrating when they entered the Betsie River.

Other estimates, using mark-recapture and maximum-likelihood methods, may
have yielded better estimates than the single census mark-recapture (Cormack 1968,
Dempson and Stansbury 1991, Schwarz and Dempson 1994, Pollock 1981, Macdonald
and Smith 1980). However, I could not obtain estimates of the number of smolts
released on any given day and smolts were released at varying times (e.g. weekdays
only and not on holidays), which precluded the use of any method other than the
single census Lincoln-Peterson index.

Mark-recapture estimates are used often to estimate trap efﬁciency for smolt
collection methods (Raymond 1979, Seelbach 1987). Raymond (1979) concluded that
the marked steelhead smolts were equally mixed with the nonmarked smolts because
of an equal capture rate at a downstream location. When used to estimate escapement
by adult salmonids, the accuracy of mark recapture ranged from 1 25% to 1 30% of the

estimated value (Cousens et al. 1982). Based on my experience in the Betsie River,

44

 

mark-recapture methods overestimated the number of emigrating smolts and I would
caution against the use of single-census Peterson estimates alone to obtain total smolt
abundance estimates with stocked, marked hatchery ﬁsh. At the very least, an
estimate of residual marked ﬁsh should be determined to make adjustments in the
number of marked migrants.

The time-lapse videography used in this study has several advantages that
make it feasible in many different ﬁeld applications. Use of such monitoring systems is
labor efﬁcient in that many stations can be operated by a single individual, allowing a
large amount of information to be recorded during this time-speciﬁc life history phase
for later analysis. The system can be used in remote locations, as demonstrated by our
use of gasoline generators which provided sufﬁcient power and did not cause
interference with the cameras or recorders. A surface view approach provided an
effective and inexpensive viewing of emigrating smolts. Although this application
involved a lamprey weir, I could have easily established a viewing station on a hydraulic
control or rifﬂe; the system should work at any location where ﬁsh could be guided
over a nanow backdrop presented at a depth where smolts could still be viewed.
Drawbacks to the surface view approach are that it requires an additional trapping
effort when multiple species are migrating and that turbidity of the water can impede
observations. However, in our study, turbidity prevented visual observations on only 3
days in four years (185 days) of viewing. A nrled backdrop and slightly higher
resolution of the images (which would require at least one additional camera in this
application) could have allowed us to use inference from lengths to determine species
composition among coho salmon, chinook salmon and steelhead based on the

differences in mean length of smolts of each species.

45

Although the video monitoring was relatively labor efﬁcient, manual review of
the videotapes was tedious, but produced accurate results. A single taping event for
one camera included 8 hours of recording and required 3-4 hours to carefully review
and count smolts. The efﬁciency of this effort could be enhanced by automated viewing
techniques (Irvine et al. 1991). Similar to Hatch et al. (1994), comparisons between
direct observation estimates and videotape results were similar. In addition, no
signiﬁcant differences were found between the 3 trained reviewers in counting smolts
from the videotapes.

Because of the nocturnal movement of smolts, lighting was required to
illuminate the area enough to provide a view through the water to the sill. I used 250
watt diffuse halogen lighting to illuminate the area for monitoring. Because the lighting
was constant, it is not likely to have affected the passage of smolts. The change in
environment for the emigrating smolts from the dark upstream areas to the lighted sill
area may have caused hesitation in passage over the weir. A typical pattern observed
by the smolts was for them to back down to the front of the sill and school together.
The school would swim back and forth across the front of the sill and continually
increase in numbers until a time when the large numbers of smolts would pass over the
sill either together or sequentially. Artiﬁcial lighting was not likely to have caused this
behavior, because the same behavior was observed during dusk when lighting was not
required and also during the few daylight observations of smolt passage. One problem
with the use of halogen lighting was that it attracted clouds of emerging insects. At
times, this caused the cameras to focus on the insects rather than the sill and
sometimes the insects were so thick that they prevented counting of the smolts. This

was usually an early evening phenomenon, and occurred when air temperatures were

46

warm. However, it could be avoided by the use of infra-red lighting, without an effect
on ﬁsh behavior (Beach 1978).

The abundance of smolts passing through any given night was variable. In a
telemetry study of coho smolts, Moser et al. (1991) found that movement was in a
saltatory fashion with movement in the direction of the current and then extended
periods of holding in areas of low current velocity. This behavior could explain some of
the variability in the abundance of smolts observed between nights, especially when
considering the small number of smolts that emigrated from the watershed.

Subsamples of the number of nights required to establish an estimate of smolts
indicated that sampling between 60-100% of the nights during smolt outmigration would
give similar estimates for the effort incurred. An effort of observation that included less
than 60% of the nights of smolt movement would likely result in an estimate with an
extremely large error. Of course, in any smolt monitoring application, determination of
the beginning and end of smolt movement is required to determine the extent of
sampling required. In some cases, the onset and cessation of smolt movement may be
approximated with water temperature trends (Wedemeyer et al. 1980, Hoar 1984).

The Betsie River watershed produced few smolts per hectare when compared
to other streams in Michigan. The Huron River, a tributary of Lake Superior and a
smaller river than the Betsie River, produced 9,141 and 1,031 wild steelhead smolts in
1987 and 1988 or 46-262 steelhead smolts/ha (Seelbach and Miller 1993). The Little
Manistee River is a high quality trout stream tributary to Lake Michigan, south of the
Betsie River, with a drainage of 59,000 ha and a mean annual ﬂow of 5-6 m3/s
(Seelbach et al. 1984). This river produces a greater number of wild smolts annually,

ranging from 11,845 to 86,425 smolts/year with a density of 425/ha (97-713/ha).

47

Conclusions and Management Implications

Currently, the Betsie River watershed produces fewer than 3,000 wild steelhead
smolts, far less than steelhead smolt abundance estimated in a high quality trout
stream such as the nearby Little Manistee River. However, this level of production may
reﬂect the productive capacity that could be expected in the marginal trout streams that
experience high summer temperatures and moderate to severe winter temperatures.

Mark-recapture methods largely overestimated the numbers of smolts leaving
this watershed. A single census, Peterson estimate should be used with great caution
and the assumptions of equal mixing and equal probability of capture should be
addressed for conclusive results.

The surface-view time-lapse videography and 20 minute observational sampling
provided estimates similar to one another. Both are inexpensive (relative to equipment
requirements) and could be conducted on weirs and natural hydraulic controls in
smaller rivers. The videography has additional advantages of providing a permanent
record and requiring less labor. Improvements to the video system I used could include
electricity at the site, infra-red lighting to decrease interference with the view frame by
ﬂying insects, and a contrasting backdrop to help clearly illuminate the ﬁsh. Further
efﬁciency could be accomplished by increasing the resolution of the cameras and

adding a ruled backdrop to measure emigrating smolts.

48

Chapter 2

DISTRIBUTION, DENSITY, AND SURVIVAL OF STEELHEAD
PARR IN A THERMALLY DIVERSE WATERSHED

Abstract

In watersheds with diverse instream habitats, such as the Betsie River in
northern Michigan, production of eariy life history stages of ﬁshes can be concentrated
into a few critical habitat units. The Betsie River watershed is atypical of the
neighboring watersheds due to its thermal properties. The main channel supports a
popular migratory salmon and steelhead ﬁshery, but is marginal habitat for resident
trout because summer temperatures are high. Tributaries in the watershed are
proportionately greater in groundwater ﬂow than the main channel and therefore
provide colder habitats in summer. The objectives of this study were to describe and
quantify the distribution, density, and mortality of steelhead (Oncorhynchus mykiss)
parr throughout the Betsie River and its tributaries and to evaluate the inﬂuence of
different thermal macrohabitats on steelhead parr production. From 1993-1996, I
estimated parr abundance at 14 sites throughout the watershed by use of depletion
sampling and I determined the number of emigrating smolts by visual methods.
Densities of steelhead parr ranged from 0/ha to 30%/ha at sampling sites. I found
62% of age-0 parr and 50% of the age-1 pen in the main channel. The remaining parr

were found in the tributaries which comprise only 11% of the total stream area in the

49

watershed. Smolts out-migrated at annual yields ranging from 11.9/ha to 22.0/ha for
the watershed channel area. Annual instantaneous mortality rates (2) ranged from
0.710 to 3.578 and were the greatest for age-1 and age-2 parr. Mortality was greater
during severe winters and at sites in the main channel. Because many parr were found
in the few tributaries sampled, and because survival was higher in these streams, the
small tributaries are highly valuable to the overall production of steelhead from this

watershed.

Introduction

Steelhead were introduced into the Great Lakes Basin in 1867 and since their
introduction, naturalized populations have become established in many Great Lakes
tributaries (Latte 1974, Biette et al. 1981). Great Lakes steelhead migrate into streams
in the fall, winter, and spring and spawn in the spring. Juveniles typically spend 1-3
years resident in the tributary watershed before migrating to the lake, where they grow
for 1-3 years before reaching sexual maturity (Biette et al. 1981, Seelbach 1993). In
most streams, the majority of juvenile steelhead spend 2 years in the river while the
remainder spend only 1 or 3 years in the watershed (Biette et al. 1981).

Water temperature can limit the distribution and abundance of a population of
salmonids by high summer temperatures or low winter temperatures that exceed the
tolerance limits for survival (0 - 25°C; Viﬁsmer and Christie 1987). Effects of
temperatures approaching or exceeding the limits may include mortality, metabolic
challenges resulting in decreased growth (Adams and Brock 1990), or ﬁsh may

distribute themselves throughout the watershed in the most preferred thermal areas

50

(Nielson et al. 1994, Peterson and Rabeni 1996) which may lead to competition for
space or a canying capacity that is a function of thermal macrohabitat availability.
Many of the southern streams in Michigan are marginal for resident trout due to high
summer temperatures, yet returns of both wild and hatchery steelhead to these rivers
can be very large (Seelbach 1989, Dexter and Ledet1993). Currently, little information
exists on the numbers of juvenile steelhead that can be produced or their population
dynamics in watersheds dominated by marginal thermal habitat (Seelbach 1989, 1993).

Severe winters can be detrimental to a cohort of steelhead with mortality of 50-
90% reported in some studies (Maciolek and Needham 1951, Seelbach 1987). Partial
explanation for the mechanisms behind winter mortality may be limiting microhabitat. ln
water below 8°C, salmonids reduce activity and seek cover (Hartman 1963, Campbell
and Neuner 1983). Preferred winter habitat for steelhead parr includes groundwater
fed side-channels, complex woody debris, interstitial spaces within boulders or rubble
along stream margins, under cobble, and in water less than 30 cm deep with near-0
water velocities (Everest and Sedell 1983, Campbell and Neuner 1984, Heifetz et al.
1986, Swales et al. 1986). In northern streams, low temperature conditions can exist
for months at a time and in predominantly sandy substrate streams, overwintering
habitat may be critically limiting (Bustard and Narver 1975, Swales et al. 1986).

Many stream rehabilitation or restoration projects focus on the improvement of
instream habitat and reduction of erosion throughout the watershed. Knowledge of the
critical production areas for wild steelhead in streams can help to prioritize and
implement important watershed management activities such as protection of riparian

vegetation, a reduction in erosion, and the addition or manipulation of instream habitat.

51

The objectives of this study were to:

1) describe and quantify the spatial and temporal distribution of steelhead parr
throughout the Betsie River and its tributaries,

2) estimate reach densities and total annual production of the number of age-0,
age-1, age-2 juveniles in the watershed,

3) evaluate mortality rates and their relationship with thermal conditions in the
Betsie River and its tributaries.

Study Site

The Betsie River watershed is located in northwestem comer of the lower
peninsula of Michigan (Figure 13). The watershed is one of the smaller watersheds in
Michigan that is tributary to Lake Michigan (67,126 ha). The river begins via an
unregulated, natural surface outﬂow from Green Lake and ﬂows for approximately 79
km before reaching Betsie Lake, which ﬂows into Lake Michigan at Frankfort (44°W,
86°N). Annual mean discharge is 2.31 m3/s at Green Lake and 7.45 m3ls near the
mouth of the river. The gradient is low (mean gradient = 0.113%, ranging from 0.007%
to 0.568%) and the river ﬂows through a landscape dominated by glacial moraines
surrounding a broad glacial outwash valley. Land use in the watershed is
predominantly light agriculture and timber production, and very little urban development
exists (Gooding 1995). Several small tributaries drain into the Betsie River. The Little
Betsie River and Dair Creek are the largest with mean ﬂows of 0.51 and 0.44 m°ls .
Small spring-fed tributaries that are approximately 1.0 m wide and 5-30 cm deep are

prevalent throughout the watershed. The river has no permanent discharge gages, but

52

it

Green Lake

Betsie Lake

   
  

Homestead

 
 
  

Dair Creek
Grass Lake

:4

10 River
9
5 \

. ' Thompsonville
f 6 D
12 am

Figure 13. Sampling sites for electroﬁshing in the Betsie River watershed,
Michigan, 1993-1996. ~

53

summer ﬂows measured through the course of the study ranged from 4.0 to 12.7 m°ls
with a mean ﬂow of 5.5 m°ls .

FISI‘I surveys indicate a predominance of white suckers (Catostomus
commersonl), homyhead chubs (Nocomis biguttatus), creek chubs (Semotilus
atromaculatus), and blacknose dace (Rhinichthys afratulus) (Wicklund and Dean 1958,
MDNR, Fisheries Division Stream Surveys, 1990) in the ﬁsh community, yet the Betsie
River supports popular steelhead, chinook salmon (Oncorhynchus tshawytscha) and
coho salmon (Oncorhynchus kisutch) sport ﬁsheries. The steelhead ﬁshery is
supplemented by stocking of hatchery ﬁsh in the river, but neither coho or chinook
salmon are stocked in the Betsie River.

Recent changes in the watershed have increased interest in the potential
production of wild steelhead. Thompsonville Darn, which was located in the mid-
reaches of the watershed (Figure 13) failed in March, 1989. Downstream
sedimentation and streambank erosion is still evident from this event. However, the
upper 22 km of the watershed is now accessible to spawning migratory salmonids and

resident juvenile salmonids, including steelhead.

Methods
l stratiﬁed sections of the watershed according to differences in temperature
regimes, discharge, and gradient, resulting in the following reach designations: upper
Betsie River (sites 1-4), lower Betsie River (sites 5-8), Little Betsie River (sites 9-10),
Dair Creek (sites 13-14), and small tributaries (sites 11-12)(Figure 13). Sites were

randomly selected within the reaches and all 14 sites were sampled in the same two

weeks in late July (1993-1996) and mid-October (1994-1995) to determine steelhead
parr abundance. I used a 250 volt DC electroﬁshing barge unit and multiple pass
depletion methods to capture pan and estimate site abundance (Zippen 1958). Block
nets were placed at the upstream and downstream margins of the sampling sites prior
to electroﬁshing to meet the assumption of a closed population. At the time of capture,
I anaesthetized steelhead parr with MS-222, and measured total length (mm) and
weight (g), and removed scale samples from an area above the lateral line, below and
posterior of the dorsal ﬁn origin (Jearid 1983) . Fish were checked for ﬁn clips to
identify their origin (hatchery or wild) and were released downstream of the sample site.
I determined the age structure of the population by use of length-frequency analyses
and scale pattern analysis. Scales were also inspected to determine if the unclipped
steelhead parr had been produced in a hatchery and then remained in the river after
stocking (Seelbach and Whelan 1988). All hatchery parr were excluded from this
analysis.

Maximum likelihood estimates of abundance and conﬁdence intervals were
calculated for each site from the removal data by use of Microﬁsh 3.0 computer
software (Van Deventer and Platts 1983, 1985, 1986). Because of small estimates for
steelhead parr at many sites, a total steelhead estimate was calculated for each site
and adjusted by the proportions represented at the site in each age-class captured to

obtain an age-speciﬁc site estimate and variance according to

n... = P; m, Var (n...) = P“ 2 ° Var (m) (Equation 1 and 2)
where,
n... = the number in each age class
P.. = the proportion captured in each age-class, and
m = the estimate of the total number of steelhead parr in the site.

55

Density and variance estimates per hectare by age-class were calculated by

d,= n, lap, and Var (d,) = 1/a,,2 0’, (Equations 3 and 4)
where,
d, = the age-speciﬁc density at site p,
as = the channel area in hectares at site p,
6 , = the variance for the estimate at site p.

Mean densities for the Betsie River, Little Betsie River, Dair Creek, and the two small

tributaries were calculated by

0.. 2 d. I n, Var (0.) = 1m2 (Equations 5 and e)
where,
D. = the density for reach or tributary t, and
n = the number of sites sampled within reach or tributary t.

Extrapolations for reach or tributary (Z) abundance of part and a variance were

calculated by
Z. = a. ‘ D., Var (Z) = A2 ' (Var D.) (Equation 7 and 8)
where
Z. = reach or tributary abundance, and
a. = area in the reach or tributary t.

And ﬁnally, the overall watershed estimate of abundance (N) was calculated for each
age-class by
N = 2 Z. , Var (N) = 2‘. Var 2. (Equation 9 and 10)

where N = total watershed abundance for age-class.

56

I used a general linear model procedure (SAS) to investigate differences in
mean lengths among the channel reaches for the summer and fall data.

Estimates of the number of outmigrating steelhead smolts were determined
each May-June, 1993-1996, by use of a visual method (Chapter 1). In addition to
estimating total abundance of smolts, I determined the species composition and origin
(hatchery/wild) of the smolts by use of traps every ﬁfth day in the smolting period
(Chapter 1). Hatchery smolts were excluded from consideration in this study.

Instantaneous mortality rates were calculated according to Ricker (1975) for
sample reaches and the watershed. Mortality estimates that I considered included
mean annual mortality and within year rates for winter (October-July) and summer
(July-October) for the watershed. I assumed that by October, the parr population had
experienced the full effects of summer conditions and the population represented pre-
winter abundance. I also assumed that by July, all age-0 steelhead were large enough
to be captured by electroﬁshing gear and that all emigrating ﬁsh had left the watershed.

I monitored temperature at 7 sites in the watershed that corresponded to
electroﬁshing sites. Continuous recording equipment (Stowawaysm, Onset
Instruments, Pocasset, Massachusetts) documented houriy water temperatures in
those sites. Prior to deployment and at the end of the study, the Stowaways'" were
tested in a water bath using an ASTM thermometer to check the accuracy and to
assess the need for calibration adjustments. Stowaways'" were anchored in the
channel where the water was continuously ﬂowing, out of direct sun, and approximately
3 cm above the substrate. I calculated daily, weekly, and monthly means from the

hourly records and also extracted the daily minimum and maximum values. A few

57

missing data points were ﬁlled in with regression analysis from upstream and
downstream locations. For each year, I also calculated mean summer (June, July, and
August) and winter (December, January, and Febnlary) temperature and determined
maximum and minimum temperatures for each monitoring location. In addition to water
temperature, I obtained air temperature data from the weather station in Cadillac
(NCDC) and calculated winter severity as a function of number of days with a mean

daily temperature less than -12°C (similar to Seelbach 1987).

Results
Beam Charageristics
The channel morphometry, discharge, and thermal characteristics varied widely
among the sample reaches (Table 4). Discharge doubled between the upper and lower

reaches of the Betsie River with only a 2.7 in increase in mean channel width and a

Table 4. Hydrological and morphological characteristics of reaches sampled in the
Betsie River watershed.

 

 

Reach & Mean July Mean Channel Mean Channel
Sites Discharge (m3ls) Width (m) Depth (m)
Upper Betsie River

(1-4) 2.2 14.4 0.4
Lower Betsie River

(5-8) 5.5 17.1 0.5
Little Betsie River

(9-10) 0.5 6.7 0.3
Dair Creek

(11-12) 0.4 5.5 0.2
Small Tributaries

(12-13) 0.1 1.7 0.1

 

58

0.1m increase in mean channel depth. Mean summer discharge varied annually. In
1995, very little snow melt and lack of spring rain resulted in low ﬂow conditions. In
1993 and 1996, the opposite occurred with large snow melt ﬂows followed by high

volumes of spring rain and higher summer ﬂows for both years (Figure 14).

Water Iemperature and Environmental Variability

Water temperature regimes varied spatially and temporally throughout the
watershed. Winters of 1993-1994 and 1995-1996 were unusually cold, with over 20
days of mean daily air temperature less than -12°C (Seelbach 1987, Tables 6 and 7).
Summer, 1995 was a drought year and the mean air and water temperatures were the

highest recorded in four years of study (Tables 5 and 6).

Table 5. Summer and winter air temperature from the Cadillac, Michigan weather
station.

 

 

Summer Mean Viﬁnter Mean

Temperature (°C) Temperature (°C)
Year (Jun-Jul-Aug) (Dec-Jan-Feb)‘ # Days < -12°C
1993 18.6 - 7.9 8
1994 18.1 -1 1.8 29
1995 20.4 - 6.3 9
1996 18.0 - 8.8 20

 

* December is from the prior year

59

Discharge In July (m’ls)

 
 
 
 

8.00 -°—Upper Betsie River
""" Lower Betsie River
700 . ° ° Ir - -Little Betsie River

—e— Dair Creek

  

 

 

 

 

1.00
__________________________ 1.46
0.00 . i : ﬁ‘
1993 1994 1995 1996

Year

Figure 14. Mean discharge measured in July at sampling reaches in the Betsie
River watershed, 1993-1996.

60

Table 6. Mean summer and winter water temperature in the Betsie River at Homestead
Weir and Little Betsie River, 1993-1996.

 

 

Summer Mean Temperature(°C) Winter Mean Temperature (°C)
Year Little Betsie Betsie Little Betsie Betsie
1993 16.0 18.3 not measured
1994 15.9 17.6 1.0 0.9
1995 17.5 19.0 1.9 1.6
1996 16.0 17.8 0.7 0.4

 

Mean and maximum summer temperatures from stations throughout the
watershed were as much as 6 to 8°C different on any given day. The upper Betsie
River exhibited the greatest extremes in temperature with the warmest summer (>25°C)
and coldest winter temperatures (< 0°C). The lower Betsie River reaches were not as
extreme, but exhibited a wider range of temperatures than the tributaries, Dair Creek
and Little Betsie River (Table 7). Dair Creek was the tributary most dominated by
groundwater and was typically the coldest in the summer and the warmest in the winter.
The Little Betsie River was similar, but usually slightly warmer in the summer and cooler

in winter (Table 7).

61

Table 7. Summer and winter water temperatures in reaches sampled in the Betsie
River watershed.

 

 

Water Temperature (°C)

Summer \Nlnter
Reach and Sites Mean Max1 Mean Min2
Upper Betsie River (sites 1-4) 20.9 24.8 0.70 0.30
Lower Betsie River (sites 5-8) 18.4 20.6 1.03 0.05
Little Betsie River (sites 9-10) 16.2 18.1 1.14 0.44
Dair Creek (sites 11-12) 14.7 16.7 1.41 0.59
Small Tributaries (sites 12-13) 7.2-12.8°

 

‘mean daily maximum temperature in July (1994-1996)
2mean daily minimum temperature in January (1994-1996)
3periodic measurements during the summer (1993-1996)

Abundance and Distribution of Parr Throughout the Watershed

From 1993 through 1996, I captured 4,130 steelhead parr in summer and
autumn electroﬁshing. Length-frequency and scale analyses (699 scale samples) from
electroﬁshing in July and October conﬁrmed the predominance of two age classes
(Figures 15 and 16). Age- 0 parr comprised a mean of 63% of the catch, age-1 parr
made up 36% and fewer than 1% were age-2 parr.

Density estimates of pan varied widely throughout the watershed (Appendix B).
Values of 0 - 20 partha were common in the upper Betsie River, sites 1-4. The lower
sites in the Betsie River (5-8) yielded higher densities, ranging from 0 to 1677 parr/ha
(Table 8). The greatest densities of pan were recorded consistently in Little Betsie
River, Dair Creek, and the small tributaries, with densities frequently more than 500
pair/ha. Age-2 parr ranged in density from 16/ha to 77/ha when they were present, but

I encountered them only in the Little Betsie River and Dair Creek in 1994 and 1995.

62

Number of Parr Captured

100 "
80 " 1993

60 .. n=135

 

30 50 70 90 110 130 150 170 190 210 230 250

1994
n = 317

 

0 30 50 70 90 110 130 15 170 190 210 230 250

  
 
  
   
   

100
so ‘ 1995
80 n = 627
40

30 50 7o 90 110130 150 170 190 210 230 250

1996
n = 261

04.. ﬁﬁﬁlnmﬂum
0 30 50 70 90 110 130 150 170 190 210 230 250

Length (mm)

Figure 15. Length-frequency distribution of steelhead pan' captured in July in
the Betsie River watershed, 1993-1996.

63

.3
O
O

   
    
  

 

   

1994
80 n=574
60
'u 40
g 20
a o '1 ;, ,’ ,. u (I. i, 1 ‘ ‘ :
8 30 50 70 90 110 130 150 170 190 210 230 250
5
0.100.- 1995
“-
o n=538
E 30.-
60
g u.-
z 40..
20“ ,

 

30 50 70 90 110 130 150 170 190 210 230 250

Length (mm)

Figure 16. Length-frequency distribution of steelhead pair captured
throughout the Betsie River watershed in October, 1994-1995.

64

0000 .05 co 20:00 :0... =0 00.0.3.8 02.0.09, 2. 00.00.05 .

 

toe o.«me o.ve «.«e «.0e o.«ee hoe 3e
«.«0 o.o00 o o 0.00. o.ooo - o.««. :0 «.«« . o o.oe o moo.
o.«oee o.oe Nooe o.«e . «ope o.«e
o 1:2 o o 0.: 300 - 0.2. E. «.8 o o 0.8 o eoo.
200$ j<u
o.«oe o..~e tke o.«oe 0.0m woue o.«oe «.o«e «Ne Noe
oooo. o.o00 o.o2. o.ooe o.80 o.o~«e . 0.000. 0.30 o.«o . «.0 o o moo.
o.o.e «owe the . 0.00e o.0«.e «ooe . 0.00.H o.«e
«o: 0.009 o.ooo o.o0~ 0.2.0 30: . 0.2.2 to. 0.0: o o 0.«« o woo.
600$ .30
0.0: «owe Pope 5e 0.oe nope . oee . .
some o.ooo 00o. 0.03 .«N 2.00 no. 0.: no. o . E o o coo.
Hope P.0«e See ooze See o.oee ooe ode «ope o.e«e ooPe «we
«.02 32. o.«o. p.00. .02 0.03 to. v.2. 0.00 to... «.8 «.0 o o moo,
o.0oee o.«o«e 0.03 0.0% o.«e o.o«e .
v.3 goo: o o «0: o.«eo o 0.00 ~.o« «.0 o o o o eoo.
ste 0.2e o.«e 0.03e 0.03
o.oo~ - o o - o.oo~. v.0. 0.02. o 0.00 - - o o «ooe
200$ «0.2.220
oooe o.0«e 5.2.9 0.0ee tome eﬁe . . 0.0«e o.««e
0.0000 o.ooo ooko 0.08 0.3 2.0. 0: 0.00 5.. «.0. . o o o ooo.
0.o:e o.«oe «.~0«e o.«ooee e.o«ee e.o«e o.«oe «.«ee «ooe o.o:e Nee toe «.eoe
:0» 980. 0.88 3.0: o.oo« .08. o0: 12.0. o.o0« «.00 to so. o.o0 o moo.
e.oo«e «.«ve oeue hose o.«oee Eoe ~.o«e meme 3% .
o.«uo o.o0~. o.ooo. oooo. 3.00. 0.22 0.8 o.ooe 000 900 o o o o 000.
0.00e «.02e 0.o««e some 0. .«e o.«o«e 0.« e 0.00 e
ooooe . o.oo- o.oooe - o.o0« ooﬁ 304 ..« 3.2 . - o o «oo.
800$ «0.22:0
$3 .0: a: a c 6: 6 S 8 so a. S .«o 5 8
00.20 023130

 

 

0007002 025.03 .021 2000 05 sooooes 82.05 563% 0:0
9.2000820 3 003E00 00:0 .0 E00 00050000 7000 0:0 0.000 .2 0.022.... 00:00:08 0000 0:0 00008000 30:00 .0 030...

65

Reach estimates were calculated from the mean of site abundance estimates to
determine the contribution of each speciﬁc reach to the watershed production of
steelhead pen. I calculated reach abundance of steelhead pan for the upper and lower
Betsie River, Dair Creek, Little Betsie River, and the small tributaries. Density
estimates for age-0 pen in Dair Creek, Little Betsie River, and the small tributaries were
2.5 to 18 times greater than the densities observed in the Betsie River (Figure 17).
Estimates of age-1 parr densities for the tributary reaches ranged from 2.6 to 18 times
their density in the Betsie River. The small tributaries were the exception in 1993 and
1994, when no age-1 parr were found.

Within the Betsie River, the upper and lower reaches also differed in pan
density estimates. In all years, densities for both age-0 and age-1 parr were greater in
the lower reach. In July, 1994 and 1996, very few parr were found in the upper
reaches of the watershed and fewer parr were estimated in the lower reaches (Figure
18). A similar trend occurred in the October sampling. However, more parr were

observed in the upper reach in October than in July (Table 9).

Table 9. Density (parr/ha) of age-0 and age-1 steelhead parr in the upper
(1-4) and lower (5-8) reaches of the Betsie River, October 1994-1995.

 

 

Year Age-Class Upper River Lower River

1994 Age-0 31.9 1 15.3 846.0 1 28.4
Age-1 9.0 1 3.6 59.8 1 1.9
Age-2 1.4 1 0.8 0

1995 Age-0 8.3 1 2.2 816.8 1 37.3
Age-1 18.3 1 2.0 109.5 1 7.5
Age-2 7.1 1 0.9 0

 

66

Steelhead Parr Per Hectare

EBetsle River ILittle Betsie U Dair Creek ESmall Trlbs

 

.IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII

 

 

 

 

Age-0

 

 

 

.IIIIIIIIIIIIIIIIII|IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII

 

 

 

 

1993 1995 1996
3000' '
2500- Age-1
1993 1994 1995 1996
3000 1 P
2500 ‘ P
.I A e-2
2000 Dair Creek 9
1500 ' '
1000 ‘ "
500 *-
77 1e 34
o m— ; -- ; g
1993 1994 1995 1996

Year Sampled

Figure 17 . Comparison of age-0, age-1, and age-2 steelhead parr densities
between the Betsie River and its tributaries, July, 1993-1996.

67

IAge-O lAge-1

 

 

    

 

 

 

 

Upper Betsie River Lower Betsie River

700 ' ' 700 - .-

50° " 600" (I
3
§ 5°°" soc
S 400 " 400"
m i.
'5 300 .. .. "
.E' 300
E 200i. 200,, its
8

1001 100*:

O1 : L— : o- “i .4 LI
1 993 1994 1995 1 996 1993 1 994 1995 1 996

Year

Figure 18. Comparison of the densities of age-0 and age-1 steelhead parr in the
upper and lower reaches of the Betsie River, 1993-1996.

68

Although densities of parr were highest in the tributaries (Figure 17), when the
estimates were extrapolated by area to account for total watershed production, the
main channel Betsie River produces more age-0 steelhead (Figure 19). Using channel
areas of 84.0 ha for the Betsie River, 5.9 ha for the Little Betsie River, and 4.5 ha for
Dair Creek, the estimates of age-0 abundance for the Betsie River main channel range
from 1,171 to 11,812 age-0 parr while the abundance in the Little Betsie River, Dair
Creek and the small tributaries ranged from 314 (1 41.2) to 9739 (1 1462) age-0 parr
(Figure 19). For age-1 pan, the differences in total abundance between the tributaries
and the Betsie River are less and in some cases, a tributary contributed more age-1
parr. In the Betsie River, the total abundance of age-1 parr ranged from 1,232
Q 2,610) to 10,505 (1 3,721). The total abundance estimates in the Little Betsie River
and Dair Creek range from 827 (1 124) to 7,080 (1 418 parr). In 1994 and 1995, the
Little Betsie River abundance of age-1 parr was greater than in the Betsie River. The
small tributaries held no age-1 parr in 1993 and 1994 and only a few in 1994 and
1995. When abundance estimates for all the sampled tributaries were combined, they
contributed a mean of 56% of the total for age-0 pan and 67% of the total for age-1
parr, yet they comprised only 11% of the total channel area in the study reach (Figure
20).

In terms of watershed production, the overall watershed abundance of each
age-class of steelhead parr varied throughout 1993-1996 (Table 11). Age-0 parr
increased from 18,795 in 1993 to 32,896 in 1995, but fewer than 10,000 age-0 parr
were estimated in 1996. In 1993, the largest abundance of age-1 parr was observed,

18,485, which was more than twice the estimates for 1994-1996. However, the large

69

lBetsie River ILittle Betsie ElDair Creek ESmall Tribs

 

 

 

 

Abundance of Steelhead Parr

 

 

1993 1994 1995 1996

50000
Age-2
40000
30000
Dair Creek
20000 Little Betsie\ River /
.0... \
0 . 906 71 _ 153 1 ,
1993 1994 1995 1996

Year Sampled

Figure 19. Estimate of abundance of steelhead parr in 4 reaches throughout the
Betsie River watershed, July, 1993-1996.

70

Abundance of Steelhead Parr

El Betsie River I 4Tributaries
50000 Age-0

40000 e
30000

20000

 

10000

 

 

 

 

50000 .. A994
40000
30000

20000
10000

 

 

:-__=I=-_I:E-___r=h-_

Age-2

50000
40000

30000 Tributaries

20000 / \
1 0000
976 . 153 ,

O I i r r I
1993 1994 1995 1996

Year Sampled

 

 

Figure 20. Comparison of abundance of steelhead parr in the Betsie River and the
four tributaries that were sampled in the watershed, July, 1993-1994.

71

conﬁdence interval should be noted for 1993. Watershed production estimates for age-

2 parr were less than 1,000 ﬁsh.

Table 10. Watershed abundance estimates and 95% conﬁdence intervals for age-0,
age-1, and age-2 steelhead parr calculated from stratiﬁed reach estimates throughout
the Betsie River watershed, July, 1993-1996.

 

 

Year of

Sampling Age-0 Parr Age-1 Parr Age-2 Parr
1993 18,795 1 27,275 18,485 1 37,708 0
1994 20,569 1 13,778 8,357 1 4,924 976 1 549
1995 32,896 1 17,081 7,621 1 3,953 153 1 16

1996 9.1301 4,789 4,5051 1,499 0

 

Num r of utmi ratin molts

In May and June (1993-1996), I captured a total of 1,775 steelhead smolts by
electroﬁshing. The mean length of wild smolts ranged from 188 to 191 mm and the
age-structure of the yeariy emigration was comprised of 48-54% age-1 ﬁsh, 40-46%
age-2 smolts and 0-10% age-3 smolts. The spring of 1996 was different in that over
78% of the smolts captured were age-1 and only 20% were age-2. Less than 3% of
the smolts in 1996 were age-3. Fewer than 3,000 wild steelhead smolts left the
watershed each year (Table 8). The Betsie River produced smolts at watershed
densities ranging from 11.9/ha in 1996 to 22.2/ha in 1993 with a mean of 17.5

smolts/ha.

72

Table 11. Age structure and watershed yield of wild steelhead smolts migrating from
the Betsie River watershed, May-June, 1993-1996.

 

 

 

Year Total % Age-1 % Age-2 %Age-3 Smolts/Ha
1993 2,096 1 716 54 46 0 22.2
1994 1,84711,152 48 44 8 19.6
1995 1,534 1 1,804 53 40 7 16.3
1996 1,12511,441 78 21 2 11.9
MEAN 1,651 58 38 4 17.5

M l' ates

Annual instantaneous mortality rates (2) were calculated for the juvenile
steelhead in the Betsie River Watershed. l excluded losses due to spring emigration by
including the number of smolts estimated each year from the cohort (Table 12). Annual
mortality rates were higher for age 1-2 pan with a mean of 1.943. The mean annual
mortality rate for parr age-0 to age-1 was 1.059. Mortality for age 2-3 parr was very
large at 2.448.

Parr estimates of mortality were problematic for summer to fall in that more age-
1 and age-2 parr were estimated for fall than late summer. This may be attributed to
the fact that parr were more widely distributed throughout the watershed in the fall than
in the summer. Watershed estimates of mortality from summer to fall were much larger
in 1994 and 1995 and were also greater than for age-1 and 2 parr in 1994 (T able 12).

In 1995, low ﬂow and high summer temperatures most likely constrained pan to a few

73

areas throughout the watershed and the larger fall estimates were due to larger
abundance estimates throughout the entire watershed.
Table 12. Annual, winter, and summer instantaneous mortality rates (Z) for steelhead

parr cohorts observed in the Betsie River watershed, 1993-1996. The annual mortality
rate includes an adjustment for the cohort’s emigrating smolts.

 

 

Mortality Mortality Mortality

Year (age-0 to age-1) (age-1 to age-2) (age-2 to age-3)
ANNUAL
1993-1994 0.710 2.395 1.881
1994-1995 0.892 2.489 2.817
1995-1996 1.810 3.576 3.507

MEAN 1.137 2.820 2.735
JULY-OCTOBER
1994 3.839 1.215 2.166
1995 0.185 ' *
OCTOBER-JULY
1994-1995 1.842 3.212 1.395
1995-1996 2.396 4.633 4.396

 

*indicates more ﬁsh captured in October than in July

Within year rates of instantaneous winter mortality were calculated from October
to July for years 1994-1995 and 1995-1996 (Table 12). An adjustment was also added
to the summer estimates to reﬂect cohort survival in the emigrant smolts. Winter
mortality rates were greater in 1995-1996 than in 1994-1995 for all age classes.

In addition to total mortality from the watershed level, mortality rates were also
calculated individually for the Betsie River, the Little Betsie River, Dair Creek, and the

small tributaries (Table 13). In this analysis, I could not adjust for emigrating smolts or

74

Table 13. Annual instantaneous mortality rates (Z) throughout the Betsie River
watershed without adjustment for outmigrating smolts, 1993-1996.

 

Mortality Age-0 to Age-1 Mortality Age-1 to
Age-2
Betsie River
1993-1994 2.260 *"
1994-1995 0.987 “
1995-1996 2.795 “
MEAN 2.014
Little Betsie River
1993-1994 ° 2.056
1994-1995 1.105 '*
1995-1996 1.910 “
MEAN 1.508
Dair Creek
1993-1994 0.116 2.545
1994-1995 0.726 3.265
1995-1996 0.722 ”
MEAN 0.521

 

* larger estimate of age-1 than age-0
“ no observations of age-2 parr

parr moving between reaches, therefore these rates should be interpreted as loss rates
from the reach rather than mortality rates. Loss rates were the greatest for age-0 parr
in the Betsie River and the least in Dair Creek. Loss rates for age-1 parr were high in

the Little Betsie River and Dair Creek.

teelhea Densi Mortali Rates and Tom erature Variables

l analyzed several relations between parr density, smolt abundance and
mortality rates with discharge and several indices of summer and winter severity.
Signiﬁcant negative correlations were found between the summer densities of age-0

and age-1 pan and the mean and maximum summer temperature (Figures 21 and 22).

75

o Dair Creek I Betsie River A Little Betsie

 

 

 

3.5--
g 3.0--
i
2.5--
E
I!
d
5? 2.0--
8.
<
“a 1.5--
E
In
8 1.0--
a
ﬁ’ 0....
0.0 i i -—-——-i
10 15 20 25 30

Maximum Summer Temperature (Degrees C)

Figure 21. Correlation between the density of age-0 steelhead parr and the
maximum summer water temperature at sites throughout the Betsie River
Watershed, 1993-1996 (adjusted r2: 0.4981, p = 0.00002).

76

o Dair Creek I Betsie River A Little Betsie

 

 

 

3.5T
3.0--
2
g 2.51-
E
t
N
‘1 2.0--
7
81
f 1.5--
O
8
c 1.0"
0
a
g 0.5--
0 i :
10 15 20 25 30

Maximum Summer Temperature (Degrees C)

Figure 22. Correlation between the density of age-1 steelhead pan and the
maximum summer water temperature at sites throughout the Betsie River
Watershed, 1993-1996 (adjusted r1 = 0.6225, p = 0.000007).

77

The correlations were the strongest with age-1 parr and maximum summer
temperature.

Signiﬁcant correlations also were observed with indices of winter severity.
Weak relations were observed between the density age-1 parr and the prior winter
minimum temperature (Figure 23) and mean winter temperature and the instantaneous
mortality rate for reaches throughout the watershed (Figure 24). Using the number of
days less than -12°C as a winter severity index, a significant negative relationship
(adjusted r2 = 0.946. p = 0.018) was observed between severe winters and the estimate
of outmigrating steelhead smolts (Figure 25).

To evaluate the effects of spring ﬂow during steelhead fry emergence, the
summer estimate of age-0 steelhead was examined for a relationship with mean spring
discharge in May and June, during steelhead fry emergence. I found a negative
relationship between spring ﬂow and the number of age-0 steelhead (r1 = 0.771 and p =
0.079) (Figure 26).

Qbserved Differences in Fish Lengths

I used length as an index of growth, and tested for differences in length within a
reach through the 4 years and among reaches each year for age-0 and age-1
steelhead parr. Too few age-2 steelhead were sampled to analyze differences in mean
length. Because I found a signiﬁcant interaction between year and reach, we analyzed
the two factors separately. For age-0 parr, there were no signiﬁcant differences in

within-reach mean lengths for the Betsie River, Dair Creek, or the small tributaries from

78

o Dair Creek I Betsie River A Little Betsie

 

 

 

 

3.0 «- o
g 2.5 «-
f 2.0 --
o
o.
E
1- 1.5 --
s.
<
‘8
a. 1.0 --
'3
c
8
3' 0.5 *r
.l
0 -—H : -——l—I : l
-1.5 -1 -0.5 0 0.5 1
Minimum Winter Temperature
(Degrees C)

Figure 23. Correlation between the density of age-1 steelhead parr and the
minimum water temperature at sites throughout the Betsie River Watershed, 1993-

1996 (adjusted r2 = 0.3391, p = 0.0275).

79

I Betsie River 0 Dair Creek A Little Betsie

 

 

 

3.0 --
I

2.5 «-
E
0
~ 2.0 --
1%
E
‘3 1.5 «-
E
O
3
O
0
g 1.0 «-
s
0
s

0.5 -

O
0 i i i f 1‘
o 0.5 1.0 1.5 2.0 2.5

Mean Winter Temperature (Degrees C)

Figure 24. Correlation between mean winter temperature and the annual mortality
rate for steelhead parr (age-0 to age-1) at sites throughout the Betsie River
Watershed, 1993-1996 (adjusted r2 = 0.389. p = 0.0581).

80

3.40 --
3.35 “r
3.30 1-
3.25 “r
3.20 '-

3.15 1-

Log Smolt Estimate
I

3.10 '-

3.05 -- I

 

 

3.00 i i i i i
5 10 15 20 25 30
Winter Severity Index

Figure 25. Correlation between the winter severity index (number of days when
air temperature is less than -12°C) and the estimate of smolts leaving the Betsie
River Watershed. 1993-1996 (adjursted r"l = 0.946, p = 0.018)

81

4.6 --

 

4.5 --
E
g 4.4 --
O
s
‘0 4.3-- '
9 I
s
<
8 4.2 --
‘t’. 4.1 --
In
8’
.1 4.0»

I
3.9 : 1 ‘ 1 ﬁ-

 

2.25 2.30 2.35 2.40 2.45 2.50
Log Mean May and June Discharge

Figure 26. Correlation between the abundance of age-0 steelhead in July and
the mean May and June discharge in the Betsie River watershed, 1993 - 1996
(adjusted r2 = 0.771. p = 0.079).

82

1993 to 1996 (Table 15). In the Little Betsie River, mean lengths were signiﬁcantly
larger in 1993 and significantly smaller in 1996 than in 1994 and 1995. Between
reaches, mean length was significantly larger in the Little Betsie River than the other
three reaches. In 1993. mean length in Dair Creek was also significantly larger than in
the other reaches.

Age-1 steelhead parr were more variable in their mean length. Within reaches,
mean lengths were significantly larger in 1993 and 1995 in the Betsie River and in
1996, mean lengths were signiﬁcantly larger in the Little Betsie River (Table 14).
Between reaches, in 1995, age-0 and age-1 parr in Dair Creek and the small
tributaries were smaller than the Betsie River and the Little Betsie River, while in 1996,
age-1 parr were signiﬁcantly larger in the Little Betsie River and signiﬁcantly smaller in

the small tributaries.

83

Table 14. Mean lengths and 95% conﬁdence intervals for age-0 and age-1 steelhead
parr sampled from four reach categories throughout the Betsie River watershed. 1993-

 

 

1996.
CHANNEL
Year Betsie River Little Betsie River Dair Creek Small Tributaries
SUMMER
AGE-0
1993‘ 55.8 1: 8.4 77.0 d: 12.21 80.6 i 10.6‘ 53.2 i 5.8
1994 50.9 d: 7.3 82.4 :t 8.8 47.2 :l: 8.0 49.7 :I: 8.2
1995 52.9 :l: 8.4 81.7 :l: 8.0 48.5 i 9.4 47.8 :l: 7.5
1996 50.9 :1: 8.5 52.1 :1: 4.51 43.8 i 7.4 44.8 :l: 7.2
AGE-1
1993‘ 155.2 a: 8.4‘ 148.1 i 8.7 128.0 :l: 8.2 0 age-1 parr
1994 137.8 :1: 8.2 145.9 d: 8.2 149.6 :1: 8.0 0 age-1 parr
1995 152.3 :l: 8.8‘ 152.8 :1: 8.9 132.1 in 8.7 127.3 1: 11.0
1996 138.3 110.0 188.4 i 95* 132.3 i 8.5 118.5 :l: 8.3
FALL
AGE-0
1994 89.1 :l: 7.4“ 94.4 :l: 7.3- 73.9 :f: 7.9 74.2 :1: 7.7
1995 83.3 1: 6.6 87.2 :1: 8.8 71.2 :l: 6.9 87.5 a: 7.7
AGE-1
1994 154.4 d: 8.5 152.9 1: 7.3 139.1 :t 9.2 0 age-1 parr
1995 157.5 :l: 10.8 188.2 21:10.1“ 150.2 110.3 0 age-1 parr

 

‘ Sampling dates 10 days later in the summer than 1994-1996.
" indicates a signiﬁcant difference (P < 0.05) within the reach among the years for the
age-class

84

Different thermal rearing conditions may explain some of the observed
differences in parr length. I combined site speciﬁc temperature with optimum growth
rate information to compare the amount of time spent in the optimal growth conditions
between sites (Figure 27). Based on information found in references (Wlsmer and
Christie 1987, Hokanson et al. 1977), I deﬁned the optimal growth range as 15-17°C
with an upper growth limit of 23°C and lower lethal limit of 0°C. In the upper Betsie
River (Sites 1 and 4), steelhead parr spent very little time in the optimal growth zone
and more time in water with temperatures above the optimum limits, approaching the
zero growth limit. In the lower Betsie River (Sites 7 and 8), the trend was similar, yet
parr spent more time during the summer near and slightly above the upper optimum
growth limit In the tributaries (Sites 9 and 14), the difference in thermal regimes had
varied results for parr. In these sites, parr spent the most time in their optimum growth
zone (> 4 months) and never exceeded the upper growth limit. During the winter
months, there were few differences in time spent near 0°C in the lower Betsie River and
the tributaries, and slightly more time was spent near the 0°C limit in the upper Betsie
River. Little Betsie River was warmer than Dair Creek which may have resulted in the

larger mean lengths observed there.

Discussion
Life history characteristics which vary from the norm may indicate that rearing
conditions such as extremes in thermal variability are facilitating trends in the
population. Over half of the emigrating smolts left the Betsie River watershed at age-1.

In most other Michigan streams, the majority of smolts leave at age-2 (Biette et al.

85

Mean Daily Temperature (Degrees C)

24 Site # 1 Site # 4

I
‘
I ._4 » "k \4 1,» h.. ‘v~" ' ‘=.‘ - v’ ‘ "-" M , ‘ '__ — ,H "- . ' ,v . ,. .‘ A,.\, A 4,. 4‘. .‘ 13..., w ,, -. ‘- '.' . -' p,
. _‘ ‘v, 7.4-.7,,.‘.~,41.._1, .» _. yk 5'54 - .1 ‘ ._, . , ,. ,_'\ _, i, 1‘ .f-
\ , 4 " _ . ~,/‘t....-. .~':. A. 4 .r'.-.-.. b, 7.3.» , . ~ ‘rw. " :91- .* .I.'~r i.-
.w ’ 4* ' pv‘ .j.‘ xvi-2521,81,} 2‘03,“ 1~§A~1¢§ " 35:228.”:- {‘34 "I as" '4‘" ~‘ 43$“
) 4W” ‘3 .1 L . .4: L . 4:. .

      

. -.--~';. - Hugh-1"", “3g. . _ ,, . .. 4;
_‘ paw-xroup or ”at '1»: ._ ~' 7%“ her-uh. '9‘ 3'74“: Pin-ﬁn“; “‘r-rs-m ,A with 7‘ "J‘ém '5’

  
  

  

  
  

 

Nov-93 Jug-94 BBC-94 Jul-95 $398 Aug-96

 

Nov-93 Jun-94 088-94 Jul-95 Jai-es AJg-se

 

L

i .mw‘. 4.‘ n. j
r > .
, , O
'. . ~- .~

0
‘
-9-44», ., v. ‘3'- :73; -3 1.4.. ,3 -.< n Lyn-”1.9. .5.- . (31:11: . .vy .‘_§“.a‘.“.‘.‘,‘¢.v., ' '-' .
. ~, .‘49' 3.; r“. r." 4.7. ‘1, “ ,1 1,3,.-. 11'. ,g - 2 .‘3 i an .7..-.. » .
"t. . r; -

 

 

Nov-93 Jun-94 Dec-94 Jul-95 Jan-96 Aug-96
Date

Figure 27. Comparison of mean monthly water temperature among 6 sites in the
Betsie River watershed. The gray shaded area respresents the optimal growth range
for juvenile steelhead, the heavy dashed line is the upper zero growth level (23° C),
the light dashed line is the lower lethal level (0°C).

86

1981, Seelbach and Miller 1993). Furthermore, although the Betsie River smolts are
leaving at a younger age, their mean length (188-190 mm) is typical of the physiological
size for steelhead smoltiﬁcation (Hoar 1988) and also similar to lengths observed for
populations in which most ﬁsh leave at age-2 (Biette et al. 1981, Seelbach 1993).
These population attributes combined with the high rate of homing often observed in
steelhead, may indicate that the Betsie River steelhead has the potential to become a
discrete stock (Biette et al. 1981). Further study is required to determine if the faster
growth observed in parr throughout the watershed is a bioenergetic manifestation or
the result of phenotypic selection for faster growth rates in an environment where
survival beyond year 1 is highly improbable .

Large smolts are advantageous and can result in a high rate of return as adults.
For example, Ward et al. (1989) found that mean lengths of outmigrating smolts from
the Keogh River, age-2 age-3, and age-4 were 153, 177 and 218 mm. However,
back-calculation of the returning adults showed that mean lengths for age-2, age-3,
and age-4 were 177, 196, and 220 mm. Survival for smolts averaging 140 mm was
only 2-3% , but was 37% for 220 mm smolts. Similarly, Seelbach et al. (1994)
observed a return rate of 10- 30% for smolts stocked at lengths greater than 200 mm.
In the late 1980’s, the Betsie River ﬁshery was found to be over 90% wild ﬁsh even
though 20,000 steelhead juveniles were stocked in the river each year (Seelbach and
Whelan 1988). Betsie River smolts at age-l, 2, and 3 average 170, 188, and 203 mm
and it is likely that these larger wild smolts also retum at high rates.

Densities of steelhead parr varied widely throughout the Betsie River watershed

and can be compared with other studies and tributaries to the Great Lakes (Table 15).

87

The production of parr in the Little Betsie River and Dair Creek is comparable with Sand
Creek and Silver Creek located in southern Michigan (Dexter 1993a, 1993b). These
streams were also found to have parr with growth rates greater than the average
growth found throughout the state. We found lower densities of in the main channel
Betsie River than what were found in the Little Manistee River in northern Michigan and

also in the southern Michigan streams, Sand and Silver Creek.

Table 15. Comparison of steelhead parr densities in select streams.

 

 

General Mean Q Total Steelhead Density per hectare

Location Stream (cms) Parr/ha Age-0 Age-1 Age-2

Southem

Michigan Sand Creek 0.14 1,536 1,045 476 15
Silver Creek 0.14 1,818 436 1,345 36

Ontario Normandale

Canada Creek 0.14 600 - 2,200

Northern

Michigan Pine Creek 0.60 4,500
Little Manistee 5.0-6.0 2,295 663 10
Little Betsie 0.5 764 604 19
Dair Creek 0.4 1,123 508 - 25
Betsie River 5.5 124 46 0

 

Incorporation of naturalized production of steelhead in Great Lakes ﬁshery
management plans is a fairly recent approach. The basic premise in management of
naturalized steelhead populations is that the protection of spawning and rearing habitat
can maintain a population and is cheaper and more cost effective than restoration or

rehabilitation that often has mixed results (Biette et al. 1981, Peck 1992). Knowledge

88

of the important rearing areas and the limiting mechanisms to juvenile steelhead
production throughout the Betsie River watershed can help to prioritize protective and
rehabilitative actions to preserve and improve these critical habitat units. By using a
watershed approach and considering all tributaries as well as the main channel in
estimating parr abundance, l was able to determine the spatial distribution of juvenile
steelhead and identify speciﬁc reaches of the watershed that were important to overall
production. In a watershed similarly limiting by high water temperatures, Roper et al.
(1994) also used a basin approach to determine the distribution and abundance of
salmonids. They found that abundance estimates based on densities from a limited
area in the watershed could have overestimated the juvenile steelhead population by a
magnitude of 5 and that signiﬁcant proportions (25%) of the older juvenile steelhead
population were constrained within 12% of the total stream area. While they sampled
throughout the entire length of the stream, they did not incorporate tributary streams in
their analysis. Similarly in the Betsie River, over 50% of the age-0 and age-1 steelhead
parr were found in the tributaries which comprise only 11% of the total channel area.
Tributaries are key to the production of juvenile steelhead in the Betsie River
watershed. Not only are the preferred thermal environments provided in the tributaries,
but there may also be an advantage to growth. Peterson and Rabeni (1996) observed
2 populations of sunﬁsh in a mainstem river and a tributary and found that the ﬁsh that
both resided in and seasonally moved into the thermally moderated tributary fed more
frequently and consistently which resulted in larger ﬁsh. Similarly in the Betsie River

watershed, larger juvenile steelhead were consistently observed in the Little Betsie

89

tributary and in some years, movement from the mainstem into the tributaries seemed
likely.

Steelhead smolt production is highly variable in some cases. Production of
smolts in the Little Manistee River ranged from 11,845 to 88,425 smolts per year (97 -
713 smolts/ha), yet a consistent number of age-1 parr were produced each year
(Seelbach 1993). Huron River, tributary to Lake Superior, is in a harsher winter
environment, but is a stream similar in size to the Little Betsie. This stream was found
to produce between 1,031 and 9,141 steelhead smolts (46 - 262 smolts/ha) (Seelbach
and Miller 1993). In contrast, the Betsie River produced a limited number of smolts,
1,125 - 2,096 but the abundance estimates were fairly consistent among years.
Density estimates for smolt production from the Betsie River watershed were less than
half of the Huron River estimates, ranging from 11.9 to 22.0 smolts/ha.

When compared with high quality trout streams, the production of steelhead in
the Betsie River is very limited, yet historically this population has supported a popular
steelhead ﬁshery. Because the watershed is recovering from a major system disruption
as a result of Thompsonville Dam failure in 1989, the current production estimates may
underestimate prior production or future potential in the mainstem Betsie River.

After a similar experience with a darn collapse in the Pigeon River, Michigan,
high silt loading and sand introduction reduced the population of trout < 200 mm by
52% (Alexander and Ryckman 1986). Suspended solids have also been shown to
cause sublethal stress effects in yearling steelhead (Redding et al. 1987). In the
Betsie River, in addition to the initial impact of Thompsonville Dam failing, resulting in a

channel disequilibrium, stream banks continue to erode and contribute sand which is

90

highly mobile and observable as bedload movement during short time frames
(Newcomb, unpublished data). This mobile sand bedload may be acting to suppress
invertebrate populations (Alexander and Hansen 1986) and in concert with high
temperatures, could result in metabolic limitations for the population of steelhead parr in
the main channel. Through time, as the channel bedload reduces as a result of the
channel again approaching equilibrium, the mobile bedload may diminish and juvenile
steelhead production may increase. Further inquire is required to determine if in fact,
this scenario is occurring.

Mean maximum summer water temperatures appeared to limit the distribution of
juvenile steelhead throughout the watershed. Age-1 ﬁsh in particular showed a
stronger relationship between maximum summer temperature and site densities.

These results are similar to Roper et al. (1994) and are expected given the ease of ﬁsh
movement throughout the Betsie River. For example, at site 2, few age-1 steelhead
were found in July, yet in October 1994 and 1995, large numbers of age-1 wild and
hatchery ﬁsh were found here. This site is 10 km upstream from the hatchery stocking
location and from areas of signiﬁcant densities of age-1 parr. Spawning activity and
steelhead redds were observed in this upper reach, yet few age-0 parr were captured
here in July.

In the drought summer, 1995, summer temperatures were greatly increased
throughout the watershed. This year also had the largest age~0 year class measured in
the 4 years with a very low mortality estimate for summer to fall of 1995. Parr mean
length in the fall for age-0 ﬁsh was smaller than lengths measured in 1994 and larger

for age-1 parr. Although the exact mechanisms for this observation are not known, it

91

may be that either density dependent factors limited the growth of age-0 ﬁsh or that
growth was limited by the high metabolic costs of surviving in warm water. On the other
hand, a larger number of age-1 parr reside in the Little Betsie River and Dair Creek,
which approached the upper optimal growth limits without going above and incurring
the higher metabolic costs.

Although summer temperatures may limit distribution of steelhead parr in me
Betsie River and its tributaries, winter effects may limit smolt abundance and cohort
size. In studying a Michigan stream, Kocik (1992) found mortality rates (Z) of 1.07
during the growing season and higher winter rates of 1.57 for age-0 steelhead. His
results compare favorably with the mortality rates in the Little Betsie River for age-0
steelhead which average 1.51 and results from the Betsie River as annual
instantaneous rates for the entire watershed ranged from 0.71 to 1.81 with a mean of
1.14. Early winter acclimatization to cold temperatures has been shown to have a large
inﬂuence on juvenile salmonids due to physiological limitations in early winter (Cunjak
1988). Furthermore, the more dramatic and sudden the winter cooling the greater the
negative inﬂuence on the ﬁsh and this may give insight to the differences in mortality
rates observed in the tributaries, where groundwater moderates the rapidly cooling air
temperature, versus the mainstem Betsie River. Areas inﬂuenced by groundwater may
be the key to winter habitat throughout the Betsie River watershed and those areas are
conﬁned to the tributaries and the main channel reach between sites 58.

Winter microhabitat may also be limiting in the Betsie River watershed. The
important winter microhabitat observed in other streams such as cobble-boulder mbble

along stream margins (Everest and Sedell 1983, Meyer and Grifﬁth 1997) are limited in

92

the Michigan watershed. Boulders and cobble are rare anomalies and woody debris
along stream margins is really the only form of complex habitat available (Newcomb,
unpublished data). Complex woody debris has been shown to be important to
overwintering steelhead (Heifetz et al. 1986), the woody debris in the Betsie Main
channel is not abundant. The absence of complex woody debris can be attributed l
part due to the logging history of the watershed that has left few old stands in the
riparian zone to contribute as windfall and clearing activities for canoe passage.

Other studies on the effects of winter severity on juvenile salmonid abundance
have had mixed results. Overwinter mortality in the Little Manistee River ranged from
13 to 90% for juvenile steelhead and was related to the severity of winter (Seelbach
1987). In the Au Sable River, Michigan, Nuhfer et al. (1994) could not ﬁnd a signiﬁcant
relationship between standing stocks of age-0 brown trout and winter severity as an
index of air temperature. However, they did not analyze any older age-classes and the
winter occurred during the egg incubation phase.

The effects that adult returns or broodstock size has on the population of
juvenile steelhead in the Betsie River is unknown. There are no recent surveys to
estimate ﬁsh harvest or return. In the late 1980's, creel harvest rates were estimated to
range between 1,000 and 3,000 steelhead per year (Rakoczy and Rodgers 1987,
1988, 1990). However, the adult spawning population or total numbers in the return

was not surveyed.

93

 

Conclusions

The production of wild juvenile steelhead in the Betsie River watershed is limited
when compared with other steelhead producing streams. Although densities of juvenile
steelhead vary both spatially and temporally throughout the watershed, consistent
numbers of smolts are produced. Tributaries are a major component to the overall
production of steelhead in this river that can be thermally marginal for trout and winter
temperatures appear to inﬂuence mortality rates which are especially large for older
ﬁsh. By the use of a watershed approach to assessing production, we were able to
quantify areas critically important for steelhead based on thermal limitations and parr

distribution.

94

CHAPTER 3

ASSESSMENT OF MANAGEMENT ALTERNATIVES FOR ALTERING THE THERMAL
REGIME OF THE BETSIE RIVER, MICHIGAN

Abstract

The Betsie River in northern Michigan is classiﬁed as a marginal trout stream
because in some reaches, instream summer temperatures exceed the tolerance limits
for trout species. In 1989 Thompsonville Dam failed, which resulted in an additional 22
km of main channel accessible to migratory steelhead (Oncorhynchus mykiss) adults
and juveniles. However, water temperatures in the upper watershed may still limit the
production and distribution of these ﬁsh. The objectives of this study were to: 1)
describe the current thermal regime throughout the Betsie River watershed and
characterize it in relation to juvenile steelhead life history requirements, 2) evaluate
empirical water temperature models based on air temperature to predict instream
temperatures throughout the watershed, 3) develop a watershed physical process
model to describe and quantify juvenile steelhead macrohabitat based on thermal
criteria, and 4) develop a stream-reach physical process temperature model to evaluate
the thermal effects of removing a remaining low-head dam in the headwaters or,
alternatively, the addition of cold water from the hypolimnion of the source lake. Under
current channel conditions, summer temperatures in the upper Betsie River routinely
exceed the optimal growth limits for steelhead and sometimes the upper incipient lethal

level, with mean summer temperatures from 21-23‘C and maximum daily temperatures

95

up to 28°C. Although instream water temperatures at sites in the watershed were
strongly correlated with air temperatures (r2 = 0.81- 0.92), air temperature was a poor
predictor of winter water temperatures. The stream network model provided temporal
and spatial predictions of the weekly mean thermal regime in the Betsie River. During
summer months, the preferred thermal habitat for juvenile steelhead was limited to less
than 52% of the main channel area. The stream-reach model predicted instream
summer temperatures on a daily time-step and provided for simulation of management
alternatives for reducing instream summer temperatures. Removal of Grass Lake Dam
would likely result in mean daily summer water temperatures 2°C lower than under
current conditions in both typical-ﬂow and low-ﬂow water years for the reach from
Green Lake to Grass Lake. In the Grass Lake to Thom'psonville reach, mean daily
water temperatures were predicted to be less than 1°C lower than under current
conditions in a typical and low-ﬂow year. The addition of hypolimnetic water would
result in temperatures 36°C lower than under current conditions in the Grass Lake to
Thompsonville reach. Although these management altematives may provide better
thermal habitat for juvenile steelhead during the summer in this 22 km reach of river,
the trade-offs, including changes in the level of Green Lake, the loss of wetland habitat
and ﬁshery and recreational boating opportunities in the current Grass Lake

impoundment areas should be considered.

I

Introduction

Water temperature is an important and controlling inﬂuence on ﬁsh ecology.
Temperature inﬂuences species distributions, timing of life history events, and growth
and survival (Hokanson et al. 1977, Adams and Breck 1990, Grifﬁth 1993). When
salmon and trout reside in the upper limits of their thermal tolerance, their growth rate
can be very high, but growth rates rapidly decline when temperature exceeds the upper
threshold (Grifﬁth 1993, Elliott et al. 1995).

The Betsie River, a tributary to Lake Michigan, located in northwestern lower
Michigan (Figure 28), drains approximately 67,126 ha comprised mainly of glacial
outwash with predominantly sandy soils (Gooding 1995). The main river channel is
classiﬁed as a marginal trout stream because summer temperatures in reaches of the
river exceed the limits for growth and survival of most trout species. Temperature in the
two largest tributaries in the watershed, Dair Creek and Little Betsie River, is better
suited for trout species because large groundwater contributions moderate ambient
temperature. Although brown trout (Salmo trutta) are stocked in the Betsie River, their
numbers are very low (Wicklund and Dean 1958, MDNR Fish Surveys 1990). Rather,
the greater portion of the ﬁsh species assemblage in the Betsie River is comprised of
suckers (Catosfomus commersoni, Moxosfoma erythrurum), chubs (Semofilus
afmmaculafus, Nocomis biguftafus), and minnows (Notropis comufus, Rhinichfhys
atratulus) (\Mcklund and Dean 1958, Bonham 1975). In spite of this marginal status,
the river has historically supported a popular ﬁshery for migratory adult steelhead
(Wicklund and Dean 1958, Rakoczy and Rogers 1987, 1988, 1990).

Prior to the early 1970’s, the river was impounded by 3 dams: Homestead,

Thompsonville, and Grass Lake. Homestead Dam, a former power generating

97

z»

Betsie Lake

 

 
      
        
  
   
 
 

Grass Lake

, Dam
k Worrel
Bﬁdge

  
 
 

 

H r
Homestead

Weir

Green Lake

Little Betsie

'H \ River
Thompsonville
Dam

Figure 28. Location of the Betsie River watershed and temperature
monitoring locations (T) and ﬂow monitoring locations (H).

98

structure, was removed in 1974 and replaced with a 2 m high lamprey weir. The
original dam contained a crude ﬁsh ladder for passage of migratory salmonids but its
functionality or success for ﬁsh passage is unknown. The lamprey weir currently in
place is designed to allow for upstream passage of migratory salmonids but other
species can only pass in a downstream direction.

Thompsonville Dam, also a former power generating structure, was
approximately 3.1 m high with an impoundment around 5.7 ha (Bullen 1972). From
anecdotal reports of local residents, a few migratory salmonids were known to pass
over this dam, but it prevented the passage of most migratory ﬁsh. Prior to 1972,
Thompsonville Dam was identiﬁed as one of the sources of heating in the Betsie River
and its removal was recommended (Wicklund and Dean 1958, Bullen 1972). However,
no action was taken on the recommendation and during a ﬂood in March 1989, the river
breached Thompsonville Dam which resulted in its collapse. As a result of the dam’s
collapse, migratory salmonids were permitted access to the upstream reach of the river.
This newly accessible reach contains several high quality rifﬂes and some of the
steepest gradients in the watershed. In addition to the main channel, juvenile and
spawning steelhead gained access to the Little Betsie River, which enters the Betsie
River upstream of the Thompsonville Dam. This is the largest tributary to the Betsie
River, contributing approximately 10% of the total ﬂow to the Betsie River at this point in
the watershed. Thermal habitat limitations in the main channel still appear to limit the
population of steelhead as evidenced by the low number of wild steelhead smolts
produced from the watershed (Chapter 1) and the large differences in the spatial
distribution and densities of age-0, age-1 and age-2 steelhead parr (Chapter 2).

Current thermal limitations in the Betsie River may, in part, be the result of the

remaining low head dam. Grass Lake darn was constructed in 1951 to improve wetland

99

r..

habitat for waterfowl and northern pike (Esox Iucius ). Grass Lake Dam is located
approximately 6.7 km downstream of the river source at the outlet of Green Lake. The
impoundment area ﬂoods approximately 463.6 ha of wetlands (Carbine 1945) which
are currently dominated by cattail (Typha Iafifolia).

Steelhead, naturalized in Michigan since 1876, possess a complex life-history
that includes a 2-3 year residency in the river before they emigrate to the receiving
ocean or lake (Latta 1974, Biette et al. 1981). In the Great Lakes, steelhead spawn in
the late winter and eariy spring (Biette et al. 1981) in water temperatures ranging from
10-15.5° C (Scott and Crossman 1973). Juvenile steelhead parr fed to satiation
experience optimal growth at temperatures between 15 and 173° C, and their upper
incipient lethal temperature is 25-26°C (Hokanson et al. 1977). In behavior studies,
juvenile steelhead preferred temperatures in the range of 11-21° C (Coutant 1977,
Cherry et al. 1977). Growth ceases for parr in temperatures greater than 22° C and
less than 8° C (Hokanson et al. 1977). Steelhead parr metamorphose into smolts and
outrnigrate in the spring in water temperatures from 10 to 13° C . At water
temperatures greater than 13° C, smolts lose the capacity to smolt, revert back to parr
and may remain in the watershed for another year (Zaugg and Wagner 1973).

Interest in improving the Betsie River steelhead ﬁshery by management of the
thermal regimes has focused on 2 options: 1) removal of Grass Lake Dam, and 2)
changing the source of the initial water supply from the surface of Green Lake, to a
hypolimnetic siphon that is diverted around the Grass Lake impoundment (Bullen
1972). Both of these options incur unique biological and ﬁnancial costs. Removal of

Grass Lake Dam would result in a large loss of wetlands, waterfowl habitat, a

100

Iargemouth bass (Micmpterus salmoides) and northern pike ﬁshery, and recreational
boating opportunities. In addition, the water level of Green Lake would likely decrease
with removal of the structure as backwater effects from the control structure can be
observed upstream to the Green Lake outlet. The second option would require
installation of a water control structure at the Green Lake outlet, where none currently
exists, and would involve a large engineering effort to transport water around the Grass
Lake ﬂooding. A water diversion would also alter the thermal and ﬂow regime of the
river’s source above Grass Lake Dam. An evaluation of the effects of these
alternatives before implementation is needed in order to predict the potential beneﬁts of
such actions.

Methods used to evaluate thermal regimes in streams include two basic
approaches which differ in levels of complexity and predictability. The two general
categories of stream temperature models are empirical models and physical process
models (Barthalow 1989). Empirical models are statistical models that incorporate
measured observations and regression or harmonic analysis (Steele 1974, Dyar
1985, Dyar1997). Empirical models can be used to ﬁll in missing data, determine
historical temperatures, and can also be useful for determining thermal suitability of
water temperature for target organisms (Stoneman and Jones 1996). Although
straightforward, empirical models generally do not consider heat ﬂux or heat transport
and are thus limited in their abilities to predict changes in then'nal regimes as a result of
physical changes in the stream or surrounding landscape.

Physical process models incorporate relationships in the energy budget to
predict instream temperatures. By incorporating an energy budget, physical process

models predict water temperature on the basis of gains and losses in thermal energy

101

from processes such as radiation, convection, conduction, and evaporation. Additional
physical parameters (e.g. stream gradient, discharge, humidity, and shading) are
incorporated into physical process models to address the energy ﬂux processes.
Because of their predictive abilities in response to a change in the ecosystem, physical
process models are very powerful. The trade-off however, is that physical process
models are usually much more difﬁcult to develop and require more data than statistical
models (Barthalow 1989).

The stream reach model and the stream network model are two physical
process models with software established by the U. 8. Fish and Wildlife Service
(Theurer et al. 1987, Barthalow 1989). Both models are similar in their algorithms and
prediction of water temperatures based on an energy budget. The stream segment
model is used to consider smaller, uncomplicated reaches of a river over a limited
number of time periods. It is a straightfonrvard interactive model that presents a
simpliﬁed modeling approach and is useful in sensitivity analyses. The stream reach
model can be very useful in predicting temperatures over a limited reach for deﬁnition
of thermal habitat characteristics.

The stream network model is more complicated to initiate than the stream reach
model, largely due to the development of a conceptual network model and
establishment of 019 input data ﬁles. The network model is very powerful when
considering several channel reaches with various input sources over several time
periods. The network model is also powerful in that the software provides post-
simulation statistical evaluations. Common applications of the stream network model
include assessment of the effects of altered thermal regimes as a result of hydropower

operations (T heurer et al. 1982, Barthalow 1987, Barthalow 1991, Zedonis 1997).

102

The objectives of this study were to: 1) describe the current thermal regime
throughout the Betsie River Watershed as it compares to the temperature requirements
of juvenile steelhead, 2) develop simple empirical statistical models that relate water
temperature at sites in the watershed to meteorological inﬂuences such as air
temperature, 3) develop a watershed physical process temperature model to describe
and quantify juvenile steelhead thermal habitat in a normal ﬂow year, and 4) develop a
stream-reach physical process model to evaluate management alternatives for

changing the thermal regime in the Betsie River.

Methods

I monitored water temperature at 7 locations in the Betsie River watershed
(Figure 28, Table 16) on an hourly schedule by use of continuous logging temperature
recorders (Stowawaysm, Onset Instruments). All recorders were tested with an ASTM
thermometer to check for accuracy prior to and after deployment Daily, weekly, and
seasonal (winter: Dec-Feb, spring: Mar-May, summer. Jun-Aug, autumn: Sep-Nov)
means were calculated for each station to use in describing the current thermal

conditions from July 1993 to July 1996.

103

Table 16. Sites and reaches used for monitoring
ﬂow and temperature in the Betsie River watershed.

 

Distance upstream of

 

Site river mouth ( rkm) Model Reach
Green Lake 74.7 I Segment A
Grass Lake 68.0

Woirel Bridge 64.4 Segment B
Little Betsie River 53.8

Thompsonville 53.1

M1 15 46.9

Dair Creek 24.0

Homestead 17.7

 

I developed empirical univariate regression models of stream temperature at 7
sites in the watershed with the independent variable of air temperature. Information
from the Manistee, Frankfort, Cadillac, and Traverse City weather stations provided
mean daily air temperature (National Climate Data Center 1993-1996). Mean daily
water temperature at each site was the dependent variable in each regression model. I
used SAS statistical software (SAS 1988) to obtain an 8. signiﬁcance level, y-intercept,
and slope for each model.

The Betsie River has no permanent gages for recording river stage or
discharge. I established ﬁve gaging stations in the watershed and measured discharge
at low, intermediate and high ﬂows to establish stage-discharge relationships (Figure
28, Appendix A) (Gordon et al. 1992). Stage levels were checked and manually
recorded weekly at each of the gaging sites. The Betsie River substratum is comprised
predominantly of sand and therefore, there was a large potential for changes in

channel morphology which could have resulted in a change in the stage-discharge

104

relationship. To ensure accurate ﬂow ratings, discharge was measured at the gaging
stations each year to monitor for a shift in the stage-discharge relationship, and
establish a new rating curve if necessary (Gordon et al. 1982).

I developed a stream network model of the Betsie River that also incorporated
the ﬂows and thermal regime of the Little Betsie River with software developed by the
US. Fish and Wildlife Service (Theurer et al. 1984). Because the model is fairly
complex, I will brieﬂy explain its required components and logic for a general

understanding of the model. The stream network model incorporates 6 features: 1) a

heat transport model to predict water temperature as a function of stream distance, 2) a
heat ﬂux model to predict the energy balance between the water and surrounding
environmental conditions, 3) a solar model for the prediction of solar radiation
penetrab‘ng the water as a function of latitude, time of year, and meteorological
conditions, 4) a shade model that predicts the effects on solar input from shade
resulting from topographic and vegetative conditions, 5) meteorological corrections as a
function of changing elevation within the watershed, and 6) regression aids for ﬁlling in
missing water temperatures (T heurer et al. 1987). Components 5 and 6 were not
necessary in this application of the model.

Information processed by each of the model components is supplied by
independent data ﬁles. Seven input data ﬁles are required for the analysis: stream
geometry data, time period information, meteorology information, study site information,
hydrology site information, hydrology data, and shade data. Each ﬁle has a number of
speciﬁed parameters that must be supplied (Table 16).

I deﬁned the watershed by a number of “nodes” recognized by the network

model software as location descriptors and site speciﬁc available information for

105

processing information. The model of the Betsie River began with a headwater node at
the Green Lake outlet and concluded with an end node at Homestead Weir (Figure 29).
The Little Betsie River was the only tributary node included in the network model
because it was the only tributary with a contribution of ﬂow of 10% or more of the main
channel (Barthalow 1989). The network structure for the model also incorporated 3
validation nodes of measured water temperature for model calibration and veriﬁcation
(Figure 29). In addition to the nodes illustrated on Figure 29, 28 output (0) nodes were
included every 2 km from Green Lake to Homestead Weir to predict instream
temperatures along the length of the study reach. The stream network software
processed the input ﬁles through a series of programs to execute the model and
estimate the output values (Barthalow 1989).

To address assumptions inherent in the network model, travel time for water
from the start to the end of the study reach must be considered as the time step for the
model. Based on the low gradient of the stream channel, the reach distance from
Green Lake to Homestead Weir, and a low mean water velocity, a time step of 1 week
was used to evaluate instream temperatures. The model was supplied with mean
weekly water temperature and ﬂow information for a normal ﬂow water year, 1994.
Groundwater contributions in Michigan streams are known to have a signiﬁcant effect in
moderating the thermal regime (Hendrickson and Doonan 1972). In this model,
groundwater contributions are added by accumulating lateral ﬂow in the channel
between two deﬁned hydrologic nodes.

After the initial data entry and ﬁrst successful run of the model, I conducted a
sensitivity analysis to evaluate the sensitivity of the model output to changes in several
parameters for the Betsie River network. Parameters evaluated in the sensitivity

analysis included air temperature, humidity, windspeed, solar radiation, percent

106

Green Lake Outlet
H -

C
C

H

V Little Betsie River

 

V

- E
Homestead Weir

Figure 29. Skeleton framework and nodes for the Besie River Watershed
network temperature model. Nodes H = headwater, C=change in shading or
stream width, T= terminus of tributary, J = junction below tributary, B= branch
above tributary, V= validation node, E = end of study reach.

107

Table 17. Input ﬁles and associated parameters required for the stream network

 

 

model.
Input File Name Parameters Sources
Stream Geometry Site Latitude& Elevation U.S.G.S. Topographic Map
Manning’s N Stream Observations
Stream Width Coefﬁcient Derived1
Stream Width Exponent Derived'
Stream Shading Minimum Observed 8. Measured
Stream Shading Maximum Observed & Measured
Ground Temperature U.S.G.S.
Streambed Thermal Gradient Program Default
Time Period Time Period Name Deﬁned as 1 week
Julian Dates for Time Periods
Dust Coefﬁcient Calculated
Ground Reﬂectivity Barthalow (1987)
Meteorology Average Annual Air Temperature N.C.D.C.-Frankfort Station
For all time periods:
Air Temperature N.C.D.C.-Frankfort Station
VVIl'Id Speed N.C.D.C.-Frankfort Station
Humidity Calculated
Sunshine Ratio Estimated
Solar Radiation SSSoIar Output2
Study File Node Conﬁguration for the
Network (Table 17)
Hydrology Data File Flow Data Measured
Water Temperature Measured

Hydrology Node File

Shade File

Lateral Inﬂow Temperature

Hydrology Node Conﬁguration

Stream Reach Azimuth

Stream Width

East and West Bank
Topographic Altitude

East and West Bank Vegetation:

Height
Offset
Density

Mean Annual Air Temp.

U.S.G.S. Topographic Map
Measured

Measured
Measured

Measured
Estimated from Observations

 

108

 

sunshine, and discharge. I used results from the sensitivity analysis when considering
potential parameters to adjust in calibrating the model. After the model was fully
calibrated, mean weekly temperatures from the output nodes were evaluated at the to
characterize the thermal habitat for juvenile steelhead in the Betsie River.

To evaluate thermal habitat management alternatives such as dam removal and
a siphon draw, I modeled two stream segments (A and B) with stream-reach physical
process model software also developed by the U.S. Fish and Wildlife Service (Theurer
et al. 1984, Barthalow 1989). Segment A extended 6.7 km from the Green Lake outlet
to Grass Lake Dam and Segment B included 14.9 km from Grass Lake Dam to
Thompsonville Darn (Table 16). Similar to the network model, the stream-reach model
required 3 types of information including stream geometry (upstream and downstream
elevations, stream width, Manning’s N, and shading), meteorology (solar radiation, air
temperature, relative humidity, cloud cover, daylength, and wind speed), and hydrology
(stream discharge in and out of the reach and temperature of the water entering the
reach). Wind speed and solar radiation were obtained from the Traverse City National
Weather Service Station and cloud cover was estimated from measured solar radiation.

Relative humidity was calculated from measures of the dew point according to:

R, = [(112 - 0.1'TA + up) I (112 + 09%)]8

where R» = relative humidity, TA = air temperature, T,» = dew point temperature.
Shading was deﬁned according to Barthalow (1989) as the amount of the stream
channel covered by shade in mid-day. Based on measurements of the channel and

stream bank vegetation, I estimated shading to be 0% for segment A and 5% for

109

 

segment B. Because of the short travel distance, a daily time step was appropriate for
this application in which I used mean daily temperature and ﬂow information.

I calibrated 4 models to evaluate the two stream reaches for typical and low ﬂow
water years. Calibrated models were developed for Segment A and Segment B for a
typical- and low-ﬂow year, 1994 and 1995 respectively. Humidity was the primary
calibration parameter and it is known to vary by as much as 20% away from the
measurement site (Barthalow 1989). Solar radiation was altered from the measured
values in Traverse City within reasonable limits (1: 10%) to also assist in calibration.
Results from the stream reach models included minimum, mean, and maximum outﬂow
temperatures, width, depth, and slope of the channel with varying ﬂows, and heat ﬂux
components which indicate atmospheric and ambient conditions that contribute to heat
gains and losses. After calibrating the model for the typical- and low-ﬂow measured
conditions, segment A input parameters were altered by increasing the gradient,
decreasing channel width, and decreasing Manning’s N values, to simulate the channel
morphology that would exist with the removal of Grass Lake Dam. Initially, high
Manning’s N values were used because of the large amount of vegetation present in
the existing channel. With the dam removed, the channel would return to a slit, sand,
and gravel substratum which would indicate a lower Manning’s N value than vegetation.
Outﬂow water temperatures predicted from the simulation of Segment A with the dam
removed were then used as starting temperatures at the upstream reach of segment B
to evaluate the effects of dam removal on this downstream reach.

Evaluation of the hypolimnetic siphon alternative required changing the water
temperature at the inﬂow of segment B to reflect a proposal to withdraw hypolimnetic
water from Green Lake, divert it around the Grass Lake ﬂooding and discharge it into

the Betsie River immediately below Grass Lake Dam (Bullen 1972). I assumed that the

110

I

 

water diversion occurred in an enclosed pipe rather than a channel and that ambient
conditions had no effect on the temperature of the diverted water. I simulated the
siphon scenario using a water temperature of 72°C as the starting temperature for
segment B and a constant mean summer monthly ﬂow for a normal- (1994) and low -

ﬂow (1995) year, reﬂecting the regulated ﬂow as a result of water control structures.

 

‘1

Results
Qgrrent Thermal Regimes Measured in the Betsie River Watershed
Water temperature varied spatially and temporally throughout the watershed. l
The upper Betsie River at Green Lake and Woirel Bridge (sites 1 and 2) experienced
the warmest water temperatures in summer while the tributaries (sites 6 and 7) had the
coldest water temperatures in summer (Table 18). Mean winter water temperature was
warmer in the upper river and in the tributaries (sites 1.2, 6, and 7) in relation to main

channel locations Thompsonville, M115, and Homestead (sites 3, 4, and 5).

111

Table 18. Mean water temperature by season and year measured at sites in the Betsie
River and its tributaries.

 

Fall Winter Spring Summer
Site#939495 939495 949596 93949596

 

1 11.8 14.9 11.9 1.6 1.8 1.1 5.4 6.1 6.3 23.2 21.4 23.4 21.0
9.7 12.9 10.7 1.2 0.9 0.4 7.9 9.0 9.0 22.5 21.6 23.4 21.2
9.3 11.7 9.3 0.7 1.0 0.2 8.3 7.8 6.7 20.5 19.2 20.9 19.1
9.1 11.3 9.3 0.9 1.7 0.8 8.1 7.8 7.5 19.0 18.1 19.8 18.1

2
3
4
5 8.711.0 8.7 0.9 1.7 0.4 8.0 8.1 7.0 18.317.619.117.6
8 7.5 9.3 7.6 1.4 2.0 0.8 6.9 7.2 5.9 14.714.016.114.1
7

8.1 10.0 8.2 1.0 1.8 0.7 7.3 7.4 6.6 15.915.917.415]

 

 

 

 

Sites: 1= Green Lake, 2 = Woirel Bridge, 3 = Thompsonville, 4 = M115, 5 =
Homestead, 6 = Dair Creek, 7 = Little Betsie River.

Cumulative probability plots of temperature (exceedence curves) were
developed with mean daily temperature data from water years 1993-1995 to illustrate
temporal pattems in which the water temperature was likely to exceed the preferred
temperature range for juvenile steelhead (Figure 30, Table 19). In the upper
watershed, (Green Lake and Woirel Bridge), the preferred temperature range and
optimal growth temperatures were exceeded 20 and 30% of the time, respectively. At
Thompsonville and Homestead, the preferred temperature range was exceeded
between 1 and 7% of the time and optimal growth limits were exceeded 12-20% of the
time. In the tributaries, Dair Creek and Little Betsie River, the preferred temperature
limits were never exceeded and the optimal growth limits were exceeded only 3% of the
time in Dair Creek and 7% in the Little Betsie River. Generally, all six sites were similar

in their probabilities to be less than the lower preferred temperature (11°C) ranging

112

.32 0.3.5.0 uca .eEvéS KB? .8 .0 525.01. 32..on 2.52... .o £32m EaEaao .0 once.
EauﬁonEo. 05 «23.2... 0.9.802 8.38m 85 can 8805.02». 2.55;. 89:80.... modaﬂoqu. 3.8.8:. 0.9.83.
098. on... @3782. Ewe» .283 62.80st .05. 0.93 05 c. v.2.» Lo. 3230 35.8096 2280..th .on 952”.

£32.28 8838.8
8. cm 8 o» ow cm 3. on ow o. 8 l8 2. 2. 8 on. 9. 8. ow. 3. 9.8. 8. 8 o." 8 om 3 cm ow. o. o

n,’ -

 

     

 

 

 

 

 

            

        

       
  

v
m
MEN”... .2
N
ha; ﬁﬁwumOEOI . «0.. ha
a e. o . a em 85. 298m 2:... em
on: ow our on. ow om. ow om om cm a 8.. co an on 8 cm ov on cm or a 2: cm on ch 00 an a? on ow or o

F- r I I I I I I I I I I F- I I I I I I I I I I I Fl
v v
a m
1 3

,... .2 3....” . M... 4.: _
amass... a...“ . we 2 e
vm QN

 

e...>:oenEo_F am am 83.8.. :35 aw

(9°) amtueduiel 1318M

113

from 52% at Woirel Bridge to 61% in Dair Creek. The largest difference between sites
was the duration of time (0 - 18.6%) when the preferred thermal range for juvenile

steelhead was exceeded (Table 19).

Table 19. Probability of exceeding, meeting, or remaining below preferred and optimal
growth temperatures forjuvenile steelhead at sites throughout the Betsie River
watershed. Calculated from mean daily temperatures recorded in water years 1993-
1996.

 

 

SITE
Green Woirel Little Dair
Lake Bridge T-ville Homestead Betsie Creek

Percent of time
likely within optimum .
growth range 3.6 4.2 6.8 10.6 13.1 7.3
likely within preferred
temperature range 23.7 26.7 35.0 39.7 39.5 35.1
likely to exceed the
preferred range 18.6 17.9 10.6 1.1 0.0 0.0

likely to be less than
the preferred range 55.3 52.3 55.3 56.3 57.2 60.8

 

Temperature and photoperiod interact to stimulate smoltiﬁcation and emigration
of juvenile steelhead. Photoperiod initiates the physiological changes, while
temperature directs the rate and duration of smoltiﬁcation (Wedemeyer 1980). Water
temperature between 7° and 13°C is optimal for smoltiﬁcation. Above this, smolts may
revert back to parr and spend another year in the stream environment (Zaugg and
Wagner 1973). In the Betsie River in most years, the lower smolting temperature was

reached in mid-April and the upper threshold was exceeded by mid-May (Figure 31).

114

Water Temperature (°C)

 

| +1994 +1995 --1996|

 

30 Wolrels Bridge 30 Thompsonville
25

20

   

0.:2::::::::°-::‘r:r:::-.
31““ 20-Apr10-May30-May19alun 940' 31-Mer 20-Apr 10-May 3W8y1Nun Nul

 

3 Homestead Weir 30 Little Betsie River

25
20
15

10‘

   

.;;;;;;;,;;0.::1:::::::
31-Mar 20—Apr 10-May 30-May 19dun 94iui 131-Mar 20-Apr 10-May 30-May 119-Jun Nul

 

Date
Figure 31. Spring water temperatures during steelhead smolt emigration in the

Betsie River watershed, 1994-1996. Shaded arera denotes smoltiﬁcation
temperatures, above which smolts may revert to parr.

115

 

In the Little Betsie River, the lower smolting temperature threshold occurred in mid-April

and suitable temperatures for smolting extend into mid-June (Figure 31).

Empirical Models of Water Temperature

Air temperature data from neighboring climatological stations were used to
predict daily mean water temperature at Betsie River stations. I initially tested data
from stations at Cadillac, Frankfort, Manistee, and Traverse City (NCDC 1993-1996).
Data from all climatological stations produced signiﬁcant relationships (p < 0.001) with
an adjusted r2 > 0.60. The Frankfort station accounted for the greatest variability in
water temperature with r2 values ranging from 0.826 to 0.908 at sites throughout the
watershed (Table 20). Water temperature at tributaries and downstream sites were
more strongly related to air temperature than at the upstream sites, Woirel Bridge and
the Green Lake outlet.

Table 20. Results of regression modeling water temperature as a function of air
temperature measured at Frankfort, Michigan. All results are signiﬁcant, P < 0.001.

 

 

Site R2 Coefﬁcient (Slope) Y-intercept
Green Lake 0.826 0.756 4.964
Woirel Bridge 0.866 0.781 4.931
Thompsonville 0.897 0.717 4.261
M115 0.906 0.858 4.549
Homestead 0.910 0.834 4.417
Little Betsie River 0.903 0.571 4.184
Dair Creek 0.908 0.511 4.015

 

Predicted temperatures were plotted with observed temperatures to evaluate
the empirical models (Figures 32 and 33). In all cases, the empirical models were poor

predictors at very low air temperatures, from -10 to -5°C, likely due to the effects of ice

116

 

Green Lake

  

20‘-

 

 

 

 

Predicted Water Temperature ( °C)

ctr-38888

-5 I
.10 .
-15

 

 

 

-5 0 5 10 15 20 25 30
Measured Water Temperature (°C)

Figure 32. Predicted water temperature (symbols) as a function of air temperature
and mean daily water temperature (line) at sites in the Betsie River, 1993-1996.

117

Predicted Water Temperature (°C)

25 M-115

 

'15. i f 1‘

 

Homestead

 

 

-15 . . . . . . . .
-5 0 5 1 0 1 5 20 25 30

Measured Water Temperature (“C)

Figure 32 (continued). Predicted water temperature (symbols) as a function of air
temperature and mean daily water temperature (line) at sites in the Betsie River,

1993-1996.

118

I

 

30
25 Little Betsie River

 

25 30

Predicted Water Temperature (°C)

 

 

I5 0 5 1'0 15 20 25 30
Measured Water Temperature (°C)

Figure 33. Predicted water temperature (symbols) as a function of air temperature
and mean daily water temperature (line) at tributary sites in the Betsie River
watershed, 1993-1996.

119

and snow cover. The largest error from the observed values was observed at the
Green Lake monitoring location (most likely due to the heat capacity of the lake) and
became progressively smaller in downstream locations, M115 and Homestead Weir.
With a few exceptions, air temperature was a very good predictor of water temperature

in the tributaries, Dair Creek and Little Betsie River at air temperatures above -7.0°C

(Figure 33).
Stream Netxgrk Temperature Model Results I

i?-

All data input ﬁles were assembled and entered for the stream network ﬁle
(Appendix C). Results of the sensitivity analysis indicated that changes to air
temperature and humidity produced the greatest changes in the predicted temperatures
(T able 21). Discharge was also sensitive although in all cases, the resulting mean error
was less than 1 degree. The model was relatively insensitive to other variables, such
as percent sunshine, solar radiation, and windspeed and changes in these parameters

resulted in little change in the temperatures predicted.

Table 20. Mean error at 3 validation nodes in the Betsie River and Little Betsie River
resulting from 5% increases and decreases a single parameter.

 

 

Site
Little Betsie Betsie River Betsie River
Hwy 669 Thompsonville Fred’s Landing
+ 5% -5% +5% -5% +5% -5%
Parameter

Air Temperature -0.64 0.64 -0.46 0.46 -0.68 0.67
Humidity -0.27 0.27 -0.24 0.24 -0.34 0.34
\Mndspeed -0.04 0.04 0.02 -0.03 0.02 -0.02
Solar Radiation 0.01 0.01 -0.12 0.12 -0.12 0.12
% Sunshine 0.00 0.00 0.01 0.01 0.01 -0.02
Discharge 0.16 0.64 0.48 -0.08 0.41 0.39

 

120

 

I evaluated model performance on the basis of 3 error values for each validation
node. Mean error values were the averages of the differences between observed and
predicted water temperatures at each validation node. Median error represented the
value in which at least 50% of the predicted water temperatures were within this range.
Maximum error was the greatest deviation of predicted temperature from the observed
water temperature. I determined my decision criteria for calibration as:

1) no more than 10% of the predicted temperatures greater than 1°C from
observed temperatures,

2) no single predicted temperature greater than 1.5°C from observed
temperatures,

3) the mean error for the validation node should be less than 0.5°C, and

4) no spatial or temporal trends in error (Barthalow 1987).
Although the results for the initial run of the network model were fairiy good (T able 22),
for several weeks, predicted temperatures ranged above the values deﬁned by my
decision criteria. For weeks and locations where the mean weekly values were beyond
the deﬁned calibration decision criteria, I altered selected input parameters to calibrate
the model. Humidity and solar radiation were altered in most cases for calibration and
discharge was altered within i 20% for the Little Betsie River in which gage
measurements were the least certain, but water temperature was known. The resulting
calibrated model predicted stream temperatures within the 4 standards outlined in the

decision criteria (Figure 34).

121

—o—OBSERVED -t:l—PREDICTED
25 .. Little Betsie River

 

 

I.

  

 

 

O 1 r i i i i r
1-Oct 20-Nov 9—Jan 28-Feb 19-Apr 8-Jun 28—Jul 16-Sep

 

25
20

 

15
10«

 

 

 

Mean Weekly Temperature (°C)

 

 

 

 

 

Date

Figure 34. Comparison of mean weekly observed water temperature and
network model predicted temperature at validation sites in the Betsie River and
Little Betsie River.

122

Table 21. Mean, probable, and maximum error for the stream network model
predictions at 3 validation sites in the Betsie River watershed.

 

 

Site
Little Betsie Betsie River Betsie River Entire Stream
Hwy 669 Thompsonville Fred’s Landing Network
Initial Model Run

Mean Error -0.51 0.36 1.66 0.51
Probable Error 2.53 2.01 1.38 2.10
Maximum Error -8.62 6.83 -5.89 -8.62

Final Calibration Run '
Mean Error 0.16 -0.07 0.40 0.16
Probable Error 0.64 0.48 0.35 0.52
Maximum Error 1.89 1.58 1.31 1.89

 

Once calibrated, mean weekly temperature results were predicted from the
stream network model for every 2 km in the Betsie River from Green Lake to
Homestead Weir. Spatial and temporal patterns emerged in the annual thermal regime
of the Betsie River. At the beginning of the water year (October) from Green Lake
downstream to river kilometer (rkm) 55, the river was warmer and in the preferred
thermal range for juvenile steelhead (Figure 35). By November, all sites in the
watershed were below the preferred range and these conditions persisted until April
(Figures 3537). Rapid cooling occurred in December and by January and through
Febniary, mean weekly temperatures hovered near 1-2°C throughout the length of the
river (Figure 36). Warming patterns emerged in early spring (Figure 37). In March, the
water quickly warmed from 1°C to 5°C downstream of rkm 50, staying slightly cooler in

the reach near Green Lake. A similar pattern was observed for April as water

123

 

October

3 o—‘l—Oct +8-Oct ~15-Oct "it-"22'0“ +29-Oct

 

3.939% 9,. 9:9
9
‘ 4 $251.54“ 65.2.
‘ ﬁzz/é; 3‘53“ $23

“0'.

 

 

November
+5-Nov»c~12-Nov~--9~--19—Nov-~«t«-26-Nov

{m2 mm 83m
5?; .
Etta" 99,“; $33 39.3 $53593

92 933

$3.?

Mean Weekly Temperature °C ~

 

 

 

Downstream U stream
River Kilometer p

Figure 35. Longitudinal distribution of mean weekly water temperature as
predicted for the Betsie River during the fall and the start of the water year,
October - November in a normal ﬂow year. The shaded box is the preferred
temperature range for juvenile steelhead.

124

Mean Weekly Temperature °C

December

+3-Dec +10-Dec ----s=9-~--17-Dec mx-24-Dec

 

‘ .9 '» = 4- can
W99.

0 ‘ , 9 Q ,, __, — -.-. Ken.“ 'Lhﬂ‘i‘v.
10 20 30 40 50 30 70 80

January
+1-Jan +7-Jan14—Jan «~21-Jan +28-Jan

  
   
   

$333 {3: ’j_ " '5
\ S ‘ A“

939 £393}
1 1‘) P

  
  

   
     
 
  
  
  
   
   
        

2:92 '
£23m

3‘3:th

  
 
  

   

C“ ‘
9 Mam
~- “$999? 3)

10 20 30 40 50 60 7O 80

February
+4-Feb +11-Feb -~.<5E:."~18-Feb -x—~25-Feb

   
 

  
 
  
 
   
 

  
  
 
    

W9. 9 9... .469;
~ 3 ~ “25% t9; 9,9393%‘3‘33 3
‘ .4}: :- .

,\

  

 

$§¢§<v ‘ k ‘

' "13" ‘9 -. ‘uv:

tNﬁg-s. ‘u . -'
(23.. , \x 5;. >.<\ ~-_ .\
9.x 2 mx ., . .-
9 ate . , ,.z .
3 Kg ‘ at; ‘§?§§§£2g€.’§:t >‘
: > >:‘-* agent-*9 4‘
3339 9323‘ 992-9392..
mt: tmtmm -.~.«

~
9.

Q ‘<
,
a“???
= is

   

-.

     

    
 

    
 

     
  

10 20 30 40 50 60 70 80

Downstream _ _ Upstream
River Kilometer

Figure 36. Longitudinal distribution of mean weekly water temperature as
predicted for the Betsie River during winter, December - February, in a
normal-flow year. The shaded rectangle is the preferred temperature range
forjuvenile steelhead.

125

Mean Weekly Temperature °C

March
+4-Mar +11-Mar 18-Mar -----><-'25-Mar

  
     

“233333333

{ti ~t2 at: -
iii Erika:
"9:3: :2 3‘ ‘

  
  
 
  

   
  
  
 
 
   
 
  
 

      
 

     
   

,
. 5 ‘ » $3
«at \x. ‘1“
Hz: 9‘

5\\\\\

   

tic;
Mu - m
t£E<<<tzzt

 

9
£39..

      
   

 

 

 

 

ﬁ“$§§
t t “
1- ' >5)?

   
  

 

 

 

 

     

53:?

um»

Ila-(av

    

.w
$5

 

Downstream , , Upstream
River Kilometer

Figure 37. Longitudinal distribution of mean weekly water temperature as
predicted for the Betsie River during spring, March - May, in a normal flow year.
The shaded box is the preferred temperature range forjuvenile steelhead.

126

temperatures warmed to 10°C below rkm 50, and while remained 2-4°C colder in the
river upstream of rkm 50. Water temperatures increased quickly between weeks in
May and water temperatures entered the preferred range for juvenile steelhead,
gaining 34°C within a week. The river remained cooler upstream, but the area

decreased to only the 10 km immediately downstream of Green Lake. In May, the

 

l

entire length of the river was within the preferred range for juvenile steelhead (Figure
37).

Summer temperatures throughout the river sometimes exceeded the preferred
range for up to 47% of the stream length. By late June, upstream of rkm 62, the water
temperature exceeded the preferred limits (Figure 38). In July, the trend continued with
the river upstream of rkm 50 beyond the preferred limits and downstream of rkm 50
within the preferred range. A cooling pattern emerged in August, however, upstream of
rkm 50, when the water temperature reached and exceeded the upper preferred

temperature limit (Figure 38). In September. nearly the entire length of the Betsie River

 

was within me preferred temperature range (Figure 39).

§tream §ggment Water Temperature Model Results

Based on the limited 4 years of water temperature and hydrological
observations, 1994 represented the typical flow and climatic conditions in the Betsie
River watershed and 1995 represented a low-ﬂow year. I used data for these two
summers to develop a stream reach temperature model to evaluate management
alternatives. The model for segment A (Green Lake to Grass Lake) was calibrated with
the water temperature monitoring station 2.4 km downstream of Grass Lake Dam. I

assumed that only minor differences in water temperature and discharge existed

127

Mean Weekly Temperature °C

June
+3-Jun +10-Jun ..

 

--17-Jun xZ4-Jun

 

 

 

August

 

30 +5-Aug --I12—Aug - - 19-Aug ----x~--- 26-Aug

 

 

Downstream , _ Upstream
River Kilometer

Figure 38. Longitudinal distribution of mean weekly water temperature as
predicted for the Betsie River during summer, June - August, in a nonnal-ﬂow
year. The shaded rectangle indicates the preferred temperature range for
juvenile steelhead.

128

September

+2-Sep ~I-9—Sep 16-Sep --x ----- 23-Sep

 

 

9
2
2
8
8
E
O
p.
>.
I
8
35
C
80. i i i i i ii
2

1 O 20 30 40 50 60 70 80

Downstream , _ Upstream
River Kilometer

Figure 39. Longitudinal distribution of mean weekly water temperature as
predicted for the Betsie River during September in a nonnal-ﬂow year.
The shaded rectangle indicates preferred temperature range forjuvenile
steelhead.

129

 

between the Grass Lake Dam outlet and Woirel Bridge. because discharge
measurements did not differ between the sites.

The calibrated temperature model for segment A had an average error of
1.57‘C and 152°C for 1994 and 1995 and most errors were less than 2°C different
(Figure 40). Predicted temperatures for Segment B had mean differences of 058°C
and 034°C for 1994 and 1995 (Figure 41). In all cases. the errors were randomly
distributed and the models did not consistently over- or under-predict the measured
values.

In the ﬁrst simulation. l modeled the effects on water temperature as a result of
removing of Grass Lake Dam by reducing the channel width and increasing the
gradient, both of which would result with the removal of the structure. Mean daily water
temperature declined in this scenario (Figure 42, Table 21 ). In segment A, during the
typical-ﬂow year. daily mean water temperature ranged from 0.88 to 6.51°C (mean =
372°C) less than under current conditions. In the low-ﬂow year, daily mean water
temperatures averaged 394°C less and ranged from 0.88 to 787°C less than under
current conditions. For the Grass Lake to Thompsonville reach. water temperature
averaged 122°C less in the typical year and 055°C less in the low-ﬂow year with the
dam removed (Figure 43).

A siphon from the hypolimnion of Green Lake to the outlet of Grass Lake has
been suggested as a remedy to the thermal limitations in the upper Betsie River (Bullen
1972). Logistics in this plan included routing the water around the Grass Lake
impoundment and directing it through either a pipe or open culvert into the Betsie River
channel below the Grass Lake Dam. l simulated this alternative with the reach model

using the climatic conditions observed for the typical and low-ﬂow year. Flows were

130

Mean Daily Temperature (° C)

 

[ -o—Predicted +Observed]

 

} .

kg} \ \
cet§$§§ E ‘ ‘ it 2 .5
w “Mt-.24

20 A ' ' ' jg:

  

0 . : : : :
6/1 I94 6/21/94 7/1 1l94 7131/94 8I20/94

30
25

20

i5k“,
)) $1?) 1‘

    

 

611/95 6I21l95 7/1 1/95 7/31/95 8I20l95
Date

Figure 40. Measured and predicted mean daily summer temperatures from the
stream reach temperature model for segment A. Green Lake to Grass Lake. in the
Betsie River during a typical-ﬂow (1994) and low-flow water year (1995). The large
shaded area indicates water temperatures preferred by juvenile steelhead and the
inner rectangle identiﬁes optimum growth temperatures forjuvenile steelhead
(Hokanson et al 1977. VWsmer and Christie 1987).

131

Mean Daily Temperature (°C)

 

-o—Predicted+ Observed

 

30--

25-

 

 

 

 

0 . i i i :
6/1/94 6/21/94 7l1 1I94 7/31/94 8I20l94

3° 1995

 

 

 

0 : : 1' :
6/1/95 6/21/95 7l1 1l95 7I31/95 8I20/95
Date

 

Figure 41. Measured and predicted mean daily summer temperatures from the
stream reach temperature model for segment B. Grass Lake to Thompsonville, in
the Betsie River during a typical-flow (1994) and low-ﬂow water year (1995). The
large shaded area denotes temperatures preferred by juvenile steelhead and the
smaller inner rectangle indicates the optimum growth temperatures forjuvenile
steelhead (Hokanson et al 1977, Wlsmer and Christie 1987).

132

Mean Daily Temperature (°C)

 

-o-With Dam +Dam Removed J

 

 

 

O : : : :
6/1/94 6/21/94 7/1 1[94 7I31/94 8I20l94

1995

 

 

0 . : : : :
6/1/95 6/21/95 7/1 1/95 7I31/95 8I20l95
Date

Figure 42. Comparison of mean daily summer water temperatures under current
conditions with Grass Lake Dam and simulated conditions of dam removal in a
typical- flow (1994) and low-flow (1995) water year for Segment A. The shaded
area indicates the temperature range preferred byjuvenile steelhead and smaller
rectangle depicts optimal growth temperatures (Hokanson et al. 1977. \Msmer
and Christie 1987).

133

 

 

 

l -0-With Dam +oam Removed

 

  
  
 

  
   

(
‘4 ~44.
N 9’55

 

   

n l I l

0 I I I I l
6/1/94 6/21/94 7l1 1l94 7I31/94 8I20l94

Mean Dally Temperature (°C)

 

n A 1 A

o I I I l I
6/1/95 6/21/95 7/1 1l95 7I31/95 8I20l95
Date

Figure 43. Comparison of mean daily summer water temperatures from Grass
Lake Dam to Thompsonville under current conditions with Grass Lake Dam
present and simulated conditions with the dam removed for a typical-ﬂow (1994)
and low-ﬂow (1995) year. The large shaded area indicates the preferred
temperature range forjuvenile steelhead and small inner rectangle depicts
optimum growth limits (Hokanson et al. 1977. Wlsmer and Christie 1987).

134

averaged from measurements in 1993-1996 to obtain mean monthly ﬂows of 57.6,
59.1, and 63.9 for June, July, and August, and l designated 72°C as the outlet
temperature. This was the measured temperature during the summer months for the
hypolimnion of Green Lake (Newcomb, unpublished data). This scenario resulted in
mean daily water temperatures less than those simulated with the dam removal (Table

23. Figure 44).

Table 22. Mean daily water temperature in June. July. and August for the river at the
downstream section of Segments A and B, in the Betsie River under current conditions
and under simulated conditions of removing Grass Lake Dam and releases of
hypolimnetic water from Green Lake at Grass Lake Dam. Data from sites at M-115,
Homestead, and the Little Betsie River are also included for comparison.

 

Typical-F low Year (1994) Low-Flow Year (1995)
°C

 

Management Altemative °C
June July August June July August
Current Conditions
Segment A 21.2 22.7 21.0 24.0 24.2 25.2
Segment B 19.0 20.3 18.7 21.0 20.7 21.5
M-115 17.9 19.0 17.4 19.8 19.5 20.2
Homestead 17.4 18.4 16.7 18.8 18.8 19.4
Little Betsie River 15.9 16.7 15.0 17.1 17.4 17.9
Dam Removal
SegmentA 18.4 19.9 19.3 21.8 21.5 21.9
Segment 8 18.3 19.6 18.1 20.9 20.1 21.2
Hypolimnetic Discharge
Segment 8 15.2 15.6 13.7 18.9 18.3 19.0

 

During both the typical-ﬂow year and the low-ﬂow year, the hypolimnetic release
resulted in water temperatures ranging from 12 to 21°C with a mean of 182°C at
Thompsonville. When compared with the dam removal alternative, the cold

hypolimnion contribution provided mean water temperatures 2.7-36°C cooler than

135

 

p.

 

 

Eo— Dam Removed + Siphon]

 

 

 

O u i i i i
6121/94 7/1 1194 7I31/94 8120/94

Mean Dally Temperature (°C)
3
E

 

 

0 . : r : :
611195 6121195 711 1195 7131195 8120195
Date

Figure 44. Comparison of mean daily water temperatures simulated for the dam
removal and the addition of cold hypolimnion water in the Grass Lake to
Thompsonville reach of the Betsie River in a typical-ﬂow (1994) and low-ﬂow
(1995) water year. The large shaded area is the water temperature range
preferred byjuvenile steelhead and smaller rectangle depicts optimum growth
ranges (Hokanson et al. 1977, Vlﬁsmer and Christie 1987).

136

 

temperatures simulated without Grass Lake Dam in Segment 8 (Figure 44). In warm,
low-ﬂow years such as 1995, instream temperatures in the Thompsonville reach still

exceed the optimum growth temperatures for juvenile steelhead (Table 23).

Discussion and Conclusions

Both empirical models and physical process models are useful tools in
assessing instream temperature. Empirical models are most valuable for predicting
water temperature based on one or a few other variables. However. empirical models
can be limited in predicting water temperature as a response to morphological changes
within the channel or other variables. such as riparian shading, which would alter the
heat dynamics within the river. Physical process models are necessary to model the
changes in thermal regime that result from morphological or riparian changes.

Both the stream network model and the stream reach model were sensitive to
changes in air temperature. Several researchers have documented the dominating
effect of air temperature on stream water temperature (Smith and Lavis 1975, Crisp
and Howson 1982. Barthalow1989, Stoneman and Jones 1996). The regression
models provided in this manuscript can be used to predict water temperature in future
modeling applications in the Betsie River under the current instream conditions. If there
were signiﬁcant changes in channel morphology or if Grass Lake Dam were removed,
new empirical models would need to be derived.

Multivariate empirical models that incorporate an additional aspect. such as
discharge, may predict water temperature more accurately than single univariate
models. The empirical models for water temperature in the Betsie River watershed

were simple univariate models (air temperature = independent variable, water

137

A }

 

temperature = dependent variable), resulting in variable but high correlation values
throughout the watershed (0.8280910). Improvement to these models might occur if
an additional variable. such as a lag time value were incorporated to account for the
ongoing heat gains and losses in preceding days.

Dom physical process models. the stream network model and the stream reach
model were useful for examining the thermal regime in detail in the Betsie River and
evaluating consequences of changes in the watershed. The stream network model
results provided for a detailed evaluation of thermal habitat availability based on
temperature data collected at 7 stations.

Juvenile steelhead residing in marginal environments can face a number of
consequences. In those environments that exceed the optimal growth limits deﬁned by
temperature, ﬁsh can experience a decline in growth rates or exhibit zero growth
(Hokanson et al. 1977). Alternatively, ﬁsh will move out of an area that is too warm and
into refuge areas which may already be occupied by resident ﬁsh (Bilby 1984),
suggesting a temporal density dependent limitation and competition for space. Finally,
when temperatures are near or at the lethal levels, mortality can occur either as a direct
result of the heat, or from a secondary cause, such as infection or starvation (Coutant
1978).

On the other hand, juvenile steelhead residing in locations where the thermal
regime remains below the optimal growth limits may experience slower growth and later
age of smoltiﬁcation. Although an additional year of residency in the stream may
subject these ﬁsh to mortality in the stream environment, they are less likely to
succumb to heat associated mortality or stress.

Under the current channel and ﬂow conditions. the most thermally suitable area

for juvenile steelhead in the summer is between Thompsonville and Homestead Weir in

138

 

 

the main channel of the Betsie River and in the tributaries. In this reach, groundwater
contributions, as evidenced by increases in downstream discharge, provide a cooling
effect to moderate summer temperatures. Locations above this reach, receive
proportionately less groundwater and may provide marginal thermal habitat in typical-
ﬂow years. but during low-ﬂow years. conditions can exceed the lethal limits. Thermal
conditions in me Little Betsie River are optimal for juvenile steelhead: summer
temperatures only rarely warm above the optimal growth rates. Dair Creek
temperatures are lower than optimal growth rates for much of the year. but this is not
likely to limit the number of juvenile steelhead produced in this stream.

Spring temperature regimes in the Betsie River give a very short window of
opportunity for the process of smoltiﬁcation and emigration to occur, particularly when
the warming trend occurs rapidly. This short duration may cause ﬁsh to stay in the river
as the water temperature increases above the required temperature for smoltiﬁcation
(Zaugg and Wagner 1973, Wedemeyer 1980) and may have serious implications for
stocking ﬁsh. If smolts are stocked too late or too far upstream in the watershed, they
will not have time to emigrate before reverting back to parr and residualizing within the
river. This would put hatchery ﬁsh in direct competition for space with wild produced
ﬁsh and this competition would be most extreme in marginal habitats during years of
low ﬂow.

If Grass Lake Dam were removed, the largest improvement in temperature
would be observed in the reach from Green Lake to Grass Lake. Because of the lack
groundwater contributions, little improvement to the thermal regime would be observed
in the Grass Lake to Thompsonville reach during either a typical-ﬂow or low-ﬂow year
(Figures 38 and 37). The hypolimnetic ﬂow diversion and release would provide no

thermal beneﬁt for trout species in the Green Lake to Grass Lake reach but it would

139

 

 

decrease the mean summer temperature in the Grass Lake to Thompsonville reach
(Figure 38). When compared with downstream locations such as M-115, Homestead
and Little Betsie River. mean monthly water temperature under the dam removal
scenario is still greater than the mean monthly temperature at sites with greater
densities of juvenile steelhead (Chapter 2). Because the river begins as a surface
outﬂow from a lake. the upper river water temperatures are dominated by the lake
surface temperatures which tend to be fairly stable as a result of the ability of the large
body of water to retain heat This stable tendency. combined with the large outﬂow of
water that begins the river (approximately 0.9 - 3.3 in3 Is during the summer) and daily
climatic conditions is what directs the thermal regime in the upper watershed until
enough groundwater enters into the channel and decreases the water temperature.
Areas where groundwater is seeping into the stream channel may provide
important micro-thermal habitat areas that give refuge to ﬁsh during periods of high
water temperature. Bilby (1984) identiﬁed 4 types of micro-thermal habitat areas as
lateral seeps, pool bottom seeps. cold tributary mouths and ﬂow through the
streambed. In areas of the Betsie River watershed, where the river ﬂows near glacial
moraines, lateral seeps are evident at the surface near the channel, although their
inﬂuence has not been quantiﬁed other than as an increase in discharge. These types
of features are limited in the reach between Grass Lake and Green Lake, and may be
too few to provide refuge to a large number of juvenile steelhead. One potential habitat
management alternative is to identify these features in this reach and provide instream
structures to prohibit mixing of the warm and cold water as much as possible (Bilby
1984). Objects, such as woody debris, that would divert the warm water or create a

small pool effect may increase these potential refuge areas. Additional research would

140

 

 

 

need to be conducted to assess the existence of these cooler micro-thennaI areas and
the feasibility of their development.

Other research found thermally stratiﬁed pools to be an important microhabitat
for adult steelhead in the summer months (Nielson et al. 1994). However, the reach
between Grass Lake and Thompsonville has a wide channel. likely a remnant effect
from logging in the 1800’s, with a steep gradient. Pools are scarce nor are they deep

enough to stratify.

141

 

SUMMARY

The results of my investigation on the Betsie River steelhead resources and
thermal limitations provided information on juvenile steelhead population dynamics in a
thennally marginal watershed. From 1993-1996. the watershed produced a limited
number of wild steelhead smolts, <3,000 per year, of which approximately 50% were
age-1. Although the smolts emigrated at a young age, they were comparable in size to
smolts observed in the Little Manistee River, 180-200 mm (Seelbach 1983). Time—
lapse videography was a useful approach to obtain an estimate of smolts, although the
20 minute direct observation samples were also effective. In both cases, sampling 60%
of the days throughout the duration in the smolt run would have produced estimates
and errors similar to those obtained when sampling the entire duration.

I found stark differences among the juvenile steelhead parr densities at sites
throughout the Betsie River watershed in the summer months. Estimates ranged from
0 to over 3,000/ha at sampling sites in the watershed. The densities of juvenile
steelhead were lowest in the upper reach of the river above Thompsonville and
temperature studies indicated a summer thermal range above the desired thermal limits
in this reach of river. The largest densities of juvenile steelhead were in the tributaries
and in the Betsie River below Thompsonville. Results of the temperature modeling
study further indicate that preferred thermal habitat is limited to the 33 km above

Homestead Weir during the peak of the summer heat in July. This reduction of

142

 

 

available habitat based on thermal suitability may indicate temporal, density dependent
competition, especially for the larger, age-1 and age-2 juvenile steelhead.

The densities that I observed for juvenile steelhead in the tributaries were similar
to those found in Sand Creek and Silver Creek in southern Michigan (Dexter 1993a and
1993b). These results may suggest similar levels of productivity that could be used to
assess production of juvenile steelhead from different watersheds. In particular, the
results from the Betsie River are most applicable in watersheds where the main

channel is marginal thermal habitat for juvenile steelhead.

 

Although summer thermal habitat appears to limit the distribution of juvenile
steelhead, a shortage of winter habitat may inﬂuence mortality rates. Annual
instantaneous mortality for juvenile steelhead in the watershed ranged from 0.710 to
3.578 and net channel losses were greatest in the Betsie River and least in the
tributaries. Mortality rates were not related to summer temperatures, but were
signiﬁcantly related to minimum winter mean daily temperatures and winter severity.
After the coldest winters, the highest mortality rates were observed and the fewest
smolts emigrated in the spring.

I used a watershed approach in evaluating the juvenile steelhead abundance
and distribution in the Betsie River and its tributaries. Had I not sampled the tributaries,
I would have neglected a large portion of the juvenile population. In terms of total
watershed abundance, I found that almost 50% of the total juvenile steelhead
abundance occurred in the 4 tributaries sampled, yet these tributaries comprised only
11% of the total channel considered in this study. The tributaries are the best rearing
environment forjuvenile steelhead with the most desired thermal regime and lowest
loss rates. This evidence can now be used to prioritize management activities in the

watershed to further protect riparian habitat and decrease soil erosion, particularly in

143

 

the tributaries which, due to their small size, can be sensitive to small changes in
bedload.

In evaluating the thermal regime for water temperature in the watershed, I
developed several empirical models and two physical process models for evaluating
temporal and spatial pattems in thermal regime. In addition, a stream reach physical
process model was used to assess management alternatives for altering the summer
temperature regimes.

The empirical models predicted water temperature reasonably well with the exception
of very cold weather. It is likely that surface ice or snow as well as the physical
properties of water at very low temperatures caused this reduced predictive power.
However, these models can be used to determine water temperature in the Betsie River
unless major landscape features change within the watershed.

The two physical process models gave insight to the spatial and temporal
thennal dynamics on a daily time-step for summer in the stream reach model and on a
weekly time step for the annual stream network model. Removal of Grass Lake Dam
would result in decreased summer water temperatures from Green Lake to
Thompsonville, although in low-ﬂow, hot summers, the beneﬁts would be minimal.
Even with the removal of Grass Lake dam, water temperatures in this reach still could
rise above the optimum growth limits and at times exceed the preferred and zero-
growth limits for juvenile steelhead. Consideration should be given to the trade-offs
between the beneﬁt to the ﬁshery and losses in recreational opportunities if this dam
were removed.

The siphon from the hypolimnion of Green Lake would lower water
temperatures to the preferred range for juvenile steelhead in the reach from Grass

Lake to Thompsonville. However this option would require the construction of a dam to

144

 

control water ﬂow at the outlet of Green Lake and construction of a diversion. Again, a
cost beneﬁt analysis would be prudent if this management option were considered.
Other management options that might result in potential thermal beneﬁts include
additional shading in the reach above Thompsonville and the addition of habitat
diversity in several reaches that are basically homogenous mns. Both of the physical
process models could be used to guide future management options in the Betsie River
and were extremely useful tools in assessing stream thermal dynamics.

Future research in the Betsie River could include several topics.
Bioenergetically, the watershed proves to be taxing in some areas and entirely
beneﬁcial in others. A study of the bioenergetics from the watershed perspective would
provide biological insight to the conditions that these juvenile steelhead face. The
mobile sand bedload that resulted from the Thompsonville Dam collapse may be
suppressing invertebrate populations and this combined with temperatures above the
preferred for juvenile steelhead could partially be the mechanism behind the
watershed’s limited capacity to produce smolts. A bioenergetic investigation would
illuminate some of these potential forage and growth limitations. Additionally, research
into the rates and periodicities of movement conducted by this juvenile population,
coupled with a bioenergetics study, may lead to an understanding between the N
watershed thermal dynamics and ﬁsh behavior regarding thermal optimization.

In terms of the steelhead ﬁshery, neither the number of adult steelhead
returning to the river or the number of steelhead harvested from within the watershed is
presently unknown. This information is crucial to determining the appropriate levels of
stocking in combination with wild production. Wild production, while seemingly small
from 1993-1998, comprised a large portion (90%) of the ﬁshery in the late 1980's

(Seelbach and Whelen 1987) when harvest was estimated to be approximately 1,200 -

145

 

 

2,800 adult steelhead (Rakoczy and Rogers 1987, 1988, and 1990). Conservative
survival and return rates (7%) from the mean number of wild smolts leaving the
watershed in 1993-1996 (1,540) predict that approximately 108 wild adults will return to
the river annually in years 1995-1999. One reason for the low number of wild smolts
observed throughout the course of my study may be due to environmental extremes:
the winter of 1993-1994 was the coldest winter in 22 years, the drought conditions of

1995 had not occurred in 17 years, and a large ﬂood occurred in 1998. Altematively,

 

the low number of smolts could reﬂect the stream’s limited carrying capacity due to
changes in habitat that resulted from the collapse of Thompsonville Dam. If this is the
case, wild smolt abundance may increase as the channel stabilizes following this
systemic perturbation. Finally, because their growth rates are high, the Betsie River
steelhead may experience higher rates of survival than observed in other streams
which would yield a larger return to the ﬁshery than I have predicted.

Future management efforts of the Betsie River watershed should incorporate
actions that may moderate the effects of extreme temperatures. For winter, options for
reducing the severity of the stream temperatures may be limited. Increased shading
may result in less radiant heat loss and subsequently less frazil and anchor ice
formation. Maintenance of loose gravel and cobble substrates, either by introduction of
the substrate or reduction of the sand/silt bedload, could also provide additional habitat
during the winter.

For the extreme summer temperatures, one possibility for increasing summer
habitat could be assessed regarding the feasibility of protecting the cooler micro-
thermal areas in the stream above Thompsonville. Most certainly, protection of the
tributaries and areas of the watershed that introduce groundwater will provide

continued sustainment of the wild production. In particular, maintenance of the riparian

146

 

and reductions in sediment loads from road crossings or other activities could provide
large beneﬁts for the continued production of juvenile steelhead from these areas.
However, given the unique hydrology of the watershed in that groundwater
contributions are not signiﬁcant until below Thompsonville Dam, in low-ﬂow, hot years,
there may not be any actions that could moderate temperatures. Variability is inherent
in natural resource management and some water years will lead to better habitat and

greater juvenile steelhead production than others.

147

 

 

APPENDICES

 

148

 

APPENDIX A

 

149

 

APPENDIX A

Procedures for Establishing and Maintaining Staff Gages and Developing Rating
Curves in the Betsie River Watershed

Qaging Sites and Comments on Gage Quality

Green Lake: DNR established gage, good quality, good place to measure
Grass Lake: DNR gage attached to Grass Lake Dam, poor rating curves

Woirel Bridge: my gage, vandalized early in study, good ﬂow and gage
relationship prior to its loss

Liitle Betsie River. ruler gage along Highway 689, good gage and ﬂow
measures

Thompsonville: several different gage placements, channel changed frequently,
poor rating curve, good ﬂow measurements

Homestead Weir. DNR gage on ﬁshing platform above darn, good relationships
between ﬂows both upstream and downstream of the dam

9.0199539

Gage Reading and Protocols for Discharge Measurements

During May and June, the Homestead Weir gage was read daily. All other gages were
recorded on a weekly basis throughout the summer. Volunteers read the gages during
the winter. Discharge was measured several times throughout the course of the study.
I tried to capture low, median, and high stream ﬂows to develop the rating curves.

Ratin urves

Measured discharges and gage readings for each site were log transformed and I used
regression analysis to determine the relationship that best ﬁt the data. Extreme outliers
were considered individually and removed when it became apparent that either a gage
was misread or discharge measures were in error. This only occurred on 3 different
instances at different gages in different years.

150

Rating Eguations r2 value signiﬁcance level
Green Lake

Log (Q) = Log(GH) " 58137-2042 0.913 0.0019
Grass Lake Dam

Log(Q) = Log(GH)‘9.3299 - 8.1032 0.956 0.0005
Woirel Bridge

Log (Q) = Log (GH)"0.221 + 1.907 0.961 0.0020
Little Betsie River

Log(Q) = Log(GH)‘0.480 + 0.766 0.871 0.0040
Thompsonville

Log(Q) = Log(GH)"0.7005 + 1.2867 0.849 0.6490
Homestead Weir

Log(Q) = Log(GH)'0.2196 + 1.7547 0.972 0.0001

Where, Q = discharge, GH= gage height.

 

W

I used regression analysis to determine relationships between gaging stations for ﬁlling
in missing data. The following equations were used to ﬁll data when necessary.

Regression Eguation if P- value
Thompsonville (TV)

Log (TV-Q) = Log (HS-Q) + Log (LBR-Q) -0.6661 0.931 0.0003
Green Lake (GL)

Log (GL-Q) = Log (TV-Q)‘1.2145 - 0.7380 0.701 0.0004
Homestead Weir (HS)

Log(l-lS-Q)= Log(TV-Q)‘0.5626+1.1421 0.758 0.0001
Little Betsie River (LBR)

Log(LBR-Q) = Log(HS)‘5.0382 + Log (TV)°4.8067+2.920 0.971 0.0001
Dair Creek (DC)

Log (DC) = Log(LBR-Q)*0.4610+Log(HS)'1.0619-1.8828 0.974 0.0120

151

APPENDIX B

 

152

 

 

 

.13 8.. 5 8.8 8.. 3 n~..~ 8.. .~ 34. 8.6 4. 852m
~........~. .2. u. «3. £8 8 8.3 :8 .~ 3 3.: a .86 Ba
3.5.525 4.22...

9.8 .5 8. 8.3 3.” a... 8...... 8.2 .v n.-~. and a. 83. was...
8.8 no... an 913 3.. 3 8.3 88 8 . . . 3.:
xmwmo :26

...o 2..“ n. 2.3 3... o... 278 2.. .3. new Box .228
«new 8.. mm 8.8.... 9.... R. 2.....4. 38 a... an-.. 8.8 .m .82, Box .258
. . . . . a.» 3.. o 80?.

ESE £33 3::

a o c o 68m .28

m 3.8 8.... mm 2... 8.” . a.-. 2.. n. .62, 8.35:

Na. .3 o. 2...... 8.. no 84. 3.... v. I” R... a Sam 9.5....
.838 cu... So 819 SN. 34 2.0.8» 8... a." 8...: ..... .n 8 m. .s. 52.9:
3.... «ad. o .44. 8... a. . . v m; .3 a £35852.
. . m 3 3... c c . . . 33.. 83. 95.

o a... a... m . .83 Box 95.

. . . o away. onI

8.3. 8.x 8 ow... ......o m. . m 9058...... 9.232,
a a o a 28m. oaucm .2232,

5:5. Enema

.3 Sum oseﬁm ._.o 35m 2258 .. .0 Sum aseﬁm ._ .0 Sam 055.3 2%

$ma Panza“ 305005 3.3 222.35 twosooﬁ 3mm Rowena anon—08w 3mm Banana Duos—03w
no: "322:”

bump KNEE?»

a: KmEEDw

an: KNEEDW

 

dam .432 60:30.53 32¢
0.90m 05 59.0385 029.com 0:» coco .5. «0.22:. 8:03:90 «one uca .Eoto Eat—5% 60.30550 tan 305003 .0 moan-5mm .m 59.09?

153

Appendix B (Continued). Estimates of abundance, standard errors, and 95%
conﬁdence intervals for steelhead parr at each sampling site throughout the Betsie
River watershed, 1993-1996.

 

 

FALL 1994 FALL 1995
Steelhead Standard 95% Steelhead Standard 95%
Site Estimate Error C.l. Estimate Error C.l.
BETSIE RIVER
Woirels Bridge (East) 0 0 . .
Woirel's Bridge(West) 13 3.90 11-21 13 1.17 12-5
Haze Road . . . . . .
King Road Upper 0 . . 2 1.00 2-15
King Road Lower 0 . . 0 . .
Thompsonville 25 8.63 20-39 1 1 0.63 10-12
Highway M1 15 385 6.66 372-398 349 8.07 333-385
Psutka Road 5 0.00 . 40 3.43 37-47
Homestead Dam
River Road
UTI'LE BETSIE
RIVER
Bentley Road East 70 2.19 64-73 44 0.70 4344
Bentley Road West 105 5.09 97-115 101 0.98 99-106
Hwy669 . . . .
DAIR CREEK
M1 15 50 2.47 50-51 32 0.53 29-32
Landls Road 50 0.39 50-51 65 0.96 65-69
SMALL
TRIBUTARIES
Red Creek 5 . . 8 0.95 8-9
Steeves 8 0.40 8-9 7 0.40 7-9

 

154

 

 

BIBLIOGRAPHY

155

 

BIBLIOGRAPHY

Adams, S. M. and J. E. Breck. 1990. Bioenergetics. Pages 389-415 in C. B. Schreck
and P. B. Moyle, editors. Methods for ﬁsh biology. American Fisheries Society,
Bethesda, Maryland.

Alexander, G. R. and E. A. Hansen. 1986. Sand bed load in a brook trout stream.
North American Journal of Fisheries Management 6:9-23.

Alexander, D. R. and H. R. MacCiimmon. 1974. Production and movement of
juvenile rainbow trout (Salmo gairdnen) in a headwater of Bothwell’s Creek,
Georgian Bay, Canada. Journal of the Fisheries Research Board of Canada
31 :1 17-121.

Alexander, G. R. and J. R. Ryckman. 1986. Effect of an abnormal discharge of
sediment from the Lansing Club lmpoundment on the trout population in the Pigeon
River, Otsego and Cheboygan Counties, Michigan. Michigan Department of Natural
Resources, Fisheries Division, Technical Report 86-8.

Angenneier, PL and J. R. Karr. 1983. Fish communities along environmental
gradients in a system of tropical streams. Environmental Biology of Fishes
9(2):117-135.

Appleby, A. E. and J. M. Tipping. 1991. Use of electronic ﬁsh counters for coho and
chinook salmon, steelhead and cutthroat trout smolts. The Progressive Fish-
Culturist 53:195-198.

Bardach, J. T., J. H. Ryther and W. O. McLamey. 1972. Aquaculture. John Wiley 8.
sons, New York, NY. 868pp.

Bartholow, J. M. 1989. Stream temperature investigations: ﬁeld and analytic methods.
Instream Flow lnfonnation Paper No. 13. US. Fish and Wildlife Service Biological
Report 89 (17). 139 pp. .

Barthalow, J. M. 1991. A modeling assessment of the thermal regime for an urban
sport ﬁshery. Environmental Management 15(6):833-845.

Beach, M. H. 1978. The use of infra-red light and closed circuit TV to validate records
from automatic ﬁsh counters. Journal of Fish Biology 13:639-644.

Berggren, T. J. and M. J. Filardo. 1993. An analysis of variables inﬂuencing the
migration of juvenile salmonids in the Columbia River basin. North American
Journal of Fisheries Management 13:48-63.

Biette, R. M., D. P. Dodge, R. L. Hassinger, and T. M. Stauffer. 1981. Life history and
timing of migrations and spawning behavior of rainbow trout (Salmo gairdnen)
populations of the Great Lakes. Canadian Journal of Fisheries and Aquatic
Sciences 38:1759-1771.

156

 

 

 

Bilby, R. E. 1984. Characteristics and frequency of cool-water areas in a western
Washington stream. Journal of Freshwater Ecology 2:593-602.

Bjomn, T. C., S. C. Kirking and W. R. Meehan. 1991. Relation of cover alterations to
the summer standing crop of young salmonids in small southeast Alaska streams.
Transactions of the American Fisheries Society 120:562-570.

Bonham, M. 1975. As assessment of the Betsie River trout population since chemical
rehabilitation in 1965. Michigan Department of Natural Resources, Fisheries
Division, Lansing, Michigan. '

Bozek, M. A. and F. J. Rahel. 1991. Assessing habitat requirements of young
Colorado River cutthroat trout by use of macrohabitat and microhabitat analyses.
Transactions of the American Fisheries Society 120:571-581.

Brege, D. A., R. F. Absolon, and R. J. Graves. 1996. Seasonal and dial passage of
juvenile salmonids at John Day Dam on the Columbia River. North American
Journal of Fisheries Management 16:659-665.

Bulkley, R. V. 1967. Fecundity of steelhead trout, Salmo gairdnen', from Alsea River,
Oregon. Journal of the Fisheries Research Board of Canada 24(5):917-926.

Bullen, B. 1972. Preliminary habitat improvement report Thompsonville Dam
removal. Fisheries Division Internal Report Michigan Department of Natural
Resources. Roscommon, Michigan.

Bustard, D. R. and D. W. Narver. 1975. Aspects of the winter ecology of juvenile coho
salmon (Oncorhynchus kisufch) and steelhead trout (Salmo gairdnen). Journal of
the Fisheries Research Board of Canada 31:667-680.

Campbell, R. F. and J. H. Neuner. 1984. Seasonal and diurnal shifts in habitat utilized
by resident rainbow trout in Western Washington Cascade Mountain streams.
Pages 39-48 in F. W. Olson, R. G. White, and R. H. Hamre, edtiors. Proceedings
of the symposium on small hydropower and ﬁsheries. American Fisheries Society,
Bethesda.

Carbine, W. F. 1945. An investigation of Grass Lake, Benzie County, to determine
whether ﬁsh would beneﬁt from any elevation of the water level. Report No. 1025,
Fisheries Division, Michigan Department of Natural Resources, Institute for
Fisheries Research, Ann Arbor.

Cherry, D. S., K. L. Dickson, J. Cairns Jr., and J. R. Stauffer. 1977. Preferred, avoided
and lethal temperatures of ﬁsh during rising temperature conditions. Journal of the
Fisheries Research Board of Canada 34: 239-246.

Collins, N. C., S. G. Hinch, and k. A. Baia. 1991. Non-intrusive time-lapse video

monitoring of shallow aquatic environments. Canadian Technical Report of
Fisheries and Aquatic Sciences 1821: 35p.

157

Cormack, R. M. 1968. The statistics of capture-recapture methods. Oceanography
and Marine Biology Annual Reviews 62455-506.

Cousens, N.B.F., G. A. Thomas, C. G. Swann, and M. C. Healy. 1982. A review of
salmon escapement estimation techniques. Canadian Technical Report of
Fisheries and Aquatic Sciences No. 1108. 122pp.

Coutant, C. 1976. Thermal effects on ﬁsh ecology. Pages 1057-1063 in J. R. Pfafﬂin
and E. N. Ziegler, editors, Encyclopedia of environmental science and engineering.
Gordon and Breach Science Publishers. New York.

Coutant, C. 1977. Compilation of temperature preference data. Journal of the
Fisheries Research Board of Canada 34:739-745.

Crisp, D. T. 8. P. A. Carling. 1989. Observations on siting, dimensions and structure of
salmonid redds. Journal of Fish Biology 1989:119-134.

Crisp, D. T. and G. Howson. 1982. Effect of air temperature upon mean water
temperature in streams in the North Pennines and English Lake District.
Freshwater Biology 12:359-367.

Cunjak, R. A. 1988. Behaviour and microhabitat of young Atlantic salmon (Salmo
salar) during winter. Canadian Journal of Fisheries and Aquatic Sciences 4522156-
2160.

Cunjak, R. A. 1988. Physiological consequences of overwintering in streams: the cost
of acclimitization? Canadian Journal of Fisheries and Aquatic Sciences 45:443-
452.

Cunjak, R. A. and G. Power. 1986. Winter habitat utilization by stream resident brook
trout (Salve/inus font/nails) and brown trout (Salmo trutta). Canadian Journal of
Fisheries and Aquatic Sciences 43:1970-1981.

Davis, S. K., J. L. Congleton, and R. W. Tyler. 1980. Modiﬁed fyke net for the capture
and retention of salmon smolts in large rivers. Progressive Fish-Culturist
42(4):235-237.

Dempson, J. B. and D. E. Stansbury. 1991. Using partial counting fences and a two-

sample stratiﬁed design for mark-recapture estimation of an Atlantic salmon smolt
population. North American Journal of Fisheries Management 11: 27-37.

Dexter, J. L., Jr. 1993a. Sand Creek. Michigan Department of Natural Resources,
Fisheries Division, Status of the Fishery Resource Report 93-2.

Dexter, J. L. Jr. 1993b. Silver Creek. Michigan Department of Natural Resources,
Fisheries Division, Status of the Fishery Resource Report 93-6.

158

 

 

Dexter, J. L., Jr., N. D. Ledet. 1994. Estimates of ﬁsh passage at St. Joseph River
Dams in fall 1992 using time-lapse video recording. Michigan Department of
Natural Resources, Fisheries Division, Fisheries Technical Report 94-5.

DuBois, R. B., J. E. Miller, and S. D. Plaster. 1991. An inclined-screen smolt trap with
adjustable screen for highly variable ﬂows. North American Journal of Fisheries
Management 1 1 :155-159.

Elliott, J. M., M. A. Hurley, and R. J. Fryer. 1995. A new, improved growth model for
brown trout, Salmo frutta. Functional Ecology 9:290-298.

Everest, F. H. and D. W. Chapman. 1972. Habitat selection and spatial interactions of
juvenile chinook salmon and steelhead trout in two Idaho streams. Journal of the
Fisheries Research Board of Canada 19:91-100.

Everest, F. H. and J. R. Sedell. 1983. Evaluation of ﬁsheries enhancement projects on
Fish Creek and Wash Creek, 1982 and 1983. Paciﬁc Northwest Forest and Range
Experiment Station. Corvallis, Oregon. Annual Report, 1983.

Giorgi, A. E. and C. W. Sims. 1987. Estimating the daily passage of juvenile
salmonids at McNary Dam on the Columbia River. North American Journal of
Fisheries Management 17:268-282.

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-Columbia River basin. North
American Journal of Fisheries Management 17:268-282.

Gooding, M. P. 1995. A watershed classiﬁcation scheme for lower Michigan. A
Practicum Paper. University of Michigan. Ann Arbor. 36pp.

Gordon, D. J. and H. R. MacCrimmon. 1982. Juvenile salmonid production in a Lake
Erie nursery stream. Journal of Fish Biology 21:455-473.

Gordon, N. D., T. A. McMahon, and B. L. Finlayson. 1992. Stream Hydology: An
introduction for ecologists. John Wiley and Sons, New York. 526p.

Grifﬁth, J. S. 1993. Goldwater streams. Pages 405-426 in C. C. Kohler and W. A.
Hubert, editors. Inland ﬁsheries management in North America. American
Fisheries Society, Bethesda, Maryland.

Hartman, G. F. 1963. Observations on behavior of juvenile brown trout in a stream
aquarium during winter and spring. Journal of the Fisheries Research Board of
Canada 20:769-787.

Hartman, G. F. 1965. The role and behavior in the ecology and interaction of

underyearling coho salmon (Oncorhynchus kisutch) and steelhead trout (Salmo
gairdneri). Journal of the Fisheries Research Board of Canada 22 (4): 1035-1081.

159

 

 

Hatch, D. R., M. Schwartzberg, and P. R. Mundy. 1994. Estimation of Paciﬁc salmon
escapement with a time-lapse video recording technique. North American Journal
of Fisheries Management 14:626-635.

Heam, W. E. and B. E. Kynard. 1986. Habitat utilization and behavioral interaction of
juvenile Atlantic salmon (Salmo salar) and rainbow trout (S. gairdnen) in tributaries
of the White River of Vermont. Canadian Journal of Fisheries and Aquatic
Sciences 43:1988-1998.

Heifetz, J., M. L Murphy, and K. V. Koski. 1986. Effects of logging on winter habitat of
juvenile salmonids in Alaskan streams. North American Journal of Fisheries
Management 6:52-58.

Hoar, W. S. 1988. The physiology of smolting salmonids. Pages 275-342 in W. S.
Hoar and J. R. Randall, editors. Fish Physiology Volume XlB. Academic Press, 1
New York. .

Hendrickson, G. E. and C. J. Doonan. 1972. Hydrology and recreation on the cold-
water rivers of Michigan’s southern peninsula. United States Geological Survey,
Water Information Series Report 3. 82p.

Hokanson, K. E. F ., C. F. Kleiner, and T. W. Thorslund. 1977. Effects of constant
temperatures and diel temperature ﬂuctuations on speciﬁc growth and mortality
rates and yield of rainbow trout, Salmo gairdneri. Journal of the Fisheries Research
Board of Canada 34: 639-648.

Houston, A. H. 1982. Thermal effects upon ﬁshes. Report NRCC No. 18566.
National Research Council of Canada. Associate Committee on Scientiﬁc Criteria
for Environmental Quality. 200p.

Hvidsten, N. A. and B. O. Johnsen. 1993. Increased recapture rate of adult Atlantic
salmon released as smolts into large shoals of wild smolts in the River Orkla,
Non/ray. North American Journal of Fisheries Management 13:272-276.

Irvine, J. R., B. R. Ward, P. A. Teti, and N. B. F. Cousens. 1991. Evaluation of a
method to count and measure live salmonids in the ﬁeld with a video camera and
computer. North American Journal of Fisheries Management 11:20-26.

Jakes, P. J. 1982. Timber resource of Michigan’s northern lower peninsula, 1980.
Resource Bulletin NC-62, St. Paul, MN; United States Department of Agriculture,
Forest Service, North Central Forest Experiment Station. 120p.

Jearld, Ambrose, Jr. 1983. Aging Techniques. Pages 301-324 in L. A. Nielsen and D.
L. Johnson, editors. Fisheries Techniques. American Fisheries Society. Bethesda,
Maryland.

Kitchell, J. F., J. F. Koonce, R. V. O'Neill, H. H. Shugart, Jr., J. J. Magnuson, and R. S.
Booth. 1974. Model of ﬁsh biomass dynamics. Transactions of the American
Fisheries Society 103: 786-798.

 

160

Kocik, J. F. 1992. Evaluation of juvenile brown trout and steelhead competition in
Great Lakes tributaries. Ph.D. Dissertation, Michigan State University. East
Lansing. 118p.

Latte, W. C. 1974. A history of the introduction of ﬁshes into Michigan. Pages 83-96
in Michigan Fisheries Centennial Report, 1873-1973. Michigan Department of
Natural Resources. Fisheries Division. Lansing, MI.

Letherberry, E. C. 1994. Forest Statistics for Michigan's Northern Lower Peninsula
Unit, 1993. Resource Bulletin NC-157, St. Paul, MN: United State Department of
Agriculture, Forest Service, North Central Forest Experiment Station. 48p.

Li, H. W., G. A. Lamberti, T. N. Pearsons, C. K. Tait, J. L. Li, and J. C. Buckhouse.
1994. Cumulative effects of riparian disturbances along high desert trout streams
of the John Day Basin, Oregon. Transactions of the American Fisheries Society
123: 627-640.

Lister, D. B., R. A. L. Harvey, and C. E. Walker. 1969. A modiﬁed wolf trap for
downstream migrant young ﬁsh enumeration. Canadian Fish Culturist 40:57-60.

Macdonald, P. D. M. and H. D. Smith. 1980. Mark-recapture estimation of salmon
smolt runs. Biometrics 36:401-417.

Maciolek, J. A. and P. R. Needham. 1951. Ecological effects of winter conditions on
trout and trout foods in Convict Creek, California, 1951. Transactions of the
American Fisheries Society 81:202-217.

Meisner, J. D. 1990. Potential loss of thermal habitat for brook trout, due to climatic
warming, in two southern Ontario streams. Transactions of the American Fisheries
Society 119: 282-291.

Meyer, K. A. and J. S. Grifﬁth. 1997. Effects of cobble-boulder substrate conﬁguration
on winter residency of juvenile rainbow trout. North American Journal of Fisheries
Management 17:77-84.

Michigan Department of Natural Resources. Fisheries Division. Stream Surveys:
Betsie River, 1990. Honor, Michigan.

Moser, M. L., A. F. Olson, and T. P. Quinn. 1991. Riverine and estuarine migratory
behavior of coho salmon (Oncorhynchus kisutch) smolts. Canadian Journal of
Fisheries and Aquatic Sciences 48:1670-1678.

Meyer, K. A. and J. S. Grifﬁth. 1997. Effects of cobble-boulder substrate conﬁguration
on winter residency of juvenile rainbow trout. North American Journal of Fisheries
Management 17:77-84.

National Climatic Data Center. Periodic. Local climatological data. Asheville, NC.

161

 

 

Nielsen, J. L., T. E. Lisle, and V. Ozaki. 1994. Thermally stratiﬁed pools and their use
by steelhead in Northern California streams. Transactions of the American
Fisheries Society 123:613-626.

Nuhfer, A. J., R. D. Clark, Jr., and G. R. Alexander. 1994. Recruitment of brown trout
in the South Branch of the Au Sable River, Michigan in relation to stream ﬂow and
winter severity. Michigan Dept Natural Resources, Fisheries Division, Research
Report 2006. 14pp.

Pajak, P. 1992. Developing integrated ﬁsheries objectives for land and water resource
management The Milwaukee River experience. American Fisheries Society
Symposum 13:175-196.

Peck, J. W. 1992. The sport ﬁshery and contribution of hatchery trout and salmon in
Lake Superior and tributaries at Marquette, Michigan 1984-1987. Michigan
Department of Natural Resources, Fisheries Division, Fisheries Research Report
1975.

Peterson, J. T. and C. F. Rabeni. 1996. Natural thermal refugia for temperate
warmwater stream ﬁshes. North American Journal of Fisheries Management
16:738-746.

Peven, C. M., R. R. Whitney, K. R. Williams. 1994. Age and length of steelhead
smolts from the mid-Columbia River basin, Washington. North American Journal
Fisheries Management 14: 77-86.

Peven, C. M. and S. G. Hays. 1989. Proportions of hatchery- and naturally produced
steelhead smolts migrating past Rock Island Dam, Columbia River, Washington.
North American Journal Fisheries Management 9:53-59.

Pfeifer, R. E. and J. S. Spencer, Jr. 1966. The Growing Timer Resource of Michigan:
Unit 3- Northem Lower Peninsula. Michigan Department of Natural Resources,
East Lansing. 112p.

Pollock, K. H. 1981. Capture-recapture models: A review of current methods,
assumptions, and experimental design. Studies in Avian Biology 6:426-435.

Rakoczy, G. P. and R. D. Rogers. 1987. Sportﬁshing catch and effort from the
Michigan waters of Lakes Michigan, Huron, and Erie, and their important tributary
streams, April 1, 1986-March 31, 1987 (Appendices). Michigan Department of
Natural Resources, Fisheries Division, Fisheries Technical Report No. 87-6b.

Rakoczy, G. P. and R. D. Rogers. 1988. Sportﬁshing catch and effort from the
Michigan waters of Lakes Michigan, Huron, and Erie, and their important tributary
streams, April 1, 1987-March 31, 1988 (Appendices). Michigan Department of
Natural Resources, Fisheries Division, Fisheries Technical Report No. 88-9b.

162

 

Rakoczy, G. P. and R. D. Rogers. 1990. Sportﬁshing catch and effort from the
Michigan waters of Lakes Michigan, Huron, and Erie, and their important tributary
streams, April 1, 1988-March 31, 1989 (Appendices). Michigan Department of
Natural Resources, Fisheries Division, Fisheries Technical Report No. 90-2b.

Rand, P. S., D. J. Stewart, P. W. Seelbach, M. L. Jones, L. R. Wedge. 1993. Modeling
steelhead population energetics in Lakes Michigan and Ontario. Transactions of
the American Fisheries Society 122: 977-1001.

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.

Raymond, H. L. 1988. Effects of hydroelectric development and ﬁsheries
enhancement on spring and summer chinook salmon and steelhead in the
Columbia River Basin. North American Journal of Fisheries Management 8:1-24.

Reddin, D. G., M. F. O'Connell, and D. A. Dunkley. 1992. Assessment of an
automated ﬁsh counter in a Canadian river. Aquaculture and Fisheries
Management 23: 113-121.

Redding, J. M., C. B. Schreck, and F. H. Everest 1987. Physiological effects on coho
salmon and steelhead of exposure to suspended solids. Transactions of the
American Fisheries Society 116:737-744.

Ricker, W. E. 1975. Computation and interpretation of biological statistics of ﬁsh
populations. Department of the Environment and Fisheries Marine Service, Bulletin
191. 382pp.

Roff, D. A. 1973. On the accuracy of some mark-recapture estimators. Oecologia
12:15-34.

Roper, B. B., D. L. Scamecchia, and T. J. La Marr 1994. Summer distribution of and
habitat use by chinook salmon and steelhead within a major basin of the South
Umpqua River, Oregon. Transactions of the American Fisheries Society 1232298-
308

SAS Institute Inc., SAS/STAT® User's Guide, Version 6, Fourth Edition, Volume 2,
Cary, North Carolina. 846 pp.

Schwarz, C. J. and J. B. Dempson. 1994. Mark-recapture estimation of a salmon
smolt population. Biometrics 50:98-108.

Scott and Crossman. 1973. Freshwater ﬁshes of Canada. Fisheries Research Board
of Canada. Bulletin 184. 966p.

Seelbach, P. W. 1987. Effect of winter severity on steelhead smolt yield in Michigan:

an example of the importance of environmental factors in determining smolt yield.
American Fisheries Society Symposium 1: 441-450.

163

 

Seelbach, P. W. 1989. Characteristics of adult steelhead populations, including
retums of hatchery yearlings, in the St. Joseph and Grand Rivers, Michigan 1979-
85. Michigan Department of Natural Resources, Fisheries Division, Fisheries
Technical Report 893.

Seelbach, P. W. 1993. Population biology of steelhead in a stable-ﬂow, low gradient
tributary of Lake Michigan. Transactions of the American Fisheries Society 122:
179-198.

Seelbach, P. W., J. L. Dexter, Jr., and N. D. Ledet 1994. Performance of steelhead
smolts stocked in southern Michigan warmwater rivers. Michigan Department of
Natural Resources, Fisheries Division, Fisheries Research Report 2003.

Seelbach, P. W. and B. R. Miller. 1993. Dynamics in Lake Superior of hatchery and
wild steelhead emigrating from the Huron River, Michigan. Michigan Department of
Natural Resources, Fisheries Division, Fisheries Research Report 1993.

Seelbach, P. W. , G. R. Alexander, R. N. Lockwood. 1984. An inclined-screen trap for
salmonids in large rivers. Michigan Department of Natural Resources, Fisheries
Division, Technical Report 84-1.

Seelbach, P. W. and G. W. Whelan. 1988. Identiﬁcation and contribution of wild and
hatchery steelhead stocks in Lake Michigan tributaries. Transactions of the
American Fisheries Society 117: 444-451.

Smith, K. and M. E. Lavis. 1975. Environmental inﬂuences on the temperature of a
small upland stream. Oikos 26:228-236.

Stauffer, T. M. 1968. Migration, age, and growth of rainbow trout parr in Black River,
Michigan. Master's Thesis. Michigan State University, East Lansing.

Stoneman, C. L. and M. L. Jones. 1996. A simple method to classify stream thermal
stability with single observations of daily maximum water and air temperatures.
North American Journal of Fisheries Management 16:728-737.

Swales, S., R. B. Lauzier, and C. D. Levings. 1986. \Mnter habitat preferences of
juvenile salmonids in two interior rivers in British Columbia. Canadian Journal of
Zoology 64: 1506-1514.

Theurer, F. D., K. A. Voos, and C. G. Prewitt. 1982. Application of lFG’s instream
water temperature model in the upper Colorado River. lntemational Symposium on
Hydrometeorology. American Water Resources Association.

Theurer, F. D., K. A. Voos, and W. J. Miller. 1984. Instream water temperature model.
Instream Flow lnforrnation Paper 16. US. Fish and Wildlife Service FWS/OBS-
84115.

Van Deventer, J.S. and WS Platts. 1983. Sampling and estimating ﬁsh populations

from streams. Transactions of the North American Wildlife and Natural Resources
Conference 48:349-354.

164

Van Deventer, J.S. and W.S. Platts. 1985. A computer software system for entering,
managing, and analyzing ﬁsh capture data from streams. Research Note INT-352.
Ogden, UT: US. Dept of Agriculture, Forest Service. lntem'iountain Forest and
Range Exp. Sta.

Van Deventer, J. S. and W. S. Platts. 1986. Microﬁsh 3.0: Microﬁsh Interactive
Program. Forestry Sciences Laboratory, Boise, Idaho.

Wagner, H. H., R. L. Wallace, and H. J. Campbell. 1963. The seaward migration and
return of hatchery-reared steelhead trout, Salmo gairdneri Richardson, in the Alsea
River Oregon. Transactions of the American Fisheries Society 92:202-210.

Wagner, H. H. 1974. Photoperiod and temperature regulation of smolting steelhead
trout (Salmo gairdneri). Canadian Journal of Zoology 52:219-234.

Waite, I. R. and R. A. Bamhart. 1992. Habitat criteria for rearing steelhead: A
comparison of site-speciﬁc criteria and standard curves for use in the Instream Flow
Incremental Methodology. North American Journal of Fisheries Management 12:
40-46.

Ward, B. R., P. A. Slaney, A. R. Facchin, and R. W. Land. 1989. Size-biased survival
in steelhead trout (oncorhynchus mykiss): Back-calculated lengths from adults’
scales compared to migrating smolts at the Keogh River, British Columbia.
Canadian Journal of Fisheries and Aquatic Sciences 46:1853-1858.

Waters, T. F., M. T. Doherty, and C. C. Krueger. 1990. Annual production and
production:biomass ratios for three species of stream trout in Lake Superior
tributaries. Transactions of the American Fisheries Society 119:470-474.

Wedemeyer, G. A., R. L. Saunders, and W. C. Clarke. 1980. Environmental factors
affecting smoltiﬁcation and early marine survival of anadromous salmonids. US.
National Marine Fisheries Service, Marine Fisheries Review 42(6):1-14.

Wicklund, R. G. and B. C. Dean. 1958. Betsie River Watershed: Survey and plans
report Michigan Department of Conservation, Fish Division, Lake and Stream
Improvement Section. Lansing.

Wismer, D. A. and A. E. Christie. 1987. Temperature relationships of Great Lakes
ﬁshes: A data compilation. Great Lakes Fisheries Commission Special Publication
No. 87-3. Ann Arbor. 165 p.

Za009. W. S., B. L. Adams, and L. R. McClain. 1972. Steelhead migration: Potential

temperature effects as indicated by gill adenosine triphosphatase activities.
Science 176: 415-416.

165

 

Zaugg, W. S. and H. H. Wagner. 1973. Gill ATPase activity related to parr-smolt
transformation and migration in steelhead trout (Salmo gairdneri): Inﬂuence of
photoperiod and temperature. Comparative Biochemistry and Physiology 45B:955-
965. '

Zaugg, W. S., 1981. Advanced photoperiod and water temperature effects on gill Na’-
K’ adenosine triphosphatase activity and migration of juvenile steelhead (Salmo
gairdneri). Canadian Journal of Fisheries and Aquatic Science 38:758-764.

Zedonis, P. 1997. A water temperature model of the Trinity River. US. Department of
the Interior, Coastal California Fish and Wildlife Ofﬁce, Arcata, California. 98pp.

Zippen, C. 1958. The removal method of population estimation. Joumal of Wildlife
Management 22:82-90.

166

 

 

MICHIGAN smrs UNIV. LIBRARIES
II|llHHII"HIIHIIWWII"lllllllllllIIIIIIIHIIIUIIIHII
31293017128947