 

.. 2M .23....

IQALCC . 1

313.6?
.

umummmmu

93w v;

32. 5
n3 chubby“.
1 5.... :‘i
£373....cttl)
51.23:. t
.~ .1 .109.
u .. It‘ll...” : 1

my?"

«mum!

i

 

gvil~

aﬂwuﬁa
. .).A"

m
l
1
u
t
.V
a.
~
‘
~
..
x
u
u
9
u
s.
I
u

'.I§II¢~‘-v'1-‘|10

‘

«nun-nu

'.Y

(n. :q'Tg-VIQ-(tl-Ilmlllﬂ‘,

to:

5.0 "no”

urn-g-

 

 

a. 1.: A
. iii

 

 

LIBRARY
“$818 Michigan State
1003 University

 

 

This is to certify that the
dissertation entitled

SEASONAL BATHYTHERMAL HABITAT USE BY
LAKE TROUT AND LAKE WHITEFISH IN LAKE
HURON AS MEASURED WITH IMPLANTED
ARCHIVAL TAGS

presented by

Roger Allen Bergstedt

has been accepted towards fulﬁllment
of the requirements for the

Doctoral degree in Fisheries and Vlﬁldlife

 

 

Major Professor’sdwore
ﬁll/ﬂ

Date

MSU is an aﬁirmative-action, equal-opportunity employer

 

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

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K:/Prolecc&Pres/ClRC/DateDueindd

 

SEASONAL BATHYTHERMAL HABITAT USE BY LAKE
TROUT AND LAKE WHITEFISH IN LAKE HURON AS
MEASURED WITH IMPLANTED ARCHIVAL TAGS

By

Roger Allen Bergstedt

A DISSERTATION
Submitted to
Michigan State University
In partial fulﬁllment of the requirements
For the degree of
DOCTOR OF PHILOSOPHY
Fisheries and Wildlife

2008

ABSTRACT
SEASONAL BATHYTHERMAL HABITAT USE BY LAKE

TROUT AND LAKE WHITEFISH IN LAKE HURON AS MEASURED
WITH IMPLANTED ARCHIVAL TAGS

By

Roger Allen Bergstedt

Temperatures occupied by 33 lake trout (Salvelinus namaycush) in Lake Huron
were recorded with surgically implanted archival tags in 1999 and 2000. Objectives
were to provide temperature data to reﬁne bioenergetics models of sea lamprey
(Petromyzon marinus) predation on lake trout, and to compare temperatures occupied
by strains of lake trout. During seasonal periods of mixing, variability was low and
followed surface temperature. During summer stratiﬁcation, temperatures occupied
varied substantially among individual ﬁsh and strains. Between June and mid
August, upper Great Lakes origin (GLO) lake trout occupied warmer water than
Finger Lakes, New York origin (FLO) lake trout, but both occupied temperatures
below laboratory preferred temperatures. In October, both origins occupied warm or
warmer water than that in summer, partially explaining the higher lethality of sea

lamprey attacks during October.

The objectives of a second study with archival tags recording depth and temperature
in 2002-2005 were to compare the observations with the 1999-2000 data, to examine
whether GLO lake trout occupy both higher temperatures and lesser depths than FLO
lake trout, and to examine changes in the temperatures occupied in light of concurrent

changes in the prey community. Temperatures occupied by GLOO lake trout were

signiﬁcantly higher than by FLO lake trout, supporting the conclusion of the ﬁrst
study. The GLO lake trout also occupied Signiﬁcantly shallower depths than FLO
lake trout. Compared to 1999-2000, both GLO and FLO lake trout occupied
signiﬁcantly lower temperatures. Concurrently, pelagic alewives (Alosa
pseudoharengus) disappeared, demersal round gobies (Neogobius melanostomus)
invaded Lake Huron, and prey biomass decreased by 90%. Switching to demersal
prey and a retreat to temperatures below the optimum for growth in response to food

scarcity are potential explanations of the movement toward lower temperatures.

Depth observations from tagged lake Whiteﬁsh (Coregonus clupeaformis),
recorded in 2003 -2005, were used to examine whether seasonal depth differences
could be exploited to reduce bycatch of lake trout in commercial gill net ﬁsheries for
lake Whiteﬁsh. Both GLO and FLO lake trout occupied signiﬁcantly deeper and
colder waters seasonally, during both daylight and dark, than lake Whiteﬁsh.
Separation was greater between FLO lake trout and lake Whiteﬁsh. Fishing to
seasonal depth limits of 25 to 35 m could target half or more of the lake Whiteﬁsh, yet
relatively low percentages of lake trout. Percentages of lake trout targeted were
lowest in late July (GLO = 8% and FLO = 11%) and early August (GLO = 4% and
FLO = 7%). Feasibility depends on lake Whiteﬁsh being available to gill nets. Modal
depths of commercial gill net effort exceeded the modal depth in the lake Whiteﬁsh
archival tag data and suggested some lake Whiteﬁsh were pelagic enough to not be

vulnerable to gill nets. The possibility of using suspended gill nets was suggested.

DEDICATION

This work is dedicated to my mother and father, Pearl and Roy Bergstedt, and to my

wife Jocelyn, who patiently supported me as I resumed my education in pursuit of a

doctoral degree.

iv

ACKNOWLEDGMENTS

I thank my graduate committee of Dr. William Taylor, Dr. Charles Krueger, Dr.

Michael Jones, Dr. Mohamed Faisal, and Dr. R. Scott McKinley for their guidance.

Funding for this research was provided by the US. Geological Survey, the US. Fish
and Wildlife Service through the Great Lakes Protection Fund, and the Great Lakes

Fishery Commission.

I thank Mrs. Karen Slaght, Mr. David Keffer, Mr. Erick Larson, and Mrs. Mary K.
Jones for their assistance in all phases of this research. Mr. Mark Ebener provided
valuable assistance in obtaining the cooperation of tribal ﬁshers. Dr. James Bence
provided valuable assistance and advice on data analysis. I particularly thank Dr. Charles
Krueger and Dr. Chris Goddard for their support in both my research and in my pursuit of

a doctoral degree.

PREFACE

Temperature is an important environmental variable affecting food consumption,
growth, and the depth and spatial distribution of the genetic strains of lake trout
(Salvelinus namaycush) stocked in Great Lakes. Studies of Salvelinus species have
shown that temperature preference, depth preference, and thermal resistance are heritable
characteristics of Salvelinus species and would likely vary among genetic strains.
Temperature preference can be inﬂuenced by selective breeding of lake trout and the
generation of hybrids between lake trout and brook trout (Goddard and Tait 1976). Tait
(1970) Showed that swim bladder gas retention of F2 hybrids of lake and brook trout (S.
fontinalis) was intermediate to either parent species. Gas retention was further
demonstrated to vary between lake trout from two lakes, that interpopulation crosses
were intermediate in gas retention, and that gas retention correlated with depth
distribution of the parent populations (Ihssen and Tait 1974). The ability to tolerate
thermal stress has also been identiﬁed as heritable (Ihssen 1973). The combined
phenotypic expression of these multi-genic traits during summer periods of thermal
stratiﬁcation affects the temperature and depth distributions of lake trout. Clear genetic
differences exist among the strains of lake trout stocked (Krueger et a1. 1989), which
suggest that strains may differ in their distributions across temperature and depth

gradients.

Hutchinson (1957) was ﬁrst to deﬁne “niche” in a strong quantitative sense and

suggested that laboratory data could be used to describe the fundamental, or non-

vi

interactive, niche of a species within a temperature gradient. Numerous laboratory
studies have contributed to our understanding of the thermal behavior of lake trout,
including preferred temperatures (e. g., McCauley, and Tait. 1970; Peterson et al. 1979;
Edsall and Cleland 2000), optimum temperatures for growth (e. g., Elliot and Hurley
1999; Edsall and Cleland 2000), and temperature tolerance or lethal limits (e. g.,
lhssen1973; Grande and Andersen 1991 ). Results of these studies tend to be consistent
(Jobling 1981) and imply an innate component to the responses of lake trout to changes
in temperature as suggested by Hutchinson (1957). However, description of a
fundamental niche for temperature considers only one of a number of environmental
variables affecting the distribution of ﬁsh and species are often found at temperatures
outside the range indicated in laboratory studies (Magnuson et al. 1979). Describing the
realized thermal niche based on assembled ﬁeld observations, which incorporate the other
variables affecting habitat will typically be of greater interest to ﬁshery ecologists
(Magnuson et al. 1979). As examples, Eaton et al. (1995) and Huff et al. (2005) used this
approach to describe temperature tolerance and Huff et al. (2005) also described the

center and width of the realized niche for some vertebrates.

Although the results of my research will be useful in describing temperatures
occupied as one aspect of the realized niche of lake trout, my interest in describing the
seasonal temperatures derived from experiences conducting ﬁsh stock assessment on
Lake Ontario, 1978-1985, and my perception of the potential importance for management
to have an accurate description of the seasonal temperatures occupied by lake trout. This

interest was originally triggered by observations of losses of lake trout to sea lamprey

vii

predation as documented by the collection of dead lake trout in bottom trawls (Bergstedt
and Schneider 1988). Although intensive trawl surveys were conducted from late-April
through October, dead lake trout were recovered only in October and early November,
when the partial autumnal circulation placed lake trout at the highest temperatures of the
year. This suggested that temperature could have a role through elevation of sea lamprey
metabolic rate, and hence rate of blood consumption. Further analyses of collections of
dead lake trout during 1982-1992 (Schneider et al. 1996) suggested that sea lamprey

attacks on the Seneca strain of lake trout showed lower lethality than on other strains.

The seasonal depths and temperatures occupied by the various genetic strains of lake
trout being stocked into the Great Lakes for rehabilitation and the commercially
important lake Whiteﬁsh (Coregonus clupeaformis) are known only as generalizations.
Our knowledge is not based on continuous observations, but is deduced from captures in
different ﬁshing gears. These capture data only provide point estimates of abundance
within the depths and temperatures sampled and suffer from the sampling bias of each
gear. As such, existing data do not accurately reveal seasonal and die] rhythms, and are
limited to the times of year when surveys are conducted in the Great Lakes. The ability
to compare the seasonal distributions of lake trout and lake Whiteﬁsh is of particular
interest because lake Whiteﬁsh are the primary commercial species in Lake Huron and
loss of lake trout as bycatch affects achievement of lake trout rehabilitation in Lake
Huron. Maintaining total annual mortality below 40% is critical to lake trout restoration

and mortality due to bycatch is one of the components of mortality.

viii

Development of archival, or data-storage, tags provided a new approach to collecting
data on seasonal bathythermal habitat use by ﬁsh in the Great Lakes. Archival tags are
considered a form of telemetry, but instead of transmitting data, they store it on ﬂash
memory. These tags can be either attached externally or implanted in the body cavity of
ﬁsh. When a tagged ﬁsh is released, they record data at predetermined intervals. This
was an important new advance in technology because investigators no longer were
required to physically follow a ﬁsh to receive data transmissions. However, to retrieve
the data ﬁsh must be recovered and the data from the tag downloaded. If a species is
exploited at a reasonable rate, as are lake trout and lake Whiteﬁsh, and if they are

externally tagged with an offer of a substantial reward, recovery is not a problem.

Advancing the knowledge of bathythermal habitat selection by strains of lake trout
(Chapters One and Two) provides an improved understanding of the mechanisms behind
differences in attack ﬁ'equency by sea lampreys and the lethality of sea lamprey attacks, a
critical component of lake trout mortality. Great Lakes origin lake trout stocked in lakes
Huron and Ontario have consistently higher sea lamprey attack rates than lake trout of
Finger Lakes origin (Eshenroder et al. 1995; Schneider et al. 1996). Differences in attack
rate and survival from individual attacks may be related to differences in seasonal
bathythermal distributions among strains (Swink 1993; Schneider et a1. 1996). If
bathythermal distributions varied, then differential overlap with the bathythermal
distribution of the sea lamprey could account for differing attack rates. Occupation of
lower or higher temperatures could, through the effects of temperature on metabolic rate

and blood consumption, also affect sea lamprey attack lethality.

ix

The importance of lake trout restoration in the Great Lakes makes the data and
analyses provided in this dissertation timely. For example, the recent signing of the 2000
Consent Decree in US. District Court, which governs ﬁsheries management in the 1836
treaty—ceded waters of the upper Great Lakes for the next two decades, clearly
demonstrated the strong commitment of the federal, state, and tribal ﬁshery management
agencies to rehabilitate lake trout. This commitment and, indeed, the Agreement itself

require reductions in the mortality of lake trout caused by commercial and sport ﬁsheries.

Lake Whiteﬁsh are the principal target of commercial species in the Great Lakes, and
lake trout are the principal bycatch in that ﬁshery (Brown et a1. 1999). Bycatch was
deﬁned by Crowder and Murawski (1988) as both the retained and discarded non-target
organisms, including unobserved mortalities of ﬁsh that are released. Unobserved
mortality includes delayed deaths caused by injury, stress, disease, or predation (Chopin
et al. 1996). Lake trout gilled in the leads and pots of trap nets account for biologically
signiﬁcant bycatch mortality (Johnson et al. 2004), although not as high per lake
Whiteﬁsh harvested as gill nets. Johnson et al. (2004) also addressed the issue of post-
release survival of lake trout released unharmed ﬁom the pots of trap nets. Using data
from Chapters One and Two, they estimated that over 98% survived after release. In
contrast, survival of live lake trout taken from gill nets was estimated at 72% (Gallinat et
al. 1997). Recommendations to reduce bycatch losses of lake trout included limiting lake
Whiteﬁsh harvest to trap nets and reducing effort in summer when the ratio of lake trout

to lake Whiteﬁsh caught was highest (Johnson et al. 2004). Die] and seasonal

bathythermal habitat-use proﬁles for lake trout and lake Whiteﬁsh provided in this
dissertation increase our understanding of lake trout losses to sea lamprey predation and
can contribute to consideration of options to reduce lake trout losses caused by

commercial ﬁsheries bycatch (Chapter Three).

In Chapter One, archival (or data-recording) tags implanted in lake trout that recorded
the temperatures occupied during 1999 and 2000 and showed that different strains of lake
trout inhabited waters of different temperatures. However, the archival tags used in that
study only recorded temperature and could not reveal whether strain differences were due
to differing depths or to occupying similar depths in different geographic areas where the
depth of the metalimnion differed (e. g., nearshore or offshore). Therefore, concurrent
measures of depth and temperature were needed for the commonly stocked lake trout

strains and lake Whiteﬁsh to fully understand their interaction.

Subsequent to the ﬁrst study (Chapter One), substantial progress occurred in
miniaturization of electronics that made it possible to implant a smaller archival tag that
recorded both depth and temperature. In Chapter Two, these newer tags were implanted .
in lake trout in 2002 and 2003 and used to assess the overlap of both temperature and
depth distributions among strains of lake trout and to compare with results from 1999-
2000. A similar approach was used for lake Whiteﬁsh during 2002-2004, where the data
were used to compare lake Whiteﬁsh distributions to lake trout, and develop
recommendations to reduce lake trout bycatch in lake Whiteﬁsh commercial ﬁsheries

(Chapter Three).

xi

The objectives in Chapter One were to:

1.

Obtain seasonal temperature-use proﬁles for the strains of lake trout being
stocked in Lake Huron

To examine whether the seasonal patterns vary signiﬁcantly among strains.
To provide more realistic temperature data for bioenergetics models of strain
differences in survival from sea lamprey attack and for models of prey

consumption by lake trout.

The objectives in Chapter Two were to use archival tag data recorded 2002-2005 to:

1.

Compare the temperatures occupied by lake trout strains in 2002-2005 to those
reported in Chapter One for 1999-2000 with regard to strain and time interval.
Examine whether lake trout of Finger Lakes, New York origins consistently
occupy both lower temperatures and greater depths during the summer months
than lake trout of Great Lakes origin.

Examine whether changes in the temperatures occupied by lake trout have altered

in light of recent changes in the abundance and composition of their prey.

The objectives in Chapter Three were to use archival tag data recorded 2003-2005 when

data from both lake trout and lake Whiteﬁsh were available to:

1.

Describe the seasonal temperature and depth distribution of lake Whiteﬁsh.

2. Compare the seasonal distribution of lake Whiteﬁsh to lake trout of Finger Lakes

and Great Lakes strain origins.

xii

3. Propose seasonal depth restrictions for commercial gear ﬁshed for lake Whiteﬁsh

for the purpose of minimizing lake trout bycatch.

xiii

References

Bergstedt, R.A., and Schneider, GP. 1988. Assessment of sea lamprey (Petromyzon
marinas) predation by recovery of dead lake trout (Salvelinus namaycush) from Lake
Ontario, 1982—85. Can. J. Fish. Aquat. Sci. 45: 1406—1410.

Brown, R.W., Ebener, M., and Gorenﬂo, T. 1999. Great Lakes commercial ﬁsheries:
historical overview and prognosis for the future. In Great Lakes ﬁsheries policy and
management, a binational perspective. Edited by W. Taylor, and GP. Ferreri.
Michigan State University Press, East Lansing, Mich. pp. 307-3 54.

Chopin, F., Inoue, Y., and Arimoto, T. 1996. Development of a catch mortality model.
Fisheries Research 25: 377-382.

Crowder, LB, and Murawski, SA. 1998. Fisheries bycatch: implications for
management. Fisheries (Bethesda) 23(6):8-17.

Eaton, J .G., McCormick, J .H., Goodno, B.E., O’Brien, D.G. Stefany, H.G., Hondzo, M.,
and Scheller, RM. 1995. A ﬁeld information-based system for estimating ﬁsh
temperature tolerances. Fisheries (Bethesda) 20(4):10-18.

Edsall, TA, and Cleland, J. 2000. Optimum temperature for growth and preferred
temperatures of age-0 lake trout. N. Am. J. Fish. Manage. 20: 804-809.

Elliott, J .M., and Hurley, MA. 1999. A new energetics model for brown trout, Salmo
trutta. Freshw. Biol. 42: 23 5-246.

Eshenroder, R.L., Payne, N.R., Johnson, J .E., Bowen, C., II, and Ebener, MP. 1995 .
Lake trout rehabilitation in Lake Huron. J. Great Lakes Res. 21(Supplement 1):
108-127.

Gallinat, M.P., Ngu, H.H., and Shively, J .D. 1997. Short-term survival of lake trout
released from commercial gill nets in Lake Superior. N. Am. J. Fish. Manage. 17:
136-140.

Goddard, CL, and Tait, J .S. 1976. Preferred temperatures of F3 to F5 hybrids of Salvelinus
fontinalis X S. namaycush. J. Fish. Res. Board Can. 33: 197-202.

Grande, M., and Andersen, S. 1991. Critical thermal maxima for young salmonids. J.
Freshw. Ecol. 6: 275-279.

Huff, D.D., Hubler, S.L., and Borisenko, A.N. 2005. Using ﬁeld data to estimate the
realized thermal niche of aquatic vertebrates. N. Am. J. Fish. Manage. 25: 346—360.

Hutchinson, GE. 1957. Concluding remarks. Cold Spring Harbor Symposium. Quant.
Biol. 22:415-427.

xiv

Ihssen, P. 1973. Inheritance of thermal resistance in hybrids of Salvelinusfontinalis and
S. namaycush. J. Fish. Res. Board Can. 30: pp. 401-408.

Ihssen, P., and Tait, J .S. 1974. Genetic differences in retention of swimbladder gas
between two populations of lake trout (Salvelinus namaycush). J. Fish. Res. Board
Can. 31: 1351-1354.

J obling, M. 1981. Temperature tolerance and the ﬁnal preferendum—rapid methods for
the assessment of optimum growth temperatures. J. Fish Biol. 19: 439-455.

Johnson, J .E., Ebener, M.P., Gebhardt, K., and Bergstedt, R.A. 2004b. Comparison of
catch and lake trout bycatch in commercial trap nets and gill nets targeting lake
Whiteﬁsh in northern Lake Huron. Michigan Department of Natural Resources,
Fisheries Research Report 2071, Ann Arbor, Mich.

Krueger, C.C., Marsden, J .E., Kincaid, H.L., May, BL. 1989. Genetic differentiation
among lake trout strains stocked into Lake Ontario. Trans. Amer. Fish. Soc. 118:
317-330.

Magnuson, J .J ., Crowder, LB, and Medvick, RA. 1979. Temperature as an ecological
resource. Am. Zool. 19: 331-343.

McCauley, R.W., and Tait, J .W. 1970. Preferred temperature of yearling lake trout,
Salvelinus namaycush. J. Fish. Res. Board Can. 27: 1729-1733.

Peterson, R.H., Sutterlin, A.M., and Metcalfe, J .L. 1979. Temperature preference of
several species of Salmo and Salvelinus and some of their hybrids. J. Fish. Res.
Board Can. 36: 1137-1140.

Schneider, C.P., Owens, R.W., Bergstedt, R.A., and O’Gorman, R. 1996. Predation by
sea lampreys (Petromyzon marinus) on lake trout (Salvelinus namaycush) in southern
Lake Ontario, 1982—92. Can. J. Fish. Aquat. Sci. 53: 1921—1932.

Tait, J. S. 1970. A method of selecting trout hybrids (Salvelinusfontinalis X S.
namaycush) for ability to retain swimbladder gas. J. Fish. Res. Board Can. 27 : 39-45.

Swink, W.D., 1993. Effect of water temperature on sea lamprey growth and lake trout
survival. Trans. Am. Fish. Soc. 122:1161—1166.

XV

TABLE OF CONTENTS

LIST OF TABLES ..................................................................................................... xvii

LIST OF FIGURES ..................................................................................................... xx

CHAPTER ONE

In situ determination of the annual thermal habitat used by lake trout

(Salvelinus namaycush) in Lake Huron ......................................................................... 1
Abstract .............................................................................................................. 2
Introduction ........................................................................................................ 3
Methods .............................................................................................................. 5
Results .............................................................................................................. 14
Discussion ........................................................................................................ 26
Acknowledgments ............................................................................................ 32
References ........................................................................................................ 33

CHAPTER TWO

Changes in bathythermal habitat use by lake trout in Lake Huron during a

period when the prey base changed rapidly ................................................................. 37
Abstract ............................................................................................................ 38
Introduction ...................................................................................................... 40
Methods ............................................................................................................ 43
Results .............................................................................................................. 55
Discussion ........................................................................................................ 65
Acknowledgments ............................................................................................ 75
References ........................................................................................................ 76

CHAPTER THREE

Seasonal and die] depth and temperature distributions of lake Whiteﬁsh and of Great

Lakes- and Finger Lakes-origin lake trout relative to exploitation and bycatch ......... 79
Abstract ............................................................................................................ 80
Introduction ...................................................................................................... 81
Methods ............................................................................................................ 84
Results .............................................................................................................. 90
Discussion ...................................................................................................... 102
Acknowledgments .......................................................................................... l 16
References ...................................................................................................... 117

xvi

LIST OF TABLES

Table 1. Summary of lake trout tagged with surgically implanted archival temperature
tags and released in Lake Huron, the number of lake trout tagged by date, location, and
source, the number known to have been captured or died, the number from which data
ﬁles were recovered, the number of ﬁsh used in the analyses, and (for the ﬁsh used in the
analyses) the total days at large ........................................................................................... 7

Table 2. Water temperatures (°C) occupied by lake trout in Lake Huron by year, half
month interval, and genetic strain. Labels for half month intervals are the midpoints
expressed as decimal months (e.g., 10.75 is the midpoint of the last half of October).
Means and SES are calculated from the individual means for the number (N) of ﬁsh
contributing data in each cell. Pairwise t tests were not corrected for experimentwise
error. Blank cells denote no data, ‘-’ denotes insufﬁcient degrees of freedom, and ‘*’ or
‘nS’ denotes signiﬁcance or non signiﬁcance at the or = 0.05 level. Strain abbreviations
are FLO, Finger Lakes origin; LLW, Lewis Lake wild; SMD, Lake Superior Marquette
domestic; and GLO, Great Lakes origin (combines LLW, SMD and two ﬁsh that could
not be assigned to strain, but could be identiﬁed from their ﬁn clips as either LLW or
SMD) ................................................................................................................................. 19

Table 3. Least square means (LSM), standard errors, and upper and lower 95%
conﬁdence intervals (CI) from a mixed mode] ANOVA predicting water temperatures
(°C) occupied by lake trout in Lake Huron by half month interval, and genetic strain for
combined data collected from mid October 1998 through June 2001. Labels for half
month intervals are the midpoints expressed as decimal months (e.g., 10.75 is the
midpoint of the last half of October). Strain abbreviations are Finger Lakes origin (FLO)
and Great Lakes origin (GLO). The GLO strain combines Lewis Lake wild (LLW),
Superior Marquette domestic (SMD) and two ﬁsh that could not be assigned to strain, but
could be identiﬁed from their ﬁn clips as either LLW or SMD ........................................ 25

Table 4. Summary of lake trout tagged with surgically implanted archival temperature
tags and released into Lake Huron, either at the Hammond Bay Biological Station
(HBBS), at Drurnmond Island (DI), or on Lake Huron (LH), the number of lake trout
tagged by dates, location, and source; the number recaptured; and the number and percent
of ﬁsh used in the analyses ................................................................................................ 45

Table 5. Results of model selection for a mixed-model ANOVA predicting the least-
squares mean temperature and depth occupied by lake trout of two genetic origins in
Lake Huron during 2002-2005. The AIC values are for two potential treatments of the
repeated nature of the observations crossed with ﬁve potential treatments of the random
components. The A AIC values are based on the lowest AIC values (in hold) for analyses
of temperature and depth. In one case the model did not converge (DNC) ...................... 54

xvii

Table 5. Results of model selection mixed-model ANOVAS of temperatures occupied in
Lake Huron during 1999-2000 and 2001-2005, run separately for Great Lakes and Finger
Lakes origin lake trout. The AIC values are given for two potential treatments of the
repeated nature of the observations and ﬁve potential treatments of the random
components. The A AIC values are based on the lowest AIC values for each genetic
origin (in bold). In one case the model did not converge (DNC) ..................................... 56

Table 7. Results of a mixed-model ANOVA examining factors inﬂuencing the mean
temperature (Temp) and median depth (Depth) occupied in Lake Huron during 2002-
2005 by lake trout of two genetic origins—Finger Lakes, NY and the upper Great Lakes.
Data were recorded with archival tags. The main effects were: Origin, Period (half-
month seasonal periods), and the interaction of Origin and Period ................................... 59

Table 8. Least-square means (LSM) of individual ﬁsh mean temperatures (°C) occupied
and the degree-days (DD) experienced by Great Lakes and Finger Lakes origin lake trout
in Lake Huron by half-month interval from a mixed-model ANOVA and results of post-
hoc comparisons (Tukey t-tests) between origins within each interval ............................. 62

Table 9. Least-square means of individual ﬁsh medians for depth (m) occupied by Great
Lakes and Finger Lakes origin lake trout in Lake Huron by half-month interval from a
mixed-model AN OVA and the results of post-hoe comparisons (Tukey t-tests) between
origins within each interval ................................................................................................ 64

Table 10. Results of a mixed model ANOVA testing whether the temperatures occupied
by lake trout of Great Lakes (GLO) and Finger Lakes (FLO) origins in Lake Huron
changed over time , as measured with archival tags. The main effects were study (two
intervals of years; 1998-2001 and 2002-2005), season (24 half-month seasonal intervals),
and the interaction of study and season ............................................................................. 65

Table 11. Summary of lake trout and lake Whiteﬁsh tagged with surgically implanted
archival temperature tags and released into Lake Huron, including the number of lake
trout tagged by date interval, Species, location, source, the number released, the number
and percent recaptured, the number and percent of ﬁsh used in the analyses, and (for ﬁsh
used in the analyses) the total days at large ....................................................................... 92

Table 12. Results of model selection for a mixed-model ANOVA predicting the least-
squares mean temperature and depth occupied by lake Whiteﬁsh and lake trout of Great
Lakes origin and Finger lakes, NY origin in Lake Huron during 2003-2005. The AIC
values are for two potential treatments of the repeated nature of the observations crossed
with ﬁve potential treatments of the random components. The A AIC values are based on
the lowest AIC values (in bold) for analyses of temperature and depth ............................ 93

Table 13. Least-square means (LSM) of individual ﬁsh means for temperature (°C)
occupied by Great Lakes origin (GLO) and Finger Lakes origin (FLO) lake trout and by

xviii

lake Whiteﬁsh (LWF) in Lake Huron by half-month interval from a mixed-model
ANOVA and the results of post-hoe comparisons (Tukey t-tests) between origins within
each interval ....................................................................................................................... 97

Table 14. Least-square means of median depths (m) occupied by Great Lakes origin
(GLO) and Finger Lakes origin (FLO) lake trout and by lake Whiteﬁsh in Lake Huron by
half-month interval from a mixed-model AN OVA and the results of post-hoe
comparisons (Tukey t-tests) between origins within each interval; Bonferroni corrections
were not applied ................................................................................................................. 99

Table 15. Results of mixed model ANOVAS assessing differences by light condition
(daylight or dark) in the mean temperatures (Temp) and median depths (Depth) occupied
in Lake Huron during 2002-2005 by lake trout of two genetic origins, upper Great Lakes
origin (GLO) and Finger Lakes, New York origin (FLO), and lake Whiteﬁsh LWF. Data
were recorded with archival tags. The main effects were: Group (GLO vs. LWF or FLO
vs. LWF), Period (half-month seasonal periods), and the interaction of Group and

Period ............................................................................................................................... 100

Table 16. Cumulative percent ﬁequency by half-month interval and upper limits of 5-m
depth intervals for Finger Lakes origin (FLO) lake trout, Great Lakes origin (GLO) lake
trout, and lake Whiteﬁsh (LWF) in Lake Huron, based on maximum depths observed with
archival tags for combinations of individual ﬁsh and hour during 2003-2005 ................ 103

xix

LIST OF FIGURES

Figure 1. Mean water temperatures occupied by four groups of lake trout in Lake Huron,
calculated by half-month interval from October 1998 to June 2002. Lake trout of Finger
Lakes, New York origin (FLO) are denoted by solid circles, the Lake Superior Marquette
Domestic (SMD) strain by open squares, the Lewis Lake Wild (LLW) strain by open
triangles, and ﬁsh that could not be assigned to strain, but known to be either SMD or
LLW strain by open diamonds ........................................................................................... 16

Figure 2. (A) Mean water temperatures occupied by lake trout of Upper Great Lakes
origin (GLO; ﬁne line and open circles) and of Finger Lakes, New York origin (FLO;
heavy line and closed circles) from October 1998 to June 2001in Lake Huron. The lines
represent the mean temperatures occupied, calculated by half-month intervals pooled
across years. Symbols and 95% conﬁdence intervals are least-squares means ﬁom a
mixed-model AN OVA treating the strain*half-month interval interaction as a random
effect and means for individual ﬁsh as repeated measures. (B) Average water
temperatures 0.6 m below the surface at NOAA buoy 45003 near the center of Lake
Huron (45.3 5° N, 82.84° W), calculated by half-month intervals, individually for 1998
through 2001 (lines) and pooled across years (symbols) ................................................... 17

Figure 3. T ernperature by depth averaged across seven sites in northern Lake Huron
(Latitude > 44.5° N) during 16 to 18 August 1999 and predicted mean temperatures
occupied by Finger Lakes Origin (FLO; ﬁlled circles) and Upper Great Lakes Origin
(GLO; open circles) lake trout during the ﬁrst half of August (large circles) and the last
half (small circles). Depth and temperature were collected with a bathytherrnograph
(Marc Tuchman, EPA Great Lakes National Program Ofﬁce, Chicago Illinois,
unpublished data). Predicted temperatures for lake trout are least squares means from a
mixed-model ANOVA using data from implanted archival tags collected during October
1998 through June 2001 ..................................................................................................... 24

Figure 4. Locations in Lake Huron of Drummond Island, the Hammond Bay Biological
Station, F orty-mile Point, Rockport, sites where lake trout were tagged at the site of
collection (squares), and boundaries of ﬁshery statistical districts MH-l to MH6 ........... 46

Figure 5. Frequency of hourly mean temperatures and depths occupied by individual ﬁsh
of Great Lakes (open boxes and circles) and Finger lakes, NY, (ﬁlled boxes and circles)
origin by month during 2002-2005. Horizontal lines are the median, boxes the 25th and
75th percentiles, vertical lines the 10th and 90th percentiles, and circles the 5th and 95th
percentiles. The dotted horizontal line shows the ﬁnal preferendum and the dashed line
the temperature for optimum growth for lake trout, as summarized by Christie and Regier
(1988) ................................................................................................................................ 58

Figure 6. Seasonal internal temperatures by half-month intervals for tagged lake trout of
Upper Great Lakes origin (GLO) and of Finger Lakes, New York origin (FLO) at large in

XX

Lake Huron as recorded with implanted archival tags. Symbols and 95% conﬁdence
intervals are least-squares means from a mixed-model AN OVA treating the origin*half-
month interval interaction as a random effect and half-month means for individual ﬁsh as
repeated measures. The data were collected from May 2002 to September 2005. Data
ﬁom 1999-2000 are from Chapter One. Panels a and b contrast GLO and FLO lake trout
within each data-collection interval and panels 0 and d contrast the data-collection
intervals for each genetic origin ......................................................................................... 61

Figure 7. Seasonal depth by half-month intervals occupied in 2002-2005 by tagged lake
trout of upper Great Lakes origin (GLO) and of Finger Lakes, New York origin (FLO) at
large in Lake Huron as recorded with implanted archival tags. Symbols and 95%
conﬁdence intervals are least-squares means ﬁom a mixed-model ANOVA ................... 63

Figure 8. The proportion of empty stomachs (dark ﬁll) and the percent of diet by weight
for nine prey items in stomachs of lake trout collected in three areas of the US. waters of
Lake Huron (Fig. 4). Data were provided by the Michigan Department of Natural
Resources, Alpena Fisheries Station ................................................................................. 69

Figure 9. Mean biomass (kg) and 95% conﬁdence limits in USGS fall trawl catches at
ﬁve ports in the Michigan waters of Lake Huron, 1976-2006 (data from Riley et al. in
press) ................................................................................................................................. 71

Figure 10. Frequency of hourly mean temperatures and hourly median depths by
half-month interval during daylight (sunrise to sunset) and dark (excludes twilight) for
Great Lakes origin lake trout (white ﬁll), Finger Lakes origin lake trout (light-gray ﬁll),
and lake Whiteﬁsh (dark-gray ﬁll) in Lake Huron during 2002-2005. In the box plots, the
horizontal line indicates the mean or median, the box the 25th and 75th percentiles, and the
whiskers the 10th and 90‘h percentiles. Horizontal lines in panels B and D Show modes for
depths of tribal gill net effort in 2003-2005 (solid) and 2006-2007 (dashed). Effort data
was provided by the Chippewa Ottawa Resource Authority, Sault Ste. Marie,

Michigan ............................................................................................................................ 95

Figure 11. Frequency of hourly maximum depths by half-month interval during daylight
(sunrise to sunset) and dark (excludes twilight) for Great Lakes origin lake trout (white
ﬁll), Finger Lakes origin lake trout (light-gray ﬁll), and lake Whiteﬁsh (dark-gray ﬁll) in
Lake Huron during 2002-2005. In the box plots, the horizontal line indicates the mean or
median, the box the 25‘h and 75th percentiles, and the whiskers the 10b and 90th
percentiles. Horizontal lines in panels B and D show modes for tribal gill net effort in
2003-2005 (solid) and 2006-2007 (dashed). Effort data was provided by the Chippewa
Ottawa Resource Authority, Sault Ste. Marie, Michigan .................................................. 98

Figure 12. Percent frequency of observations of maximum depth for combinations of
individual lake Whiteﬁsh, day, and hour during the ﬁrst half of august 2003-2005.
Percents add to 100% within each hour. The area between the lines at 05:30 and 20:30 is
the approximate period of daylight. Arrows Show the modal depths for tribal gill net
effort during the study and the following two years ...................... 102

xxi

CHAPTER ONE

Bergstedt, R.A., Argyle, R.L., Seelye, J .G., Scribner, K.T., and Curtis, G.L. 2003.171 situ
determination of the annual thermal habitat used by lake trout (Salvelinus namaycush)
in Lake Huron. .1. Great Lakes Res. 29 (Suppl. 1):347—361 .

Abstract

Records of the temperatures occupied by 33 lake trout (Salvelinus namaycush) at
large in Lake Huron were obtained for up to 14 months per ﬁsh, at 75-minute intervals,
ﬁ'om surgically implanted archival temperature tags. The dataset covered nearly three
years, from October 1998 to June 2001, and included 160,000 observations. The
objectives of the tagging were to obtain temperature data to reﬁne bioenergetics models
of sea lamprey (Petromyzon marinas) predation on lake trout, and compare the
temperatures occupied by strains of lake trout stocked in Lake Huron. The seasonal,
thermal-use proﬁles of lake trout followed the general warming and cooling pattern of
Lake Huron. During periods when the zone of surface water mixing extended below the
depth range occupied by lake trout, variability among individual ﬁsh and strains was low
and followed surface temperature. However, during the period of summer stratiﬁcation,
the average temperatures occupied varied substantially among individual ﬁsh and strains.
Strains (Lake Superior and Lewis Lake) originating from the upper Great Lakes occupied
similar temperatures. Between June and mid August, upper Great Lakes lake trout
typically occupied water several degrees warmer than that occupied by lake trout of
Finger Lakes, New York origin. Most of the lake trout occupied summer temperatures
lower than the preferred temperatures suggested by laboratory studies. In October, all
strains occupied water as warm as or warmer than that occupied in summer, which may

partially explain the higher lethality of sea lamprey attacks during October.

Introduction

Restoration of native lake trout (Salvelinus namaycush) in the Great Lakes continues
to be a high priority for US. and Canadian ﬁshery agencies. Although there is some
natural reproduction in Lakes Huron and Ontario, restoration has only been successful
enough to create self-sustaining stocks in Lake Superior. Possible impediments to
restoration in the other lakes include a loss of genetic variability, effects of non-native
species, habitat degradation, mortality from predation by sea lampreys (Petromyzon
marinas), and mortality ﬁom ﬁshing (Krueger et al. 1995; Ebener 1998). Research and
assessment considered necessary to evaluate lake trout restoration in Lake Huron include
community interactions, stocking strategies, and population and community modeling
(Ebener 1998). A key information need in each category is seasonal, temperature-use

proﬁles for the various hatchery strains of lake trout stocked.

Temperature is an important parameter governing food consumption and growth of
both lake trout and sea lampreys, yet descriptions of temperatures occupied by lake trout
as hosts are mostly derived from laboratory studies or based on captures in ﬁshing gear.
Temperature at capture in active gear reﬂects only the temperature occupied by the ﬁsh at
that moment. For ﬁsh captured in passive gear, only the temperature at deployment or
retrieval is known, which may not reﬂect the temperature at capture; in the Great Lakes
that difference can be substantial due to internal seiches. In the absence of in-situ data,
Stewart et al. (1983) assumed a preferred temperature of 10°C for lake trout in Lake
Michigan, hypothesizing that lake trout occupy the warmest available water as long as it

does not exceed the preferred temperature. The preferred temperature of 10°C was about

two degrees cooler than the temperatures preferred by yearling lake trout (11.7°C) in the
laboratory (McCauley and Tait 1970). The lower temperature was used because ﬁeld
observations indicated that lake trout normally occupy temperatures about two degrees
lower than in the laboratory (Spangler and Berst 1976; McCauley and Tait 1970).
Madenjian et al. (1995) derived temperature regimens for lake trout from the mean

temperatures in which lake trout were caught in gill nets and trawls.

Reduction and control of sea lamprey populations remains critical for restoration of
lake trout, and justiﬁcation for continued sea lamprey management ultimately rests on its
expected beneﬁts (Sawyer 1980; Koonce et al. 1993; Christie and Goddard 2003).
Deﬁning beneﬁts requires estimating the damage done by sea lampreys to their host
ﬁshes. A promising approach to damage estimation has been bioenergetics modeling
(Kitchell and Breck 1980; Kitchell 1990). The diurnal and seasonal temperatures
occupied by sea lampreys when attached to lake trout (their preferred host in the Great
Lakes) are necessary for developing and reﬁning models of sea lamprey grth and for
assessing the damage from attacks. Madenjian et al. (2003) integrated seasonal data on
sea lamprey growth (Bergstedt and Swink 1995), energy density (Cochran et al. 2003),
and temperature (this study) to construct a more realistic bioenergetics model of blood
consumption by sea lampreys, and to examine some new hypotheses regarding the
seasonal effects of sea lamprey predation on lake trout (Bergstedt and Schneider 1988;

Schneider et al. 1996; Bence et al. 2003).

To better understand their seasonal bathythermal distribution and to provide data for
bioenergetics models, archival (i.e., data recording) temperature tags were surgically
implanted in several strains of lake trout collected from Lake Huron during 1998-2000.
The objectives were to obtain seasonal temperature-use proﬁles for strains of lake trout in
Lake Huron, examine whether the seasonal patterns vary signiﬁcantly among strains, and
provide more realistic temperature data for bioenergetics models of strain differences in

survival from sea lamprey attack and for models of prey consumption by lake trout.

Methods

Tags and tagging

The archival temperature tags used in this study were manufactured by Vemco
Limited, Nova Scotia, Canada. Two types of archival tags were used, the standard TX
Minilog, containing a single thermistor, and a customized TX Minilog with a second
thermistor mounted on a 70-mm stalk. The purpose of the second thermistor was to
provide capability for simultaneous collection of internal and external temperatures.
Regardless of the tag used, all electronics were housed within the cylinder that was
surgically implanted in the abdominal cavity of the ﬁsh. In order to obtain external
temperatures, the stalk of the customized tag protruded through the body wall. External
water temperatures are not reported in this paper, and all recorded temperatures refer to
temperatures measured inside the abdominal cavity. The cylinder containing the
electronics and battery measured 16 mm in diameter by 71 mm long. The tags were

capable of measuring temperatures from —5 to 35°C with 0.2°C resolution and :1: 0.3°C

accuracy. The calibrations were checked against a calibrated thermometer when tags
were recovered; observed accuracy was always within the speciﬁed limits. The standard
tag weighed 23.5 g in air and 9.0 g in water, and the modiﬁed tag weighed 26.8 g in air
and 9.8 g in water. Fish were also marked externally with dart tags manufactured by
Floy, Inc., Seattle, Washington, that advertised a $100 (US) reward and provided contact

information.

Tagging of adult lake trout was conducted from 1998 through 2000 (Table 1). In
1998, lake trout were tagged and released during October at Drummond Island and
during November at the Hammond Bay Biological Station (HBBS). In 1999, tagging
was at Drummond Island during May and November and at the HBBS during May, June,
and August. In 2000, all tagging was conducted at the HBBS during May. The lake trout
tagged at Drummond Island were obtained from gill nets ﬁshed by the Chippewa Ottawa
Resource Agency (CORA) for lake trout assessment near Scarnmon Cove along the south
shore of Drummond Island, MI. The lake trout tagged at HBBS were obtained from
commercial trap nets ﬁshed by Gauthier and Spaulding Fisheries along the western shore

of Lake Huron between the HBBS and Alpena, MI.

Lake trout obtained from gill nets were transferred to aerated containers and
transported to shore and placed in 1000-L tanks supplied with a constant ﬂow of aerated,
ambient-temperature Lake Huron water. Lake trout ﬁom trap nets were placed in a
ﬂoating pen adjacent to the trap net pot and then transferred in aerated containers—either

directly by boat to the HBBS or indirectly by boat to shore and then by truck to the

HBBS. At the HBBS, they were held in raceways or in 1000-L tanks in aerated,

ambient-temperature Lake Huron water. After transfer to the tanks

Table 1. Summary of lake trout tagged with surgically implanted archival temperature
tags and released in Lake Huron, the number of lake trout tagged by date, location, and
source, the number known to have been captured or died, the number from which data
ﬁles were recovered, the number of ﬁsh used in the analyses, and (for the ﬁsh used in the
analyses) the total days at large.

 

 

 

Tagging Number Number Number
and reported of data of ﬁsh Total
Date release Number captured ﬁles re- used in days
tagged location Source released or dead covered analyses at large
Oct. 1998 D1 Gill net 90 13 l 1 9 6,303
Nov. 1998 HBBS Trap net 13 3 2 1 268
May 1999 D1 Gill net 12 l 0 0 0
May 1999 HBBS Trap net 56 10 9 8 1,549
Jun. 1999 HBBS Trap net 34 5 5 0 0
Aug. 1999 HBBS Trap net 8 0 0 0 0
Oct. 1999 D1 Gill net 43 7 7 7 3,648
May 2000 HBBS Trap net 62 10 8 8 1,396
Totals 318 49 42 33 13,164

 

or raceways the lake trout were held from four hours to four days before undergoing
surgery. Fish that were lethargic, appeared severely stressed, or showed visible injuries

were not used in the study.

Fish were anesthetized with clove oil (Anderson et al. 1997). Once anesthetized, the
ﬁsh were weighed, measured, and ﬁn clips recorded. For the surgical procedure, lake
trout were placed on a foam rubber-covered table and a small submersible pump was
used to maintain a ﬂow of water containing a maintenance concentration of anesthetic
across the gills. A 2- to 3-cm incision was made about 1 cm to either side of the ventral
midline anterior to the pelvic ﬁns. The thermal tag was inserted through the incision,
and, for the customized thermal tags, the external thermistor was threaded posteriorly out
through a small hole in the body wall just behind the pelvic ﬁns. Three or four
monoﬁlament sutures were used to close the incision and 3-M Vet-Bond® adhesive was
applied to the knots. A numbered F loy tag was inserted just below the dorsal ﬁn. The
procedure generally took from 3 to 5 minutes, after which the ﬁsh were transferred to a
recovery tank. For those ﬁsh tagged with the standard Minilog, the tag was simply
inserted in the incision, the incision closed as described, and the ﬁsh double tagged (one
tag on each side of the dorsal ﬁn) with the numbered Floy tags. Following surgery, most
ﬁsh regained equilibrium within 5 to 10 minutes and were held from 24 to 48 hours
before release, except during fall 1998, when most ﬁsh were released when they appeared
fully recovered from the anesthetic (generally within 3 to 4 hrs).

Information posters describing the study, advertising a $ 100 reward for return of the
tag and ﬁsh, and providing contact telephone numbers were displayed in tackle and
convenience stores, at launch sites, and provided to commercial and charter ﬁshers. The
Michigan Department of Natural Resources and the Ontario Ministry of Natural

Resources also publicized the study. In addition to the posters, the stalk of the external

thermistor on the customized Minilogs and the external Floy tags also canied reward

notices and contact information.

Other data sources

Seasonal surface water temperatures and measures of thermal stratiﬁcation in Lake
Huron were obtained from two sources. To describe the seasonal pattern of warming for
surface waters, temperature data for spring through fall of 1998 through 2001 were
obtained from a 3-meter discus buoy deployed by NOAA at station 45003 near the
middle of Lake Huron (45.35° N, 82.84° W). Temperature at the buoy was measured at
0.6 m below the surface, and the buoy is deployed from April through November of each
year. As an indication of the pattern of thermal stratiﬁcation and the likely depths
associated with observed temperatures, data collected with a bathytherrnograph during 16
to 18 August 1999 (Marc Tuchman, EPA Great Lakes National Program Ofﬁce, Chicago
Illinois, unpublished data) were averaged across seven sites in northern Lake Huron

(Latitude > 44.5° N).

Genetic strains of lake trout

In 1985 a study was initiated in Lake Huron to evaluate relative performance of the
Superior Marquette Domestic (SMD) strain and several alternative strains. The
ﬁsh included in this study were implanted with coded-wire tags (CWT) and marked
externally by removing the adipose ﬁn. Alternative strains included the Lewis Lake Wild
strain (LLW), the Jenny Lake Wild strain (J LW), the Seneca Lake Wild strain (SLW),

and the Lake Ontario Wild strain (LOW). The Lewis and Jenny Lake strains originated

from fry hatched from eggs collected in northern Lake Michigan and stocked in
Shoshone and Lewis Lakes in Yellowstone Park in 1889 (Crossman 1995). The Lake
Huron Technical Committee recommended J LW and LLW for Lake Huron because they
might contain genes from original Great Lakes stocks (Ebener 1998). The Seneca Lake
strain (SLW) originated from eggs taken in Seneca Lake (one of the New York Finger
Lakes) and was a strain that showed evidence of being less vulnerable to sea lamprey
attack (Schneider et al. 1996). The Lake Ontario strain (LOW) originated from eggs
taken from feral ﬁsh of mixed strains in Lake Ontario. All of these strains are considered

to be shallow water or “lean” forms of lake trout.

The LOW strain was included in Lake Huron stockings as a surrogate for the SLW
strain because of a Shortage of SLW eggs. Only the 1989, 1991, and 1992 year classes of
the LOW strain were stocked. Several sources provide evidence that the LOW and SLW
strains are genetically similar. Most of the LOW strain stocked were propagated from the
1983 year class of brood stock derived from gametes collected in Lake Ontario and held
at the Allegheny National Fish Hatchery. Ninety percent of the 1989 year class, 100% of
the 1991 year class (Elrod et al. 1996a), and 100% of the 1992 year class of LOW (D.
Blick, Allegheny National Fish Hatchery, personal communication) were from this 1983
brood stock. Marsden et al. (1993) estimated that the parental contribution to the 1983
LOW brood stock was approximately 81% SLW, and that this strain provided the only
statistically signiﬁcant contribution among the six strains previously stocked into Lake
Ontario. Krueger et a1. (1989) compared allelic variation in 16 collections of lake trout,

including SLW and the 1983 LOW brood stock, and concluded the two strains were

10

genetically similar. Therefore, the SLW and LOW strains were treated as one group.
Because the lake trout in the New York Finger Lakes likely experienced some mixing
over the years due to hatchery practices (Charles Krueger, Great Lakes Fishery
Commission, personal communication), the combination of SLW and LOW is hereafter

referred to as being of Finger Lakes Origin (FLO).

The genetic strains of ﬁsh in this study were identiﬁed from CWTS or a combination
of ﬁn clips and DNA analysis. Fish with CWTS were readily identiﬁed to strain if the
ﬁsh was recovered along with the thermal tag. In a few instances, anglers had discarded
the head but a tissue sample from a ﬁllet was obtained for DNA analysis. For ﬁsh with
only ﬁn clips, strain could not be determined from ﬁn clips because they were only year-
class speciﬁc and all strains were usually marked identically. For 10 ﬁsh where the strain
could not be determined from CWTS, individuals were assayed at eight microsatellite loci
(Page 2001). For eight of those ﬁsh, the ﬁn clips indicated the strain was either LLW or
SMD, and for the other two, ﬁn clips indicated either LLW, SMD, SLW, or LOW.
Individual ﬁsh were assigned to hatchery strain of origin based on the likelihood of
observing individual multilocus genotypes, given the expected ﬁ'equency of occurrence
in each strain in individual assignment tests (Paetkau et al. 1995). Estimates of statistical
conﬁdence in strain assignments for individual ﬁsh were derived from the posterior
probabilities using a Bayesian approach (Pritchard et al. 2000, Topchy et al. in review)
and considering the prior information on the possible strains associated with each ﬁn clip.
The tissue sample from one ﬁsh had deteriorated and strain could not be assigned. The

other nine ﬁsh were assigned a strain based on the DNA analysis; the conﬁdence in

11

correct assignment ranged ﬁom 0.74 to 0.98 with a mean conﬁdence of 0.90 and median

conﬁdence of 0.91.

Data intervals and statistical tests

The recording interval of the archival tags was set to 75 minutes, which resulted in a
maximum temperature record of 424 days. When the memory ﬁlled, recording ceased
and the data were retained until downloaded. For ﬁsh remaining at large until the
memory ﬁlled, the 424-day record insured that data collected during the recovery period

following surgery could be discarded while still retaining an entire year’s record.

Temperature data were summarized by seasonal intervals of half-month duration.
The ﬁrst half of each month included the ﬁrst 15 days and the second half the remaining
days in the month. Half-month intervals were not initially combined across years.
Because these data were intended to be used in bioenergetics models of sea lamprey
growth (Madenjian et al.2003), the arithmetic mean temperature (°C) in each half-month
interval was used to best represent the average number of degree-days experienced by a
ﬁsh. With each ﬁsh, there was a period following release when it was recovering from
surgery and the diel temperature pattern was suspect. Each plot of temperature was
inspected and a subjective judgment was made as to the date when behavior appeared
normal. For ﬁsh captured within 424 days of release, when the tag was still actively
recording, the date when the ﬁsh was caught in the ﬁshing gear was evident from a
cessation of diel temperature changes. Temperature observations before behavior

appeared normal and after the ﬁsh was caught were not used in the analyses. Data for

12

each combination of ﬁsh and half-month interval were used only if there was more than
seven days of data. To avoid pseudoreplication, subsequent analyses were based only on
the means for individual ﬁsh by half-month interval. Analyses also considered the
repeated nature of the measures among subsequent half-month periods for an individual

ﬁsh.

The temperatures analyzed and reported in this study were only those recorded inside
the abdominal cavity. At any instant, these could vary from the temperature of the
surrounding water. However, over the temporal scale of a half-month interval, the mean
internal temperature should approximate the average temperature of the water occupied

by a ﬁsh. To simplify, these are hereafter referred to as temperatures “occupied.”

To initially determine if any of the strains did not differ in temperature use and could
be combined, mean temperatures were compared in each of the possible two-way strain
combinations within each half-month interval using t-tests. Signiﬁcance was assumed at
the or = 0.05 level, and a Bonferonni correction or other method of controlling the

experimentwise error rate were not applied in these initial analyses.

The overall effect of strain on the temperatures occupied was evaluated with a

mixed-model ANOVA (SAS 1999). The underlying model for the analysis was

Th,s,y,f = .11 +ah +16s +7h,s +6h,y +8h,s,y,f (1)

l3

in which T is temperature and h, s, y, and f denote the half-month interval, strain, year,
and individual ﬁsh. The ﬁxed effects were the overall mean (it), the seasonal or
half-month adjustment (or), the strain adjustment (,8), and the interaction between
half-month interval and strain (y). Potentially, the random components are a year by
half-month variation (6) and residual error (6). Consecutive half-month means for the
same ﬁsh were treated as repeated measures with a ﬁrst order autocorrelation. Initially,
three versions of the model were ﬁt, ﬁrst treating 6 as a ﬁxed effect, second treating it as
a random effect, and third treating it as a random effect with a ﬁrst order autocorrelation.
Each successive iteration was evaluated with a likelihood ratio test. The best of those
three models were then reexamined to determine if there was any additional improvement
by allowing separate errors by strain. The least-squares means and conﬁdence intervals
from the selected model were used to describe the temperature occupied by each strain in

each half-month interval.

RESULTS

Releases and Recoveries

Sixteen of the 161 lake trout that underwent surgery at Drummond Island (10%) and
one of the 174 at the HBBS (1%) died before release for a total initial mortality of about
5%. A total of 3 1 8 tagged lake trout were released during the study—145 at Drummond

Island and 173 at the HBBS (Table 1), but a few of those likely died shortly after release.

Tagged ﬁsh (including those that died) ranged from 455 to 861 mm and averaged 657

mm in total length. The smallest ﬁsh weighed 0.87 kg and the average ﬁsh weighed 2.93

14

kg. The ﬁsh from the trap nets were Signiﬁcantly smaller (t = 10.23; df= 333; P < 0.001)
in length (average 618 mm) than those from the gill nets (average 700 mm). The wet
weight of the tag never exceeded the suggested maximum of 1.25% of body weight
(Winter 1996); the maximum was 1.1% for the smallest ﬁsh and 0.3% for the average

ﬁsh.

Of the 318 lake trout released, 49 (15%) were either captured or died. Three were
caught in a trap net and released unharmed just a few days after tagging by the same
commercial ﬁsherman who initially caught them, one was caught in a gill net the
morning after tagging, the external Floy tag from one was recovered from a commercial
ﬁsh house (but no archival tag), one archival tag was found on a beach (indicating the
ﬁsh had died), and one recovered tag had malfunctioned. Valid data ﬁles were recovered
from 42 (13%) of the ﬁsh released (Table 1). Of those, strain could not be determined for
two ﬁsh and seven ﬁsh had not been free long enough to meet the requirement for the
minimum number of temperature observations (7 days). Thirty-three ﬁsh provided

useable data from at least one half-month period and were included in analyses.

All recovered ﬁsh appeared healthy. Incisions showed no signs of redness and ﬁsh
captured after about a month at large showed visible healing. There was never more than
a slight redness where the tag rested against the body wall. The site where the external
temperature probe passed through the body wall usually exhibited some irritation and
redness. There was a small weight loss for many of the ﬁsh recaptured, but the loss

(average, 135 g) was not signiﬁcant (paired t-test; t = 1.569; df = 21; P = 0.131). The

15

change in mean length of ﬁsh (-2 mm), measured at release and again at recapture, was

also not signiﬁcant (paired t-test; t = 0.328; df= 23; P = 0.745).

Between November and the following April, the mean recorded temperatures within
half-month intervals varied little among individual ﬁsh (Fig. 1). There were no

bathythermal proﬁles available seasonally for Lake Huron, but the general pattern of

Fig. 1. Mean water temperatures occupied by four groups of lake trout in Lake Huron,
calculated by half-month interval from October 1998 to June 2002. Lake trout of Finger
Lakes, New York origin (F LO) are denoted by solid circles, the Lake Superior Marquette
Domestic (SMD) strain by open squares, the Lewis Lake Wild (LLW) strain by open
triangles, and ﬁsh that could not be assigned to strain, but known to be either SMD or
LLW strain by open diamonds.

14 r TV v r r I v v v I I f T

 

VVVVVVVVVVV

IIIIII

12- o
o

 

Temperature (C)

 

 

 

 

 

ONDhFMAMJJASONDhFMAMJJASONDJFMAMJ
1998 1999 2000 2001
Year and Month

 

l6

increasing temperatures occupied by lake trout in spring and decreasing temperatures
occupied in late autumn, irrespective of strain (Fig. l and 2A), approximated that of the
surface temperature at the midlake buoy (Fig. 28). From about June through October all

lake trout occupied cooler water than was available near the surface. The FLO ﬁsh

Fig. 2. (A) Mean water temperatures occupied by lake trout of Upper Great Lakes origin
(GLO; ﬁne line and open circles) and of Finger Lakes, New York origin (FLO; heavy
line and closed circles) from October 1998 to June 2001in Lake Huron. The lines
represent the mean temperatures occupied, calculated by half-month intervals pooled
across years. Symbols and 95% conﬁdence intervals are least-squares means ﬁ'om a
mixed-model AN OVA treating the strain*half-month interval interaction as a random
effect and means for individual ﬁsh as repeated measures. (B) Average water
temperatures 0.6 m below the surface at NOAA buoy 45003 near the center of Lake
Huron (45.3 5° N, 82.84° W), calculated by half-month intervals, individually for 1998
through 2001 (lines) and pooled across years (symbols).

 

12 Tﬁ'r'rETYTrl'rrr'T'r'

10-

Temperature (C)

 

 

 

'r'r'rrrfrirrrrrrrrr'r'
JFMAMJJASOND

 

 

 

 

j r T I f l V I 1 I I I T 7 Y I I r r 1 r Tﬁ A
O , “r I
V 20 1 «15 \g‘tgz -
e .1 .' ,‘./’l , ~- K“&\
3 15 a ‘ ‘
‘9" ,
S _ 9. ‘ 3\
m 10 "‘ ll'l . "'
E 5 i ‘ .
I1) ‘ 0'93"“ /

0 — _
TT WI'TTT ITITT rrrr
J F M A M J J A S O N D
Month

occupied their warmest waters of the year during late September through early November
(Fig. l). The SMD and LLW ﬁsh also occupied relatively warm waters in late September
and early October, but occupied their warmest waters of the year during several
half-month intervals earlier in the summer. By early December, lake trout were
occupying water at temperatures averaging 6°C or less, and by late December they were
in water around 3°C (Fig. 1, Table 2). From January through mid March, all lake trout
occupied water that was 1°C or less (Fig 1, Table 2). Surface water temperatures were
not available for comparison during the winter period because the buoys are removed

each year.

The mean water temperatures occupied by each possible combination of strains were
initially compared with t-tests in each combination of year and half-month interval (Table
2); the objective was to determine if any of the strains could be combined. Because the
number of comparisons was large and the experiment-wise error not controlled, these
tests were liberal and a number of signiﬁcant differences were expected due to chance.
However, none of the 26 comparisons between the LLW and SMD strains were
signiﬁcant. The LLW and SMD strains were therefore combined as being of Upper
Great Lakes origin (GLO) for further analyses. Two of the ﬁsh for which genetic strain
could not be determined had ﬁn clips identifying them as either LLW or Slvﬂ) and their

strain was also assigned as GLO.

Analysis of all data collected from October 1998 through June 2001 in a single

mixed-model, repeated-measures ANOVA, indicated there were signiﬁcant seasonal

l8

 

 

 

 

 

m: 84 N v6 .. a 3» $6 e Ne megs 33
we mm; N m.m - H e._ de A. A: mN;q 39
we N_.o N m._ - _ e; ONd e m._ med 32
8 3.0 N wd - H NA 86 5 ed mN.m e02
... ovd N ad - _ mg Nod 5 ed meN oom—
... wﬁo N md - M o; mod n We mN.N e92
... Rd N Nd - _ o; 86 A. v.0 we; 33
m: 86 N md - _ 9o ood e md mN._ 32
we mmd N Ym - _ Em NN.o e mm KN“ mom“
2 Rd N 9m - _ o6 3.0 e 5m mN.N_ meg
me cod N we - H we Ed 5 m6 3.: 33
m: - _ 0.x - _ 9w Ed e mm mN.: M33
... - _ NS - ﬂ NN ONd v N.o_ 3.3 M32

92m 92m 3.: mm 2 $02 mm. 2 $02 mm 2 :82 mm. 2. 522 BEBE 80>

.m> .m> .m> 5:08

>915 04m 04m Ed:

$81 815 040 02m 3‘5 0.:

£3 mo 33QO :53 03280

 

.Ezm 8 >91: 556 mm 320 am .55 Bob “accuse“: on 2:8 Sn dream 9 ccewmmme

on 8: 2:8 55 Ema 95 use 02m .315 mocBEoov Ewto 88:3 $.20 .040 98 63888 0303.85 .8596 8:3 .sz

6:3 8:5 $33 $7.: ”Ewes 80:3 Swan 6.: ca Eczema/Bane 58% 432 36 n o 2: “a 088$:me no: no ooﬁowawmm
882% .2: co 2... new .Eoeoot mo moocwoe 25653,: mowoeoe .-. See on 802% £00 x55 coho omEHeoEtomxo com 38880
8: 953 $83 Begum .=8 :08 5 See 3:33:00 2mm mo AZC c0285: 2: 8m £82: 36365 2: 88m poem—3030 8m mmm new
£8.32 .CBBoO mo was 3.3 05 mo 2:328 06 mm m2: swov 3:88 Remove me. 398on $5828 2: 0.8 338:: ﬁeoadﬁn
new £33 .265 ouoeow 98 .338E 5:08.23 .80» .3 :05: 8:3 5 Bob 83 .3 @2958 Gov 855388 833 .N «Baa.

l9

 

 

 

 

we we m: 3.0 v :c - _ Nd w: m 3. mNd c 5N meNﬁ add—
me we we Ed v hm - g 3. odd m dd 3:0 0 mm mN.N~ 092
m: m: we wed d We NNd m Nd mod N md cNd a md 3.: 32
we we we med o ms emd m ed and m 3. 2d a E. mN.: 32
me we we wmd o w.» Nd m Nd Nod m ed med o Nd m2: 3%
me we we de m dd MNd N Nd omd m ed Ed m md_ mNd_ 32
me me we omd m dd oNd N md amd m Md 3d m dd med 3%
m: m: m: oNd c md omd m as NNd m ed cmd m 5.5 mNd 82
we we we 2d e ed de v ed 8d m md Nd o Nd med 8%
m: ... .. med e dd Nwd V dd Nvd m We VNd 0 ed mNd 32
m: m: ... mNd 2 ad de o md Sd m Nd ~nmd 0 d6 mes 82
we ... ... id 2 ad omd o wd mod m dd 3d d wd NH 32
m: ... ... Ed 2 Nd mmd o md A: m dd oNd 0 ed med 32
me ... we NNd e We oNd v «d omd N Nd mod o as mNd 32
as Ed m g VNd N dd 9d e in mud add—
me 8d N dd - L ed omd e ed mN.m 32
02m 92m 313 we. 2. 982 ma. 2 :82 mm 2 :82 mm. 2 $02 REESE;
.m> .m> .m> 588
315 01$ 01—..— $32
38: 33> 040 92m 3.5 04m
$8 mo $38M £86 3650

 

20

dosﬁﬁoo .N 03.3.

 

 

 

 

 

- - _ Nam - d 5m - _ _.N 3.2 dooN
- - _ Em - ﬂ Nam - ﬂ Nd mN.N~ OOON
m: - _ vs - _ v8 NNd m Nd m5: doON
._. - a .2. - ~ We VNd m wd mN.: OQON
... - ~ 0.x - ﬂ ed nod m 0.2 med“ OOON
m: - _ ed— - _ ed— mNd m md_ mNd_ 88
we - _ 5.2 - ~ ed“ 5d d as med OOON
m: wed N ed vodN ed wmd o md mNd OOON
we no; N as SAN as mmd o md we.» OOON
m: odd m 2. cddm ms oNd o no mNd OOON
... mod v Nd mode Nd omd e md mes 88
we Nod o wd Node wd de w NS NS OOON
m: me me mNd v wd - _ dd mmdm wd Ed 0 md KN oddN
m: m: m: :d v wd - ﬂ ed wmdm ad 3d 0 wd mNN OQON
m: m: we end v v; - _ ed modm e; mod 0 ed we; oOON
we we ... dad v 1N - ﬂ dd 8.; eN eod o wd mN._ oddN
92m 92m 315 Mm Z 502 we. 2. 2.82 mm 2 SEE mm. Z 982 Etooetmoar
.m> .m> .m> 5:08
3.5 DIE 0.: Arum
983 cm?» 040 92m >71: 0.5
£8 mo 833m 58% 36:00

 

83:80 .N 2.3

21

 

 

 

 

- - H v.3 - ~ 0.2 - ﬂ wd who SON
- - _ N.w - _ N.w - ~ as mNd SoN
- - H We - _ m6 - ﬂ m6 mud SON
- - _ m6 - _ m6 - _ ad mN.m SON
- - _ w; - ﬂ w; - M YN mud SON
- - — o4 - _ o; - _ m; WNzw SON
- - _ 9o - _ 0.0 - _ md mud SON
- - ~ ed - ﬂ 9o - ~ No mN.m SON
- - _ 5o - _ 5o - _ N.o mnN SON
- - M n; - ﬂ 5; - ﬂ ed mN.N SON
- - _ _.N - _ _.N - ﬂ m4 mm; SON
.. .. ﬂ wN - ﬂ w.N - H to NA SON
92m 92m >941 NW 2 982 Wm. 2. 582 mm 2 S32 Mm. 2 302 333580.».
.m> .m> .m> 5:08
315 0‘: 01E .bmm
38: 83> 0A0 92m 344 04m
$3 mo 333m 595m 2850

 

3:538 .N 3.3

22

differences between the FLO and GLO strains. The model (equation 1) was ﬁrst ﬁt
with the year by half-month variation (8) as a ﬁxed effect. That model explained a
signiﬁcant portion of the variation (likelihood ratio test; P < 0.0001). Treating 5 as a
random effect resulted in a sigﬁﬁcant additional improvement (likelihood ratio test; P <
0.0001) and both of these models provided a good ﬁt to the data in Table 2. Treating 5
then as a random effect with ﬁrst-order autocorrelation did not improve the model
(likelihood ratio test; P = 0.09), and the resulting model also yielded a questionable ﬁt
to the data, with predicted temperatures in winter below 0°C. Treating 5 as a random
effect without autocorrelation, but allowing separate errors by strain, resulted in a
ﬁirther improvement (likelihood ratio test; P = 0.001 ). In that model, the coefﬁcients
for strain (P = 0.009), half-month interval (P < 0.0001), and the strain by half-month
interval interaction (P = 0.026) were all signiﬁcant. The least squares estimates of mean
temperature and 95% conﬁdence intervals in each strain by half-month interval
combination were also calculated (Table 3). The conﬁdence intervals for these
estimates did not overlap between FLO and GLO lake trout from mid June to mid

August (Fig 2A).

To assess how different the depths occupied by FLO and GLO lake trout might be,
bathythermograph data (Marc Tuchman, EPA Great Lakes National Program Ofﬁce,
Chicago Illinois, unpublished data) collected during 16 to 18 August 1999 at seven
locations in northern Lake Huron (Lat > 44.5°N) were used to calculate the average
temperature by half-meter depth interval (Fig. 3). The seasonal locations of tagged ﬁsh

were not known and bathythermal proﬁles can vary widely by location. Therefore,

23

plotting the least-squares mean temperatures occupied by FLO and GLO lake trout
(Table 3) on the temperature-depth proﬁle can only provide a general sense of how
different the depths occupied might have been. The dates for the bathythermograph
data are near the temporal boundary between the early and late August intervals.
Because temperatures occupied by FLO and GLO ﬁsh in early August were different
and those in late August were not (Fig. 2A), the means were plotted from both periods.
Their locations on the depth-temperature proﬁle suggest that the FLO ﬁsh were located
near the lower boundary of the metalimnion. The GLO ﬁsh occupied shallower depths,

although the maximum difference is only about 6 m (Fig. 3).

Fig. 3. Temperature by depth averaged across seven sites in northern Lake Huron
(Latitude > 44.5° N) during 16 to 18 August 1999 and predicted mean temperatures
occupied by Finger Lakes Origin (FLO; ﬁlled circles) and Upper Great Lakes Origin
(GLO; open circles) lake trout during the ﬁrst half of August (large circles) and the last
half (small circles). Depth and temperature were collected with a bathytherrnograph
(Marc Tuchman, EPA Great Lakes National Program Ofﬁce, Chicago Illinois,
unpublished data). Predicted temperatures for lake trout are least squares means from a
mixed-model ANOVA using data from implanted archival tags collected during
October 1998 through June 2001.

 

Depth (m)

 

 

 

 

140 IIIITITI
2468101214161820

Temperature (C)

24

Table 3. Least square means (LSM), standard errors, and upper and lower 95%
conﬁdence intervals (CI) from a mixed-model ANOVA predicting water temperatures
(°C) occupied by lake trout in Lake Huron by half-month interval, and genetic strain for
combined data collected from mid October 1998 through June 2001. Labels for
half-month intervals are the midpoints expressed as decimal months (e. g., 10.75 is the
midpoint of the last half of October). Strain abbreviations are Finger Lakes origin
(FLO) and Great Lakes origin (GLO). The GLO strain combines Lewis Lake wild
(LLW), Superior Marquette domestic (SMD) and two ﬁsh that could not be assigned to
strain, but could be identiﬁed from their ﬁn clips as either LLW or SMD.

 

 

 

 

FLO GLO
Half-month Lower Upper Lower Upper
interval LSM SE c1 CI LSM SE Cl C]
1.25 0.6 0.39 -0.1 1.4 1.9 0.40 1.1 2.7
1.75 0.6 0.39 .01 1.4 1.3 0.40 0.5 2.1
2.25 0.6 0.39 .01 1.4 0.9 0.40 0.2 1.7
2.75 0.7 0.39 -0.1 1.4 0.8 0.40 0.0 1.6
3.25 0.9 0.39 0.2 1.7 0.9 0.40 0.1 1.6
3.75 1.3 0.39 0.5 2.1 1.3 0.41 0.4 2.1
4.25 2.0 0.39 1.3 2.8 2.6 0.44 1.7 3.5
4.75 4.0 0.39 3.2 4.8 4.0 0.44 3.1 4.9
5.25 6.2 0.39 5.5 7.0 6.3 0.42 5.5 7.2
5.75 7.2 0.38 6.4 8.0 7.6 0.32 7.0 8.2
6.25 8.0 0.38 7.2 8.7 9.0 0.28 8.5 9.6
6.75 7.9 0.39 7.2 8.7 9.4 0.26 8.9 9.9
7.25 7.4 0.41 6.6 8.2 8.9 0.28 8.3 9.5
7.75 7.1 0.42 6.2 7.9 8.6 0.30 8.0 9.2
8.25 6.3 0.42 5.5 7.2 8.4 0.34 7.8 9.1
8.75 7.4 0.43 6.6 8.3 8.1 0.36 7.4 8.9
9.25 8.2 0.43 7.3 9.1 8.2 0.38 7.5 9.0
9.75 8.5 0.43 7.6 9.4 9.1 0.42 8.3 10.0
10.25 10.6 0.43 9.7 11.4 9.7 0.42 8.8 10.5
10.75 9.8 0.35 9.1 10.5 8.5 0.38 7.8 9.3
11.25 8.4 0.34 7.7 9.1 7.5 0.38 6.7 8.2
11.75 6.3 0.34 5.6 7.0 6.7 0.36 5.9 7.4
12.25 5.4 0.38 4.6 6.1 5.6 0.39 4.8 6.4
12.75 2.8 0.39 2.0 3.5 3.8 0.39 3.0 4.6

 

25

Discussion

There are no published accounts dealing speciﬁcally with the effects of internal tags
on adult lake trout performance or behavior, and such an evaluation was not done in this
study. However, the wet weights of the implanted tags were always a smaller
percentage of ﬁsh weight than the upper limit of 1.25% suggested by Winter (1996).
Furthermore, recent tests with rainbow trout (Oncorhynchus mykiss) suggest that the
acceptable limit could be set higher without affecting performance (Brown et al. 1999).
Adult Atlantic salmon (Salmo salar) with weights ranging from 1,021 to 2,388 g,
showed no decline in swimming performance or blood physiology when tested 6 to 20
days after tagging with internal tags weighing 7.9 g in water and ranging from 0.3 to
0.8% of ﬁsh weight (Thorstad et al. 2000). In the present study, the wet weight of only
nine tags were a larger percentage of the ﬁsh weight (max 1.1%), and the average
(0.3%) was equal to the lowest percentage in their study. Mock tags adjusted to 1%
(wet weight) of body weight did not affect the metabolic rate of sea bass (Dicentrarchus
labrax), also indicating that small additions to body weight should not affect
performance (Lefrancois et al. 2001). In the present study, the wet weight of only one
tag exceeded 1% of the ﬁsh’s body weight. It is always possible that tags could affect
behavioral or physiological processes in ways that are difﬁcult to anticipate. For
example, in a study by Perry et al. (2001), juvenile Chinook salmon (Oncorhynchus
tshawytscha) compensated for transmitter mass through changes in air bladder volume,
but changes in depth affected the buoyancy of tagged ﬁsh (whose buoyancy depended
on a larger gas volume) more rapidly than untagged ﬁsh. Perry et a1. cautioned that

reduced buoyancy at depth could affect behavior and physiology. Although questions

26

still remain about how internal tags affect ﬁsh, the available data suggest that for tags

within weight guidelines, the effects are probably minimal.

Although interpretation of the data in the present study was hampered by a lack of
depth information coincident with the temperature data ﬁ'om the tags, some inferences
can be drawn about the winter depths occupied by lake trout from the temperatures they
occupied and published winter bathythermal proﬁles for the other Great Lakes.
Regardless of the strain, lake trout remained at temperatures less than 1°C during
winter, which suggested they did not move deeper in search of warmer water during the
period of inverse winter stratiﬁcation. If the winter thermal proﬁle in Lake Huron was
similar to that described for Lake Michigan during February 1964 (Wells 1968), they
must not have been deeper than 40 to 50 111, although temperatures near 3°C were
probably available at 100 m. Temperatures in Lake Ontario during February 1984 were
1°C at 15 m and 1.7°C at 35 m, with temperature rising to 19°C at 95 m and 3.7°C at
215 m (Bergstedt and O’Gorman 1989). Based on the Lake Ontario proﬁles, lake trout

would have been shallower than 35 m during the winter months.

The variation in temperatures occupied was lowest among individual ﬁsh ﬁom late
autumn through April. In autumn there is a partial circulation (Ruttner 1963) due to
decreasing temperature in the epilimnion that reduces the stability of stratiﬁcation and
allows wind events to begin mixing the epilimnion into the hypolimnion. From October
or November through the winter, surface waters can be homothermous to a considerable

depth. For example, on October 14, 1964 in Lake Michigan, the temperature varied by

27

only 07°C in the upper 50 m of water (Wells 1968). Likewise, during February 1984 in
Lake Ontario, there was only a 08°C difference in temperature throughout the upper
110 m of water. During this time of the year, lake trout are forced to occupy cold water
within a narrow range of temperatures, unless they actively select a depth below the
zone of partial mixing in autumn (Bergstedt and Swink 1995) or below the inverse

stratiﬁcation in winter (Wells 1968; Bergstedt and O’Gorman 1989).

The signiﬁcance of the effect of strain and of the strain by half-month interaction in
the ANOVA suggests that the strains occupied different temperatures, but not in all
seasons. The pattern of overlap of the conﬁdence intervals for the two strains (Fig. 2A)
shows that the differences were the most apparent from mid June to mid August when
stratiﬁcation was strongest. Gill nets ﬁshed in Lake Ontario in September showed that
the Seneca strain of lake trout, age-3 and older, was caught in deeper and colder water
than the Superior strain (Elrod et al. 1996b). Temperature differences between strains
had disappeared by early September in Lake Huron. This could be due to the more
northern latitude of Lake Huron and an earlier breakdown of stratiﬁcation, although a
difference attributable to annual variation is also a possibility. Judging from the August
1999 bathythermal proﬁle in Lake Huron (M. Tuchman, EPA, Chicago, IL, unpublished
data), the difference in temperatures occupied by the strains implies a relatively small
but consistent difference in depths occupied, unless these strains consistently occupied
different areas of the lake with differing average depths of the thermocline. Archival
tags that determine both depth and temperature would provide additional insights by

providing concomitant measures of the depth and temperature occupied.

28

Differences in sea lamprey attack rates and mortality of lake trout from those attacks
were thought to be related to differences in seasonal bathythermal distributions of
strains of lake trout (Schneider et a1. 1996). Mortality of lake trout induced by sea
lampreys in Lake Ontario was documented using incidental recoveries of dead,
wounded lake trout in bottom trawl catches—but only during a brief period from late
September to early November (Bergstedt and Schneider 1988; Schneider et al. 1996).
Also, the Lake Superior strain of lake trout was observed to suffer consistently higher
rates of sea lamprey attacks in both Lakes Huron (Eshenroder et al. 1995a) and Ontario
(Schneider et al. 1996). If the strains differed in their bathythermal distributions and
one strain more closely approximated the bathythermal distribution of sea lampreys, it
could explain the difference in attack rates and in any subsequent mortality per attack
due to changes in metabolic rate and blood consumption (Swink and Hanson 1989;
Madenjian et al. 2003). The FLO and GLO strains of lake trout do occupy, at least

during portions of the year, different temperatures, and presumably, different depths.

Seasonal differences in the water temperatures available to lake trout may also
explain the occurrence of most lamprey-induced mortality during a brief period in
autumn (Bergstedt and Schneider 1988; Schneider et al. 1996) and the coincident rapid
growth of sea lampreys during that same seasonal period (Bergstedt and Swink 1995).
Bergstedt and Swink speculated that lake trout occupy the warmest water of the year
during the partial autumnal circulation (Ruttner 1963). The archival tag data support

that speculation. Higher temperatures and the resulting higher metabolic rates would

29

provide a partial explanation for the increased growth of sea lampreys in autumn
(Bergstedt and Swink 1995) and partially account for the sharp increase in sea lamprey

predation observed during that season (Bergstedt and Schneider 1988).

Although the data presented here advance knowledge of the actual temperatures
occupied by two strains of lake trout, it is still unclear how “preference” for depth and
temperature might interact. Because of that concern, the term “occupied” was
intentionally chosen instead of “preferred.” When presented with a choice of
temperatures, lake trout show a consistent preference for temperature in laboratory tests
(McCauley and Tait 1970; Goddard and Tait 1976). However, native stocks of lake
trout in the Great Lakes have historically inhabited a wide range of habitats and
exhibited distinct phenotypic differences (Krueger and Ihssen 1995). There is evidence
that other factors that would lead to differing selections of depth (and hence
temperature) are heritable characteristics of Salvelinus species. Fat content is thought to
affect depth distribution and vertical migration (Eshenroder et al. 1995b), and has been
demonstrated to be a heritable characteristic, with hybrids having an intermediate fat
content (Eschmeyer and Phillips 1965). Depth preference is also related to swim
bladder gas retention, which determines the depth of neutral buoyancy. Gas retention
can be controlled through selective breeding of lake trout with hybrids of lake trout and
brook trout (Tait 1970; Goddard and Tait 1976). It has also been demonstrated that gas
retention varied between lake trout from two lakes, that interpopulation crosses were
intermediate in gas retention, and that gas retention correlated with depth distribution of

the parent population (Ihssen and Tait 1974). During the period of stratiﬁcation,

30

preference for depth could have a strong interaction with temperature preference in

determining the actual temperatures occupied.

It is also possible the strain differences observed in this study were related to
differences in geographic distribution of strains corresponding to differences in prey
distribution and abundance. The temperature for optimal growth is affected by ration
size (Brett 1971), and a positive effect of ration size on temperatures selected by lake
trout has been demonstrated in the laboratory (Mac 1985). The relation of food
availability to the temperatures occupied by Paciﬁc salmon (Oncorhynchus spp.) is
being considered as an important factor controlling their seasonal distribution (W elch et
al. 1995). If food availability differed among lake trout strains, this could affect
temperature selection. Separating the likely importance of various environmental and
hereditary factors that might affect temperature selection will be a difﬁcult task that
initially requires collecting concomitant depth and temperature data and potentially

assembling gee-referenced, strain-speciﬁc data on stomach contents and food habits.

Temperature selection inﬂuences the interaction of lake trout with other species by
controlling metabolism (and the amount of prey consumed) and by determining spatial
overlap with prey and with other predators. The thermal requirements of lake trout
were formerly derived from laboratory studies or ﬁ'om temperatures at capture in
various ﬁshing gears. The data presented in this study indicate that during the period of
summer stratiﬁcation, lake trout tend to inhabit temperatures somewhat lower than the

generalization of 10°C used in models such as that of Stewart et al. (1983) to predict

31

prey consumption or of Kitchell and Breck (1980) to predict predation of sea lampreys
on lake trout. Bence et al. (2003) and Madenjian et al. (2003) found that the higher
temperatures lake trout actually inhabit in October helped to explain the strong pulse of
sea lamprey predation in October documented by Bergstedt and Schneider (1988).
However, these differences do not appear to create large differences in model output.
The predicted consumption of prey by lake trout in a model under development was
reduced by only about 10% at the lower temperatures occupied by FLO ﬁsh (N.
Dobiesz, Michigan State University, personal communication) and Madenjian et al.
(2003) found a 7% increase in lake trout mortality from sea lamprey attack (assuming
no change in sea lamprey growth) when the temperature regimes reported here were
used. Although these differences are not large, they are the result of replacing
generalizations with estimates based on actual thermal histories. Changes in model
output likely reﬂect gains in accuracy, and replacing generalizations with observations

certainly results in increased conﬁdence in the output.

Acknowledgments

I thank Mark Ebener of the Chippewa Ottawa Resource Authority and John
Gauthier of Gauthier and Spaulding Fisheries for their help in obtaining lake trout for
tagging, and Richard Stickel, USGS, for his diligence in the ﬁeld and many useful
suggestions. Charles Bronte, US. Fish and Wildlife Service, provided valuable review

comments and J arnes Bence, Michigan State University, assisted with analyses.

32

References

Anderson, G.W., McKinley, RS, and Colavecchia, M. 1997. The use of clove oil as an
anesthetic for rainbow trout and its effects on swimming performance. N. Am. J.
Fish. Manage. 17: 301-307.

Bence, J .R., Bergstedt, R.A., Christie, G.C, Cochran, P.A., Ebener, M.P., Koonce, J .F.,
Rutter, M.A., and W.D. Swink. 2003. Sea lamprey (Petromyzon marinas)
parasite-host interactions in the Great Lakes. J. Great Lakes Res. 29(Suppl. 1):
253-282.

Bergstedt, R.A., and O’Gorman, R. 1989. Distribution of alewives in southeastern Lake
Ontario in autumn and winter: A clue to winter mortalities. Trans. Am. Fish. Soc.
118: 687-692.

Bergstedt, R.A., and Schneider, GP. 1988. Assessment of sea lamprey (Petromyzon
marinas) predation by recovery of dead lake trout (Salvelinus namaycush) from
Lake Ontario, 1982-85. Can. J. Fish. Aquat. Sci. 45: 1406-1410.

Bergstedt, R.A., and Swink, W.D. 1995. Seasonal growth and duration of the parasitic
life stage of landlocked sea lampreys (Petromyzon marinas). Can. J. Fish. Aquat.
Sci. 52: 1257-1264.

Brett, J .R. 1971. Energetic response of salmon to temperature: a study of some thermal
relations in the physiology and freshwater ecology of sockeye salmon
(Oncorhynchus nerka). Am. Zool. 11: 99-113.

Brown, R.S., Cooke, S.J., Anderson, W.G., and McKinley, RS. 1999. Evidence to
challenge the “2% rule” for biotelemetry. N. Am. J. Fish. Manage. 19: 867-871.

Christie, GO, and Goddard, C]. 2003. Sea Lamprey International Symposium (SLIS
II): advances in the integrated management of sea lampreys in the Great Lakes. J.
Great Lakes Res. 29(Suppl. 1): 1-14.

Cochran, P.A., Hodgson, J.Y., and Kinziger, AP. 2003. Change in energy density of the
sea lamprey (Petromyzon marinas) during its parasitic phase: implications for

modeling food consumption and growth. J. Great Lakes Res. 29(Suppl. 1): 297-306.

Crossman, E. J. 1995. Introduction of the lake trout (Salvelinus namaycush) in areas
outside its native distribution: a review. J. Great Lakes Res. 21(Suppl.1): 17-29.

Ebener, M.P. (Editor). 1998. A lake trout rehabilitation guide for Lake Huron. Great
Lakes Fishery Commission, Ann Arbor, Mich.

33

Elrod, J .H., O’Gorman, R., and Schneider, GP, and Schaner, T. 1996a. Geographical
distributions of lake trout strains stocked in Lake Ontario, 1978-1993. J. Great
Lakes Res. 22: 871-883.

Elrod, J .H., O’Gorman, R., and Schneider, C.P. 1996b. Bathytherrnal distribution,
maturity, and growth of lake trout strains stocked in U. S. waters of Lake Ontario,
1978-1993. J. Great Lakes Res. 22: 722-743.

Eschmeyer, RH, and Phillips, A., Jr. 1965. Fat content of the ﬂesh of siscowets and
lake trout from Lake Superior. Trans. Am. Fish. Soc. 94: 62—74.

Eshenroder, R.L., Payne, N.R., Johnson, J .E., Bowen, C. II, and Ebener, M.P. 1995a.
Lake trout rehabilitation in Lake Huron. J. Great Lakes Res. 21(Suppl.1): 108-127.

Eshenroder, R.L., Crossman, E.J., Meffe, G.K., Olver, CH, and Pister, E.P. 1995b.
Lake trout rehabilitation in the Great Lakes: an evolutionary, ecological, and ethical
perspective. J. Great Lakes Res. 21(Suppl.1): 518-529.

Goddard, CL, and Tait, J .S. 1976. Preferred temperatures of F 3 to F5 hybrids of
Salvelinusfontinalis X S. namaycush. J. Fish. Res. Bd. Canada 33: 197-202.

Ihssen, P., and Tait, J.S. 1974. Genetic differences in retention of swimbladder gas
between two populations of lake trout (Salvelinus namaycush). J. Fish. Res. Bd.
Canada 31:1351-1354.

Kitchell, J .F. 1990. The scope for mortality caused by sea lamprey. Trans. Am. Fish.
Soc. 119: 642-648.

Kitchell, J .F., and Breck, J .E. 1980. Bioenergetics model and foraging hypothesis for
sea lamprey (Petromyzon marinas). Can. J. Fish. Aquat. Sci. 37: 2159-2168.

Koonce, J .F., Eshenroder, R.L., and Christie, GO 1993. An economic injury level
approach to establishing the intensity of sea lamprey control in the Great Lakes. N.
Am. J. Fish. Manage. 13: 1-14.

Krueger, CC, and Ihssen, RE. 1995. Review of genetics of lake trout in the Great
Lakes: History, molecular genetics, physiology, strain comparisons, and restoration
management. J. Great Lakes Res. 21(Suppl.1): 348-363.

Krueger, C.C., Marsden, J .E., Kincaid, H.L., and May, B. 1989. Genetic differentiation

among lake trout strains stocked into Lake Ontario. Trans. Am. Fish. Soc. 118:
317-330.

34

Krueger, C.C., Jones, M.L., and Taylor, W.W. 1995. Restoration of lake trout in the
Great Lakes: challenges and strategies for future management. J. Great Lakes Res.
21(Suppl. 1): 547-558.

Lefrancois, C., Odion, M., Claireaux, G. 2001. An experimental and theoretical analysis
of the effect of added weight on the energetics and hydrostatic function of the
swimbladder of European sea bass (Dicentrarchus labrax). Mar. Biol. 139: 13-17.

Mac., M.J. 1985. Effects of ration size on preferred temperature of lake charr Salvelinus
namaycush. Environ. Biol. Fishes 14: 227-231.

Madenjian, C.P., Cochran, P.A., and Bergstedt, R.A. 2003. Seasonal patterns in growth,
blood consumption, and effects on hosts by parasitic-phase sea lampreys in the
Great Lakes: an individual-based model approach. J. Great Lakes Res. 29(Suppl. 1):
332-346.

Madenjian, C.P., Whittle, D.M., Elrod, J .H., O’Gorman, R., and Owens, R.W. 1995.
Use of a simulation model to reconstruct PCB concentrations in prey of Lake
Ontario lake trout. Environ. Sci. Technol. 29: 2610-2615.

Marsden, J.E., Krueger, C.C., Grewe, P.M., Kincaid, H.L., and May, B. 1993. Genetic
comparison of naturally spawned and artiﬁcially propagated Lake Ontario lake trout
fry: evaluation of a stocking strategy for species rehabilitation. N. Am. J. Fish.
Manage. 13: 304-317.

McCauley, R.W., and Tait, J .S. 1970. Preferred temperature of yearling lake trout
(Salvelinus namaycush). J. Fish. Res. Board. Can 27: 1729-1733.

Page, KC. 2001. Genetic diversity and interrelationships of wild and hatchery lake trout
in the upper Great Lakes: inferences for broodstock management and development
of restoration strategies, MS thesis, Michigan State Univ., East Lansing, MI.

Paetkau, D., Calvert, W., Stirling, I., and Strobeck, C. 1995. Microsatellite analysis of
population structure in Canadian polar bears. Mol. Ecol. 4: 347-354.

Perry, R.W., Adams, N.S., Rondorf, D.W. 2001. Buoyancy compensation of juvenile
chinook salmon implanted with two different size dummy transmitters. Trans. Am.
Fish. Soc. 130: 46-52.

Pritchard, J .K., Stephens, M., and Donnelly, P. 2000. Inference of population structure
using multilocus genotype data. Genetics 155: 945-959.

Ruttner, F. 1963. Fundamentals of limnology. Toronto, Ontario: University of Toronto
Press.

35

SAS Institute, Inc. 2002. Statistical analysis system. Version 8.2. SAS Institute Inc.,
Cary, NC.

Sawyer, A.J. 1980. Prospects for integrated pest management of the sea lamprey
(Petromyzon marinas). Can. J. Fish. Aquat. Sci. 37: 2081-2092.

Schneider, C.P., Owens, R.W., Bergstedt, R.A., and O’Gorman, R. 1996. Predation by
sea lampreys (Petromyzon marinas) on lake trout (Salvelinus namaycush) in
southern Lake Ontario, 1982-92. Can. J. Fish. Aquat. Sci. 53: 1921-1932.

Spangler, G. R., and Berst, AH. 1976. Performance of the lake trout (Salvelinus
namaycush) back crosses, F1 splake (S. F ontinalis x S. namaycush), and lake trout
in Lake Huron. J. Fish. Res. Bd. Canada. 33: 2402-2407.

Stewart, D.J., Weininger, D., Rottiers, D.V., and Edsall, TA. 1983. An energetics
model for lake trout, Salvelinus namaycush: Application to the Lake Michigan
population. Can. J. Fish. Aquat. Sci. 40: 681-698.

Swink, W.D., and Hanson, L.H. 1989. Survival of rainbow trout and lake trout after sea
lamprey attack. N. Am. J. Fish. Manage. 9: 35-40.

Tait, J .S. 1970. A method of selecting trout hybrids (Salvelinusfontinalis X S.
namaycush) for ability to retain swimbladder gas. J. Fish. Res. Bd. Canada 27:
39-45.

Thorstad, E.B., Okland, F., and Finstad, B. 2000. Effects of telemetry transmitters on
swimming performance of adult Atlantic salmon. J. Fish Biol. 57: 531-535.

Topchy, A., Scribner, K.T., Punch, W., and Guinand, B. In review. Conﬁdence
estimation for individual assignment using multi-locus genotypic data. Conserv.
Genet.

Welch, D.W., Chigirinsky, A.I., and lshida, Y. 1995. Upper thermal limits on the
oceanic distribution of Paciﬁc salmon (Oncorhynchus spp.) in the spring. Can. J.
Fish. Aquat. Sci. 52: 489-503.

Wells, L. 1968. Seasonal depth distribution of ﬁsh in southeastern Lake Michigan. U. S.
Fish. Wildl. Serv. Fish. Bull. 67: 1-15.

Winter, J .D. 1996. Advances in underwater biotelemetry. In Fisheries techniques.

Edited by ER. Murphy and D.W. Willis. American Fisheries Society, Bethesda,
Md. pp. 555-590.

36

CHAPTER TWO

Changes in bathythermal habitat use by lake trout in Lake Huron during a period when

the prey base changed rapidly

37

Abstract

Before development of small archival (or data-recording) telemetry tags, knowledge
of temperatures and depths of lake trout (Salvelinus namaycush) in the Great Lakes
came from point estimates of abundance within the zones sampled with net gear. An
initial archival tag study recording temperatures occupied by lake trout in Lake Huron
during 1999 and 2000 (Chapter One) showed that lake trout strains of Great Lakes
origin (GLO) occupied signiﬁcantly higher temperatures than strains of Finger Lakes,
New York, origin (FLO). The objectives of this study (Chapter Two) were to compare
the temperatures recorded on tags recovered from 2002 through 2005 to those reported
in 1999 and 2000, to examine whether lake trout of Finger Lakes, New York origin
occupy both lower temperatures and greater depths than lake trout of Great Lakes
origin, and to examine any changes in the temperatures occupied in light of changes in
the abundance and composition of their prey. Archival tags recording depth and
temperature were surgically implanted in 447 lake trout in Lake Huron in 2002 and
2003. Through 2005, 93 records of depths and temperatures occupied by GLO lake
trout (38) and FLO lake trout (55) were recovered (21% of the tags deployed).
Temperatures occupied by GLO lake trout were signiﬁcantly higher than those
occupied by FLO lake trout, supporting the conclusion of the ﬁrst study. The GLO lake
trout occupied signiﬁcantly shallower depths than FLO lake trout, averaging 4.5 m
shallower (range 2.3 to 8.3) across 24 half-month intervals. In comparison to the ﬁrst
study, both GLO and FLO lake trout occupied signiﬁcantly lower temperatures; from
May through August, GLO lake trout averaged 1.7°C lower temperatures and from May

through early November FLO lake trout 1.6°C lower than in 1999-2000. Over the same

38

period, the prey ﬁsh community changed dramatically. The formerly abundant alewife
(Alosa pseudoharengus), a key pelagic prey item, virtually disappeared and a new
demersal species, the round goby (Neogobius melanostomus), invaded Lake Huron.
Prey biomass also decreased by 90% during this period. Switching to more demersal
prey and retreating to temperatures lower than optimum for growth in response to food
scarcity could increase conversion efﬁciency could explain the movement toward lower

temperatures.

39

Introduction

Temperature is an important environmental variable that governs food consumption
and growth, and hence affects the distribution of ﬁshes in lakes (Fry 1947).
Temperature selection (Goddard and Tait 1976) and the ability to maintain position at
different depths (Ihssen and Tait 1974) are known to be heritable characteristics in lake
trout (Salvelinus namaycush). The combined expression of these multi-genic traits
among ﬁsh species during summer periods of thermal stratiﬁcation, along with other
environmental variables (Fry 1947), yields the temperature and depth distributions of
predators and their prey. Numerous laboratory studies have contributed to our
understanding of the thermal behavior of lake trout, including preferred temperatures
(e. g., McCauley and Tait 1970; Peterson et al. 1979; Edsall and Cleland 2000),
optimum temperatures for growth (e. g., Elliot and Hurley 1999; Edsall and Cleland
2000), and temperature tolerance or lethal limits (e. g., Ihssen1973; Grande and
Andersen 1991). The temperature relationships described by these studies tend to be
consistent among each other (J obling 1981), which suggests that a component of
temperature selection is genetic and deﬁnes the fundamental niche axis as deﬁned by
Hutchinson (1957). However, description of a fundamental niche for temperature
considers only one of a number of environmental variables affecting the distribution of
ﬁsh; thus, a species is not necessarily found at temperatures indicated in the laboratory
(Magnuson et al. 1979). Describing the realized thermal niche as recorded by ﬁeld
observations, and which incorporate these other variables, will be of greater interest to
ﬁshery ecologists (Magnuson et al. 1979). As examples, Eaton et al. (1995) and Huff et

al. (2005) used this approach to describe temperature tolerance. Huff et al. (2005) also

40

described the center and width of the realized niche. Little is known about the depths
and temperatures occupied by lake trout on a daily and seasonal basis in the Laurentian
Great Lakes. That knowledge is deduced from captures in different ﬁshing gears. These
capture data only provide point estimates of abundance within the depths and
temperatures sampled (e. g., Elrod et al. 1996), and suffer from the sampling bias of
each gear. Observations of temperature at capture are also limited to where gear is set,

and do not necessarily reﬂect the range of habitats occupied.

During 1999 and 2000, archival tags implanted in lake trout of two genetic strains
ﬁrst enabled monitoring of the environmental conditions experienced by deep-ranging
ﬁsh in the Great Lakes and provided the ﬁrst reasonably continuous (75-minute
intervals) seasonal measures of the temperatures occupied in Lake Huron (Chapter
One). Although that information contributed substantially to our knowledge of lake
trout behavior in the Great Lakes, the data were collected with tags that only measured
temperature but not depth. The lack of depth data was problematic in that it was not
possible to directly conclude that the higher temperature occupied by Great Lakes origin
lake trout, compared to Finger Lakes origin lake trout, was due to the use of lesser
depths. Uncertainty existed because the average therrnocline depth can be affected by
prevailing winds and bathyrnetry, and can vary geographically within one of the Great
Lakes. Therefore, the temperature difference could have been due to a similar depth
distribution but with differences in geographic distribution between strains (Chapter

One).

41

Subsequent to this ﬁrst study (Chapter One), substantial progress occurred in
miniaturization of electronics that made it possible to implant a smaller archival tag that
recorded both depth and temperature. By 2001, an affordable tag became available that
would measure both temperature and depth--and that also had substantially more
memory. This newer tag could be implanted in lake trout and used to assess the
overlap of both temperature and depth distributions among strains of lake trout and to

compare with results from 1999-2000 (Chapter One).

Since the temperature distribution of lake trout reported in Chapter One (1999-
2000) were collected, substantial changes have taken place in the food web in Lake
Huron. Phosphorus reductions from pollution abatement (Nalepa et al. 2007), along
with the probable effect of the zebra mussel (Dreissena polymorpha) and quagga
mussel (Dreissena bugensis) colonizing Lake Huron (N alepa et al. 2007; Hecky et al.
2004) have likely reduced pelagic plankton production and also food for benthos.
Nalepa et al. (2007) reported that the densities of four key taxa (Diporeia, Oligochaeta,
Sphaeriidae, and Chironomidae) had not changed substantially between 1970 and 2000,
but then decreased signiﬁcantly between 2000 and 2003. The decrease was greatest for
Diporeia and Sphaeriidae, which dropped by 57 and 74% lakewide. The abundance of
most prey ﬁsh species in Lake Huron also declined sharply in the early 20008 (Riley et
al. in press), with alewives (Alosa pseudoharengus) virtually disappearing between
2003 and 2004 (Warner et al. 2005). The distribution of predators such as lake trout in
Lake Huron would potentially shift after the loss of a key pelagic prey species, such as

alewife.

42

The combination of the data presented in Chapter One and here bracket this period
of sweeping changes in the biota of Lake Huron. The data collected provide an
Opportunity to examine the temperatures occupied by lake trout relative to a substantial
shift in the prey-ﬁsh community, including the virtual disappearance of alewives
between 2003 and 2004 and the appearance of a new invasive species, the round goby
(Neogobius melanostomus). The objectives of this chapter were to compare the
temperatures recorded on tags recovered from 2002 through 2005 to those reported in
Chapter One (1999-2000), to examine whether lake trout of Finger Lakes, New York
origin occupy both lower temperatures and greater depths than lake trout of Great Lakes
origin, and to examine any changes in the temperatures occupied in light of changes in

the abundance and composition of their prey.

Methods
Tags and tagging

In 2002 and 2003, LTD-1110 archival tags manufactured by Lotek Wireless Inc.,
Newmarket, Ontario, Canada, were surgically implanted in lake trout released to Lake
Huron. Each tag recorded temperature (0 - 25 °C) and depth (0-500 m). With a battery
life of two to three years, those tags provided temporal coverage from 2002 through
2005. The tags were cylindrical—1 1 mm in diameter by 32 mm long and rounded on
one end. Weight was about 5 g in air and 2 g in water. Recording continued until tag

recovery and download, or until the battery expired. Observations of date, time,

43

temperature, and depth were initially recorded at lS-second intervals until the memory
was full. Software then replaced previous observations to maintain a complete set of
observations from initialization to download, but with the intervals between
observations gradually increasing. About 33,000 observations were recorded, with
average intervals increasing from 15 seconds by about 1.3 minutes for each month aﬂer

deployment and by 16 minutes for each year.

Tag speciﬁcations were 0.2°C resolution with :1: 0.3 °C accuracy for temperature and
0.4% of the current depth scale with :l:1% accuracy. The potential depth range was 0 to
500 Hi, but, to improve resolution and accuracy, software initially set the range to one-
quarter of that scale (0 to 125 m). If ﬁsh moved deeper than the initial scale, the range
would progressively shift (permanently) to one-half scale (0 to 250 m) and then full
scale (0 to 500 m). Resolution and accuracy (in parentheses) at these scales were 0.5 (i

1.25) m, 1.0 (i 2.5) m, and 2.0 (i 5.0) m, respectively.

Recovery of archival tags depended on the recapture of ﬁsh by sport and
commercial ﬁsheries, and return of the tags. To promote returns, a reward of $100
(U SD) was offered. Tagged ﬁsh were also marked externally with orange dart tags.
The external tags bore identifying serial numbers, reward advertisement, a contact
telephone number, and instructions to not freeze the tag. The internal tags also bore a
reward notice and contact information. Information posters describing the study and

reward were displayed in tackle and convenience stores, at launch sites, and were

44

provided to commercial and charter ﬁsheries. The Michigan Department of Natural

Resources and the Ontario Ministry of Natural Resources also publicized the study.

Table 4. Summary of lake trout tagged with surgically implanted archival
temperature tags and released into Lake Huron, either at the Hammond Bay
Biological Station (HBBS), at Drummond Island (D1), or on Lake Huron (LH),
the number of lake trout tagged by dates, location, and source; the number
recaptured; and the number and percent of ﬁsh used in the analyses.

 

Re— Source No.
lease (gill or No. No. % in % in
loca- trap re- recap- recap- ana- ana-
Dates tagged tion net) leased tured tured lyses lyses

 

15-30 May 2002 HBBS trap 77 21 27 15 19

18-20 Jun. 2002 HBBS trap 107 l7 16 15 14

9-10 Oct. 2002 DI gill 38 5 l3 3 8
Subtotal 222 43 19 33 l 5

07-28 May 2003 LH trap 64 24 34 20 28
15-16 May 2003 HBBS trap 55 17 31 15 27
5-17 Jun. 2003 HBBS trap 106 28 27 25 24
Subtotal 225 69 31 60 27

Total 447 l l l 25 93 2 l

 

All lake trout tagged were obtained from commercial trap nets ﬁshed during May
and June of 2002 and 2003 along the Michigan shore of Lake Huron between Rockport
and Forty Mile Point—with the exception of 37 lake trout from the Chippewa-Ottawa
Resource Authority (CORA) assessment gill nets at Drummond Island, Lake Huron in
October 2002 (Table 4; Fig. 4). Lake trout from gill nets were transported to shore in

aerated containers and placed in 1000-L tanks supplied with aerated, arnbient-

45

temperature Lake Huron water. Lake trout from trap nets were placed in a ﬂoating pen
and either tagged and released at the capture site or transferred in aerated containers—
either directly by boat to the Hammond bay Biological Station (HBBS) or indirectly by
boat to shore and then by truck to the HBBS. At the HBBS, they were held in raceways

or in 1000-L tanks in aerated, arnbient-temperature Lake Huron water.

Fig. 4. Locations in Lake Huron of Drummond Island, the Hammond Bay Biological
Station, Forty-mile Point, Rockport, sites where lake trout were tagged at the site of
collection (squares), and boundaries of ﬁshery statistical districts MI-I-l to MH6.

 

Drummond Island

North Channel

 

 

 

 

Lake trout were anesthetized, weighed, measured, ﬁn clips recorded, and a small
piece of tissue taken from the upper lobe of the caudal ﬁn for genotyping, as described

in Chapter One. Archival tags were inserted through 1.5-cm incisions made slightly to

46

the side of the ventral midline anterior to the pelvic ﬁns, the incisions closed with two
monoﬁlament sutures, 3-M Vet-Bond® adhesive applied to the knots, and a Floy tag
inserted on each side of the dorsal ﬁn. The procedure took about 3 minutes, during
which the gills were continually irrigated. The ﬁsh were then transferred to a recovery
tank or ﬂoating pen, where they regained equilibrium within 5 to 10 minutes. If tagging
was done in the ﬁeld, weights were not taken and they were immediately released. If
tagged at the HBBS, they were held from 24 to 48 hours for observation before release

from shore at the station.

Genetic origins of lake trout

Lake trout used in this study were shallow water, or lean, lake trout of hatchery
origin. A number of genetic strains have been stocked into Lake Huron, but all have
origins either in Lakes Superior or Michigan or in the Finger Lakes of New York. No
difference was found in temperatures occupied by strains derived from Lake Superior or
Lake Michigan stocks, and therefore they were collectively designated as Great Lakes
origin (GLO; Chapter One). All New York hatchery strains were designated as Finger
Lakes origin (FLO). More detailed information on the sources of GLO and FLO lake

trout is provided in Chapter One, Krueger et al. (1989), and Elrod et al. (1996).

Other than ﬁsh bearing a CWT, strain of a lake trout used in this study could not be
determined directly from ﬁn clips because marks were only year-class speciﬁc and all
strains were marked identically. The geneticstrains of ﬁsh in this study were therefore

identiﬁed from coded wire tags (CWTS) aﬂer recapture or via a combination of genetic

47

analysis of tissue samples and ﬁn clips (clips were used to narrow the set of potential

strains).

Precautions were taken to equalize the releases and hence returns ﬁom the GLO and
FLO groups, because of previous knowledge that temperatures and depths would likely
differ with genetic origin (Chapter One). One strategy used was to tag ﬁsh with certain
ﬁn clips based on previous knowledge of the likely strain compositions. Fin clips and
relative numbers to tag were chosen based on stocking records and previous genotyping
(Chapter One). On 15-30 May and 8 October 2002, 114 lake trout were selected based
on ﬁn clips and tagged and released in the ﬁeld. A second strategy used in 2002 relied
on genotyping. Lake trout were transported to the HBBS, marked with the external
Floy tags, and tissue samples were taken. Tissues from each ﬁsh were genotyped (Page
2001). Fish were held for about three weeks while genetic analyses were performed
and strain assignments made. The results were used to choose another 107 ﬁsh for
tagging. In 2003, information from the 2002 genotyping were combined with
information ﬁom ﬁn clips and coded wire tag data from Michigan Department of
Natural Resources lake trout surveys in 2002 (Ji He, Alpena, Michigan Fisheries
Research Station, personal communication). This information was used to choose

which ﬁn clipped lake trout to tag.

48

Statistical analyses

Temperature and depth data from the tags provide an array of possibilities for their
summarization. These data were summarized by half-month seasonal intervals (similar
to Chapter One) -- an interval selected in consultation with potential users of the data
for bioenergetics models of prey consumption (Charles Madenjian, USGS, Great Lakes
Science Center; James Bence, Michigan State University; J i He, Michigan Department
of natural Resources, personal communication). The ﬁrst half of each month included
the ﬁrst 15 days and the second half the remaining days in the month. Because these
data were likely to be used in bioenergetics models of growth or prey consumption, the
arithmetic mean temperature (°C) in each half-month interval was used to best represent
the average temperature experienced by each ﬁsh; this summary statistic, when
expanded to the interval of interest, most accurately represents the number of degree-
days experienced. For depth, the median depth occupied by each ﬁsh was used in each
interval. With each ﬁsh, there was a period after release, during recovery from surgery,
when the data were suspected to not represent normal behavior. Each plot of
temperature was inspected and a subjective judgment was made as to the date when
behavior appeared consistent with the remainder of the data. For ﬁsh captured when the
tag was still actively recording, time of capture in the ﬁshing gear was evident from a
cessation of temperature and depth variation. Temperature observations before
behavior appeared normal and after the ﬁsh was caught were not used in the analyses.
Because of the expanding interval of observations over time and the greater number of
observations per hour for ﬁsh recaptured sooner after release, and to give ﬁsh equal

weight (the experimental unit used in this study was the individual ﬁsh), subsequent

49

analyses were based only on the mean temperatures and median depths for individual
ﬁsh by half-month interval and considering the repeated nature of those measures.
Mean temperature was chosen for the reason stated above, and their likelihood of use in
bioenergetics models. The median depth was chosen as best representing the most

common depth frequented.

The temperatures analyzed and reported in this study were only those recorded
inside the abdominal cavity. At any instant, these could vary from the temperature of
the surrounding water. Based on laboratory studies (HBBS, unpublished data), internal
temperature after a rapid change in external temperature would move 50% toward the
new external temperature within about 5 minutes, and 90% toward the new temperature
within 15 minutes. Over the temporal scale of a half-month interval, the mean internal
temperature should closely approximate the average temperature of the water occupied
by a ﬁsh. To simplify, these mean temperature data are hereafter referred to as
“occupied” temperatures, referring to both the approximation of a half-month average
temperature occupied and the fact that these measures do not necessarily reﬂect a
“preference,” but a realized niche as deﬁned by Magnuson et al. (1979). The
commonly used term “preference” only has meaning in the context of laboratory
studies, where all other dimensions of the ecological niche are controlled. In this study,
where temperature and depth are simultaneously measured, it is not possible to know
along which of those niche axes a preference is being exercised or whether the

controlling variable is along another axis.

50

To describe the variability in the observations, percent frequency distributions for
temperature and depth were compiled by genetic origin. Because the time between
release and recapture varied with individual ﬁsh, and because of the gradually
increasing intervals between observations, and hence the changing number of
observations per time interval, it was necessary to create a summary statistic to ensure
that all ﬁsh contributed equally to the distribution. Therefore, the average temperature
and median depth were calculated for combinations of genetic origin, individual ﬁsh,
half-month interval, and hour and used to describe those statistics along with the 10th,

25m, 75m, and 90th percentiles by half-month interval.

The overall effect of genetic origin on the temperatures and depths occupied was
evaluated with a mixed-model ANOVA (SAS 2001). The null hypothesis was that no
difference existed between genetic origin of lake trout in temperatures and depths

occupied. The underlying model for the analysis was

71h,s,y,f 0rDh,s,y,f : :u +ah +ﬂs +7h,s +6)”: +8h,s,y,f (l)

in which T h, S, y f or thwf is a half-month mean for temperature or depth and h, s, y,

and f denote the half-month interval, genetic origin, year, and individual ﬁsh. The ﬁxed
effects were the overall mean (u), the seasonal or half-month adjustment (a), the genetic

origin adjustment (,6), and the interaction between half-month interval and genetic

51

origin (y). Potentially, the random components are a year by half-month variation (5)
and residual error (8). Ten versions of the model were ﬁt, testing each of ﬁve
treatments of the random effects versus two treatments of the repeated aspect of the
data. The repeated nature of the data was treated either as a ﬁrst-order autoregressive
structure or as a ﬁrst-order autoregressive structure with the subjects identiﬁed as
individual ﬁsh within genetic origin. For each repeated treatment, the random
component (5) was ﬁrst treated as a ﬁxed effect, second treated as a random effect, third
treated as a random effect with a ﬁrst order autocorrelation, fourth treated as a random
effect allowed to differ between genetic origins, and ﬁfth treated as a random effect
allowed to differ between genetic origins and with a ﬁrst order autocorrelation. These
models were ﬁt for depth and temperature and the Akaike Information Criterion (AIC)
used to select the best model for each. The least-squares means (LSMs) and conﬁdence
intervals from the selected models were used to describe the depths and temperatures

occupied by lake trout of each genetic origin in each seasonal, half-month interval.

The same procedure was used for model selection to compare temperatures
occupied by lake trout of each genetic origin between 1999 and 2001 (Chapter One) to
those occupied between 2002 and 2005 (current study). The null hypotheses were that
for each genetic origin no difference in temperature existed between the two studies.
The two intervals of years (studies) were substituted for genetic origin in the model
above, with the analyses performed separately by genetic origin. Model selection was
performed as described above, except that subjects were identiﬁed as individual ﬁsh

rather than ﬁsh within origin. Because the tags used during 1999-2000 did not record

52

depth, the comparative analysis was only between occupied temperatures. Because
these datasets were the ﬁrst of their type for lake trout in the Great Lakes and to
minimize type-2 errors, significance was evaluated at 01 = 0.1. Where post-hoe t-tests
were made comparing LSMs in each half-month interval, the intent was to describe the
relative signiﬁcance and pattern of differences throughout the year and Bonferroni

corrections were not made.

Differences in recapture rates between genetic origins and tagging years were
evaluated using Fisher’s exact test. The deviation of numbers of GLO and FLO lake
trout recaptures from the planned equal numbers to be recovered was evaluated with a

Chi-square test assuming the expectation of equal proportions.

Model selection

Comparisons between genetic origins

In the mixed-model ANOVA comparing depths and temperatures occupied by genetic
origin during 2002-2005, the best treatment of the repeated nature of the data was a
ﬁrst-order, autoregressive structure that was allowed to differ for ﬁsh within genetic
origin (Table 5). For both temperature and depth, the best treatment of the random
effect was treating the coefﬁcient for the year by half-month variation (5) as a random

effect, and allowing it to differ between genetic origins (Table 5).

53

Table 5. Results of model selection for a mixed-model ANOVA predicting the least-
squares mean temperature and depth occupied by lake trout of two genetic origins in
Lake Huron during 2002-2005. The AIC values are for two potential treatments of the
repeated nature of the observations crossed with ﬁve potential treatments of the random

components. The A AIC values are based on the lowest AIC values (in bold) for

analyses of temperature and depth. In one case the model did not converge (DNC).

 

 

Repeated statement Random statement AIC A AIC
Temperature
Type=ar(l) None 8138.7 1194.2
bimonth*year 8088.7 1 144.2
bimonth*year/ type=ar(l) 8090.7 1 146.2
bimonth*year/ group=ori gin 8042.7 1098.2
bimonth*year/ type=ar(l)
group=origin 8045.1 1 100.6
Type=ar(l ),
sub=ﬁsh(ori gin) None 6984.0 39.5
bimonth*year 6954.0 9.5
bimonth*year/ type=ar(l) 6956.0 1 1 .5
bimonth*year/ group=ori gin 6944.5 0.0
bimonth*year/ type=ar(l)
group=ori gin 6947.2 2.7
Depth
Type=ar( 1) None 18092.5 2945.5
bimonth* year 1 8070.2 2923 .2
bimonth*year/ type=ar(l ) 1 8072.2 2925 .2
bimonth*year/ group=ori gin 1 8071 .3 2924.3
bimonth*year/ type=ar(l)
group=ori gin 18075.1 2928.1
Type=ar(! ),
sub=ﬁsh(ori gin) None 15150.40 3 .4
bimonth*year 15149.4 2.4
bimonth*year/ type=ar(1) DNC -
bimonth*year/ group=ori gin 15147.0 0.0
bimonth*year/ type=ar(l)

 

 

54

Comparisons between year intervals 1999-2000 and 2001-2005

The mixed-model ANOVA options were evaluated separately for the GLO and FLO
lake trout. For both genetic origins, the best treatment of the repeated nature of the data
was a ﬁrst-order, autoregressive structure that was allowed to differ within year
intervals (Table 6). The best treatment of the random effect for GLO ﬁsh was treating
the coefﬁcient for the year by half-month variation (5) as a random effect and allowing
it to differ between year intervals. For the FLO ﬁsh, the random component providing
the best ﬁt was a ﬁrst-order, autoregressive structure without separate coefﬁcients for

the two year intervals.

Results

Tag returns

Support from the commercial and sport ﬁsheries in returning tags was strong, with
25% (111 tags returned of 447 released) returned through the end of 2005 (Table 4).
Returns by tagging year were 43 of 222 tags deployed in 2002 (19%) and 69 of 225
(31%) deployed in 2003. The return rate for ﬁsh tagged in 2003 was signiﬁcantly
higher (Fisher’s exact test, P = 0.035). Data from 18 of the 111 returned tags could not
be used due to tag loss (1 ﬁsh), due to tag malfunction and corrupted data (4 ﬁsh),
recovery too soon after release for behavior to be representative (9 ﬁsh), or our inability
to assign strain and genetic origin (4 ﬁsh). Analyses were therefore based on data from
93 ﬁsh, of which 38 were the GLO strain and 55 ﬁsh were the PLO strain. The lower

return of GLO ﬁsh was different than the expectation of 46.5 (Chi-Square = 3.12, df =1 ,

55

P =0.078), but likely due to the imprecision in predicting strain from ﬁn clips and

genotyping.

Table 6. Results of model selection mixed-model ANOVAS of temperatures occupied
in Lake Huron during two studies, 1999-2000 and 2001-2005, run separately for Great
Lakes and Finger Lakes origin lake trout. The AIC values are given for two potential
treatments of the repeated nature of the observations and ﬁve potential treatments of
the random components. The A AIC values are based on the lowest AIC values for
each genetic origin (in bold). In one case the model did not converge (DNC).

 

 

 

 

 

A
Repeated statement Random statement AIC AIC
Great Lakes origin
Type=ar(1) None 3505 .3 526.5
bimonth*year 3464.4 485.6
bimonth*year/ type=ar(l ) 3463 .3 484.5
bimonth*year/ group=study 3446.5 467.7
bimonth*year/ type=ar(1) group=study 3448.0 469.2
Type=ar(1) sub=ﬁsh None 2992.6 13.8
bimonth*year 2986.9 8.1
bimonth*year/ type=ar(1) 2987.8 9.0
bimonth*year/ group=study 2978.8 0
bimonth*year/ type=ar(l) group=study 2981.1 2.3
Finger Lakes origin

Type=ar(1) None 6270.5 830.6
bimonth*year 6236.8 796.9
bimonth*year/ type=ar(1) DNC -
bimonth*year/ group=study 6236.9 797.0
bimonth*year/ type=ar(l) group=study DNC -
Type=ar(1) sub=ﬁsh None 5473.9 34.0
bimonth*year 5442.5 2.6
bimonth*year/ type=ar(1) 5439.9 0
bimonth*year/ group= study 5443.6 3.7
bimonth*year/ type=ar(l) group=study 5443.3 3.4

j

56

 

Individual variability

Variation in hourly temperatures occupied among individual lake trout within half-
month intervals was substantial overall and within strains (Fig. 5). The overall
median hourly mean temperature for GLO ﬁsh was 5.3 °C. The range for the monthly
medians was 8.8 °C—from a minimum of 0 °C in February and March (5.3 °C or 97%
colder than the overall median) to 8.9 °C in October (3.6 °C or 68% warmer). The
range in hourly mean temperatures observed for GLO ﬁsh in individual months (5th to
95th percentiles) was greatest in July, ranging from 4.2 to 11.6 °C and least in March,
ranging from 0.0 to 1.4 0C. For FLO ﬁsh, the overall median hourly mean
temperature was 4.8 °C. The range for the monthly medians

was 7.9 °C———from a minimum of 0.3 °C in February (4.6 °C or 94% colder than the
overall median) to 8.2 0C in October (3.3 °C or 69% warmer). The range in hourly
mean temperatures observed for FLO ﬁsh in individual months was greatest in July,

ranging from 4.5 to 9.7 0C and least in March, ranging from 0.0 to 1.5 °C.

Substantial variation in the depths occupied existed among individual lake trout
within strains (Fig. 5). However, seasonal median depths did not vary proportionally
as much as the median temperatures. The overall median hourly mean depth for GLO
ﬁsh was 34 m. The range for the monthly medians was 13 m—from a minimum of 27
m in July (7 m or 22% shallower, than the overall median) to 40 min March (5 m or
16% deeper). The range in hourly mean depths observed within months (5th to 95th
percentiles) was greatest in February, ranging from 8 to 85 m and least in May,

ranging from 20 to 52 m. For FLO ﬁsh, the overall median hourly mean depth

57

Fig. 5. Frequency of hourly mean temperatures and depths occupied by individual ﬁsh
of Great Lakes (open boxes and circles) and Finger lakes, NY, (ﬁlled boxes and circles)
origin by month during 2002-2005. Horizontal lines are the median, boxes the 25th and
75th percentiles, vertical lines the 10th and 90th percentiles, and circles the 5th and 95th
percentiles. The dotted horizontal line shows the ﬁnal preferendum and the dashed line
the temperature for optimum growth for lake trout, as summarized by Christie and
Regier (1988).

 

........................................................

Temperature C)
-h 0) co 8
O
O
O
O 41:23—20
c-czzr— .0
o ——:=—O
O
O
O
om—c
o-—1:_—_1—o
o—m—O
O-ED-O
c-m—o
E—o
O—GZB-O
1 a I J

 

 

N
4 L
m—o
83°
_ —D-

 

 

 

 

 

 

 

0- I I . I I
J F M A M J J A S O N D
0
o o. 0 o
oo
0. o c
20- 0' ° -
5,40- ii _
5'1
*5. o. o 00 o
d) 601 o -
D o . o
o o. O
80‘ -
i o
100 I T I I I j I I I I I I I
J F M A M J J A S O N D
Month

was 38 m. The range for the monthly medians was 11 m—from a minimum of 32 m in
June (6 m or 14% shallower) to 43 m in February (6 m or 15% deeper). The range in
hourly mean depths observed for FLO ﬁsh was greatest in January, ranging from 11 to

75 m and least in September, ranging from 13 to 55 m.

58

Temperatures occupied

Overall, GLO lake trout occupied signiﬁcantly higher temperatures (mixed-model
ANOVA; F [1.59] = 9.22; P = 0.004) than FLO lake trout (Table 7; Fig. 6). Temperatures
occupied differed signiﬁcantly among seasonal periods, but more importantly, the
interaction between genetic origin and seasonal period was not signiﬁcant (mixed-
mOdel ANOVA; F9117] = 0.80; P = 0.696). Together, these results indicate that the
overall pattern of seasonal temperatures is similar between genetic origins, but with

GLO ﬁsh occupying higher temperatures during periods of stratiﬁcation.

Table 7. Results of a mixed-model ANOVA examining
factors inﬂuencing the mean temperature (Temp) and median
depth (Depth) occupied in Lake Huron during 2002-2005 by
lake trout of two genetic origins—Finger Lakes, NY and the
upper Great Lakes. Data were recorded with archival tags.
The main effects were: Origin, Period (half-month seasonal
periods), and the interaction of Origin and Period.

 

Dependent Fixed

 

variable effects df F P > F
Temp Origin 1, 59.3 9.22 0.0035
Period 23, 16.8 32.91 <0.0001

Origin*Period 23, 16.8 0.80 0.6960

Depth Origin 1, 167 3.14 0.078
Period 23, 34 3.90 0.000
Origin*Period 23, 34 0.51 0.953

 

From the onset of autumnal mixing in Lake Huron through the onset of summer
stratiﬁcation, temperatures occupied varied little between genetic origins (Fig. 6a).

Between those events, temperature varies little within the upper 100 m (see Bergstedt

59

and O’Gorman 1989), and choice is limited. As a consequence, little difference can
exist between genetic origins with the temperature following the warm-up and cool-
down of the lake. During stratiﬁcation, the temperatures occupied by GLO ﬁsh were
consistently higher. Within speciﬁc half-month periods, the LSMs differed the greatest
in early July and in late August to late September (Tukey t-test, P<0.l), although the
probability of a greater t in late July and early August were nearly as small (Table 8).
Based on the least-squares means for temperatures expanded in each interval, the Great
Lakes origin ﬁsh would annually experience 1,839 degree days versus 1,593 for the

FLO ﬁsh—about 15 percent greater.

Depths occupied

Median depths occupied were signiﬁcantly different between genetic origins across
the entire data set (mixed-model ANOVA; me] = 3.14; P = 0.078). The LSMs for
the median depths occupied were consistently lower for GLO than for FLO lake trout
(Fig. 7). Within speciﬁc half-month periods, the LSMs differed signiﬁcantly between
genetic origins (Tukey t-test, P<0. 1) in early June through early July, and in late
September through early October (Table 9). The minimum difference between genetic
origins was 2.3 m in late summer and the greatest difference was 8.3 m in early

October.

Comparison of temperatures occupied in 1999-2000 versus 2002-2005
The GLO lake trout occupied signiﬁcantly lower temperatures in 2002-2005

compared to 1999-2000 (Table 10; mixed-model ANOVA; F [1,107] = 5.77; P = 0.018).

60

Fig. 6. Seasonal internal temperatures by half-month intervals for tagged lake trout of
Upper Great Lakes origin (GLO) and of Finger Lakes, New York origin (FLO) at large
in Lake Huron as recorded with implanted archival tags. Symbols and 95% conﬁdence
intervals are least-squares means ﬁom a mixed-model AN OVA treating the origin*half-
month interval interaction as a random effect and half-month means for individual ﬁsh

as repeated measures. The data were collected from May 2002 to September 2005.
Data from 1999-2000 are from Chapter One. Panels 3 and b contrast GLO and FLO
lake trout within each data-collection interval and panels c and d contrast the data-
collection intervals for each genetic origin.

 

12

 

  

 

 

 

 

 

 

  

 

a. 2002-2006 c. GLO ’ ‘ A
10- ’ - - ' g ‘ °..
8 8: GL0\ : 1999-2000 .-‘ j * '
e I]. ‘ ~11" °
3 1- [I ‘ : . \. ‘
E 6‘ ’ '0' "l . . - . .‘
m ‘3 l i 7 _
°- ° 2002-2006 .\
g 4‘ o/ " ‘ 3 . I:
l" 1 ./ '. j. ' ' '
2-°\ " f ’l
‘ “an.' , iotwaﬂ‘ ' .
0 1"1' 1’ r 1 1 1 r 1 r m T 1 I r r 1 r I 1 I 1
12 T I I I I I I I I I I I I I I I I I I I I I
b. 1999 - 2000 , ‘ d. FLO
10‘ )0 '1 d .
G GL0 \3- \, J 1999-2000
7"” 8* /° ° 1!. ' l 1 J ‘ 3 : '
5 ° 0 , ‘_ i ..
t9 6- ° \ '\ ~ ~
é ‘ FLO 4 .
m 4- o _
1— , 4
24 - - ~° . . '
' I
d o a"... ./. I "o -. .t’ol’.
0 I I I I I I I I I I Iw *I I I I I I I I I I I
JFMAMJJASONDJFMAMJJASOND
Month Month

61

 

Table 8. Least-square means (LSM) of individual ﬁsh mean temperatures (°C)
occupied and the degree-days (DD) experienced by Great Lakes and Finger Lakes
origin lake trout in Lake Huron by half-month interval from a mixed-model ANOVA
and results of post-hoe comparisons (Tukey t—tests) between origins within each
interval

 

 

 

 

Half- Great Lakes Finger Lakes
month Origin Origin

interval LSM DD LSM DD A LSM df t P > | 11
Jan 2.6 38.8 2.1 32.1 -0.4 21 -0.51 0.614
1.2 18.6 1.0 16.2 -0.2 21 -0.18 0.861
Feb 0.4 6.1 0.6 9.4 0.2 21 0.25 0.804
0.3 4.3 0.5 6.7 0.2 21 0.21 0.837
Mar 0.4 6.4 0.4 6.6 0.0 21 0.02 0.983
0.6 10.0 0.7 10.9 0.1 21 0.06 0.952
Apr 1.3 19.7 1.3 19.0 0.0 21 -0.05 0.960
2.6 38.7 2.6 38.7 0.0 21 0.00 0.999
May 3.7 54.8 3.8 57.7 0.2 20 0.23 0.822
5.1 81.2 4.8 76.8 -03 22 -O.38 0.708
Jun 5.8 87.4 5.3 79.5 -0.5 20 -0.76 0.459
7.5 112.9 6.4 96.6 -1.1 19 -1.57 0.134
Jul 8.5 127.5 6.8 102.6 -1.7 19 -2.40 0.027
7.5 119.8 6.4 101.7 -1.1 19 -1.63 0.120
Aug 7.1 106.8 6.0 89.9 -1 .1 20 -1.61 0.123
7.7 122.4 6.1 97.8 -15 21 -217 0.042
Sep 8.4 125.7 6.5 97.5 -1.9 22 ~26] 0.016
9.7 144.8 6.7 100.8 -2.9 24 -3.97 0.001
Oct 9.4 141.6 8.1 121.1 -1.4 20 -1.59 0.129
8.9 143.1 7.9 125.8 -1.1 20 -1.26 0.224
Nov 7.6 114.6 6.9 104.0 -0.7 20 -O.81 0.428
. 6.3 94.0 5.8 87.0 -0.5 21 -0.53 0.600
Dec 4.4 66.6 4.4 65.5 -0.1 21 -0.09 0.932
3.4 53.7 3.1 49.5 -0.3 21 -0.30 0.767

Total 1 840 l 593

 

The interaction between study and season was not signiﬁcant (Table 10; mixed-model
ANOVA; F9115] = 1.19; P = 0.372), indicating that the seasonal pattern did not differ

over all seasons between the two studies. Differences were greatest in May-June, and

62

Fig. 7. Seasonal depth by half-month intervals occupied in 2002-2005 by tagged lake
trout of upper Great Lakes origin (GLO) and of Finger Lakes, New York origin (FLO)
at large in Lake Huron as recorded with implanted archival tags. Symbols and 95%
conﬁdence intervals are least-squares means ﬁ'om a rrrixed-model AN OVA.

 

0 I I I I I I I I I I I

10- -

N
C
1

: G'ﬁ‘ i

' . .1. *

m'.’J..M“l'J o'e‘ol'l
§FLO

J FIMIAIMIJIJIAISIOINID
Month

0)
O
r

 

Depth (m)

 

40 lllll‘l‘m

 

 

50

early August (Fig. 6c). Finger Lakes origin lake trout also occupied signiﬁcantly lower
temperatures in 2002-2005 than in 1999-2000 (Table 10; mixed-model ANOVA; F1196]
= 84.13; P < 0.0001). The interaction between studies and seasonal intervals was also
signiﬁcant (mixed-model ANOVA; F3165] = 3.84; P < 0.0001), indicating that the
seasonal pattern differed between the two studies. The greatest differences were in
May-August for GLO and May-early November for FLO lake trout. Compared to
1999-2000 (Chapter One), GLO lake trout averaged 1.7°C lower temperatures from
May-August and FLO lake tout 1.6°C lower from May through mid-November (Table

5; Table 2 in Chapter One).

63

Table 9. Least-square means of individual ﬁsh medians for depth
(m) occupied by Great Lakes and Finger Lakes origin lake trout in
Lake Huron by half-month interval from a mixed-model ANOVA
and the results of post-hoe comparisons (Tukey t—tests) between

 

 

o_rigins within each interval.
Half-
month GLO FLO A P >
Interval LSM LSM LSM df t I tl

Jan 36.6 41.3 4.8 298 1.06 0.291
38.1 41.1 3.1 299 0.68 0.497
Feb 36.9 40.6 3.7 298 0.83 0.409
36.1 40.4 4.3 297 0.96 0.338
Mar 34.3 40.1 5.9 293 1.33 0.186
34.7 40.8 6.1 286 1.4 0.164
Apr 35.2 40.4 5.3 280 1.23 0.220
35.2 38.4 3.1 261 0.75 0.452
May 31.5 35.0 3.5 229 0.88 0.382
32.3 35.4 3.1 260 0.88 0.378
Jun 30.1 36.6 6.6 241 1.96 0.051
27.2 33.6 6.4 216 1.96 0.051
Jul 24.9 31.0 6.1 224 1.86 0.064
34.4 36.7 2.3 236 0.69 0.491
Aug 35.8 38.1 2.3 248 0.66 0.511
33.6 38.8 5.2 263 1.47 0.143
Sep 34.1 37.5 3.4 281 0.91 0.363
28.8 36.3 7.6 283 1.97 0.050
Oct 25.4 33.7 8.3 236 2.04 0.043
27.1 31.2 4.1 260 0.98 0.328
Nov 30.5 33.9 3.4 277 0.79 0.430
31.4 34.0 2.7 287 0.61 0.543
Dec 33.9 37.6 3.7 293 0.84 0.404
35.6 39.2 3.7 297 0.81 0.416

 

Table 10. Results of a mixed model ANOVA testing whether the
temperatures occupied by lake trout of Great Lakes (GLO) and Finger
Lakes (FLO) origins in Lake Huron changed over time , as measured
with archival tags. The main effects were study (two intervals of years;
1998-2001 and 2002-2005), season (24 half-month seasonal intervals),
and the interaction of study and season.

 

 

Origin Fixed Effects df F P > F

GLO Study 1, 107 5.77 0.0181
Season 23, 15 19.09 <0.0001

Study *Season 23, 15 1.19 0.3722
FLO Study 1, 66 84.13 <0.0001
Season 23, 65 124.60 <0.0001
Study * Season 23, 65 3.84 <0.0001
Discussion

All original concern leading to this study was the question Of whether lower
temperatures occupied by FLO lake trout in 1999 and 2000 might not be due to an
innate difference in preference but rather related to geographic distribution and potential
differences in therrnocline depth as inﬂuenced by prevailing winds. The question was,
could lake trout of one genetic origin or the other tend to be either inshore versus
offshore or on the east or west side of the lake rather than in deeper or shallower water?
The consistent depth difference reported here (Fig. 7) argues against that explanation
and for the consistent selection of lower temperatures and greater depth by FLO lake

trout, regardless of geographic location.

65

The results of Chapter One and this study provide new evidence to suggest that
innate differences exist between the lake trout origins, which cause them to occupy
different temperatures and depths. The difference in temperatures and depths occupied
by GLO and FLO lake trout during summer stratiﬁcation in 1999-2000 (Chapter One),
here reported for the period 2002-2005, and reported in Lake Ontario in 1984-1993 by
Elrod et al. (1996) were consistent. In all cases, the GLO lake trout occupied the higher
temperatures. Additionally, lake trout of both origins appeared to respond similarly to
the environmental variables responsible for the decrease in occupied temperatures
between Chapters One and Two (Fig. 6). Swim bladder gas retention, which is related
to depth distribution, has been demonstrated to be heritable (Ihssen and Tait 1974) and,
thus, could affect depth preference and distribution Observed in this study. Likewise,
temperature preference in laboratory studies was shown to be heritable among
Salvelinus species (Goddard and Tait 1976). Elrod et al. (1996) presented data ﬁom gill
net surveys conducted in September 1984-1993, concluding that the Superior strain
(SUP=GLO) was found signiﬁcantly shallower and warmer than the Seneca strain
(SEN=FLO). For GLO and FLO ﬁsh, Elrod et al. (1996) presented least-square, mean
temperatures for September (interpolated from their Fig. 8) of about of 7.8 and 6.5 °C
versus my results of 8.3 and 7.7 °C in Chapter One (Table 2) and 8.4 and 6.4 °C in
Chapter Two (Table 8). The corresponding LSM depths for GLO and FLO were about
33 and 42 m in Elrod et al. (1996) versus my results of 34.1 and 37.5 m in Chapter Two
(Table 9). These results were consistent despite spanning two lakes, two decades, and
two degrees of latitude. In interpreting their results, Elrod et al. (1996) focused

primarily on the potential for strains to reﬂect phenotypic traits selected for in the area

66

where they evolved, citing the example of the Clearwater Lake, Manitoba, strain, which
were consistently captured at the shallowest depth and highest temperatures in Lake
Ontario. Clearwater Lake is a shallow northern lake with a mean depth of 13 m and
only 15% of its area greater than 24 m. This explanation of the difference between
GLO and FLO lake trout is less plausible because Seneca Lake is slightly deeper than
Lake Michigan based on average depth (85 versus 82 m), but is not deeper based on
maximum depth (188 versus 282 m). Furthermore, Seneca Lake is much shallower than
Lake Superior, which has an average and maximum depth of 147 and 406 m. It seems
more likely that, if Seneca Lake strain has a preference for depths greater than selected

by GLO lake trout, selection would have occurred along a niche axis other than depth.

The decrease observed in temperatures occupied between 1999-2000 and 2002-2005
coincided with recent changes in the food web in Lake Huron. Decreases in abundance
of Diporeia spp. (Nalepa et al. 2007), and the abundance and species composition of the
prey ﬁsh community (Riley et al. in press) could have inﬂuenced bathythermal habitat
use. The most likely causes of the shiﬁ are changes in the distribution of available prey
species, necessitating changes in search patterns for prey, or selection of lower
temperatures to maximize conversion efﬁciency of food as prey availability becomes

severely limited, as predicted by Mac (1985).

Changes in composition of the prey ﬁsh community could have put pressure on
predators to move deeper. Since their ascendance as a dominant pelagic species in the

Great Lakes, alewives became the most prevalent prey of salmonines (O’Gorman and

67

Stewart 1999). During the course of this study, alewife numbers dramatically declined
in Lake Huron. Acoustic surveys by the USGS estimated their biomass at over seven
kg-ha"l in 1997, but only 0.002 ltg-lia'l in 2004 (Warner et al. 2004). In 2005, the
estimate was 0.012 kg-ha’l (Schaefer et al. 2005), still almost two orders of magnitude
lower than in 1997. In the absence of a key pelagic prey base, lake trout may have been
forced to prey upon demersal species, and thus, used greater depths and lower
temperatures in 2002-2005 than they did in 1999-2000, when pelagic prey were more
abundant. Food habit data collected across US. waters of Lake Huron over this time
period document a switch from pelagic to demersal prey species (Fig. 8). The pelagic
species found in lake trout stomachs were primarily alewives and ninespine sticklebacks
(Pungitius pungr'tius), with rainbow smelt (Osmerus mordax) important seasonally.
Alewives systematically disappeared ﬁ'om the diet (Fig. 8). By 2004, alewives were
largely absent from the diet except in MHl, which is nearest the connection with Lake
Michigan, where alewives remained moderately abundant (Madenjian et al. 2005). The
round goby, a recent invader and a strongly demersal species, has emerged as a major

prey item of lake trout (Fig. 8).

Prey availability decreased severely between the periods of data collection for
Chapters One and this study. This decrease in prey could have caused a selection of
lower temperatures than in the earlier study to conserve energy and make more efﬁcient
use of resources. The mean biomass of all demersal prey species taken in USGS bottom
trawl tows in Lake Huron had declined since the mid 1990S, and dropped sharply to

very low levels in 2003 (Riley et al. in press; Fig. 9). For all pelagic ﬁsh species

68

combined, biomass decreased by about 90% between 1997 and 2005 (Schaefer 2005).

In general during the present study (2002—2005), substantially fewer prey were

available to lake trout, and empty stomachs were more prevalent (Fig. 8) than in the

previous study (Chapter One).

Fig. 8. The proportion of empty stomachs (dark ﬁll) and the percent of diet by weight
for nine prey items in stomachs of lake trout collected in three areas of the US. waters
of Lake Huron (Fig. 4). Data were provided by the Michigan Department of Natural
Resources, Alpena Fisheries Station.

Empty stomachs and prey species eaten

Northern (MH-1)

Empty stomachs h I 1| L L *

Salmonine sp. ~
Trout-perch - a 3 4 ~
Round goby ~ 4 {l {l - “j
Burbot 4 . - - . ‘
Stickleback 4 « —
Sculpin - . -1 .
Coregonine sp. ~ . ~ ~ 4 3

Rainbow smelt 4:: -: Cl -:1

Alewife a... 14%. are. 1%.. 1W

North-central (MH-2)

 

 

J l

 

 

 

 

 

 

 

 

Empty stomachs 4| *1 I ‘I h
Salmonine sp. ~ 4 . . 4
Trout-perch ~ . 4 1 4
Round goby ~ ~ *3 4] ~
Burbot ~ . . i
Stickleback ~ . - .
Sculpin - 4
Coregonine sp. -

Rainbow smelt 1:1 C3 Cl

:1
Irv—mm
l-
J
‘ _ it
43 _
Alewife W 27% am @ﬁ - E
h
l
L

 

 

L L44

 

 

 

Southern half (MH-3,4,5)
Empty stomachs 1i 1| + +

Salmonine sp. 1 1
Trout-perch ~ . - ~ ~
Round goby . - {I {l ~:|
Burbot ~ 3 , 1 ~
Stickleback 4
Sculpin . ~ 4
Coregonine sp. ~ . 4| «I -

Rainbow smelt C1 C1 . i «:3 3::

Alewife am m p Q -| ,
0 1 0 1 0 1 0 1 0 1 0 1
1999 2000 2002 2003 2004 2005

Year and proportions of empty stomachs and prey species by weight

 

 

 

 

 

 

 

 

 

 

69

The drastic reduction in prey availability over recent years could have triggered
physiological responses affecting temperature preference. Good agreement exists
between laboratory determinations of the ﬁnal temperature preferendum and the optimal
temperature for growth of many ﬁsh species (McCauley and Casselrnan 1981; Jobling
1981). Determinations of the optimal temperature for growth require feeding at an
optimal or excess rate; however, piscivorous ﬁsh in oligotrophic lakes will rarely have
prey supplies at excess levels and are hence not capable of supporting maximum
growth. Brett et al. (1969) hypothesized that “the optimum temperature for growth
would drop as the ration decreased.” The hypothesis was based on the assumption that
decreased metabolic cost for maintenance at lower than optimum temperatures would
make more energy available for growth when food was limited. By varying ration size
for sockeye salmon (Oncorhynchus nerka), optimum temperature for growth decreased
from 15°C at a ration of 6% of body weight per day to 5°C at a ration of 1.5% per day
(Brett et a1. 1969). This relation could be expected to have evolutionary signiﬁcance
and result in observable behavior. Such a behavior was demonstrated experimentally
with juvenile lake trout, where a linear decrease in preferred temperature occurred when
rations decreased from 5.0% to 0.3% of body weight (Mac 1985). The preferred
temperature at the lowest ration tested was 9 °C, just slightly higher than observed for
GLO lake in this study from mid June through September, and two to three degrees
higher than observed for FLO lake trout. Citing unpublished data by D.V. Rottiers,
Mac (1985) suggested that reductions in metabolic costs could be gained by using
temperatures as low as 5 °C, which is identical to the results of Brett et al. (1969).

Using techniques similar to Mac (1985), preferred temperatures were determined for

70

brown trout (Salmo trutta) and arctic char (Salvelinus alpinus) (Larsson 2005). The
preferred temperature of brown trout (16 °C) was close to the optimum temperature for
growth of 17 °C (Elliot and Hurley 1999), but the preferred temperature for Arctic char
(11.8 °C) was 4.5 °C less than the optimum temperature for growth (Larsson and
Berglund 1998). Reasoning that Arctic char typically inhabit low productivity waters,
Larson (2005) speculated that char may tend to optimize conversion efﬁciency rather
than growth. In Lake Huron, lake trout were rarely found near their optimum
temperature for growth (Christie and Regier 1988; Fig. 5). Retreating to temperatures
lower than optimum in response to a growing scarcity of food could increase conversion
efficiency and partially explain the observed decrease in temperatures occupied by lake

trout during this study.

Fig. 9. Mean biomass (kg) and 95% conﬁdence limits in USGS fall trawl catches at
ﬁve ports in the Michigan waters of Lake Huron, 1976-2006 (data from Riley et al. in
press).

 

 

 

 

50 I I I I I I I
40 4 + ,
3 3° ‘ ‘
C
8 20 + i
2
10 - f + -
0 I I I I I j j
1999 2000 2001 2002 2003 2004 2005

Year

71

By occupying lower temperatures, lake trout may cause parasitic sea lampreys to
reduce their blood consumption rate. Temperature data from Chapter One have already
been incorporated into an individual-based model (IBM) of sea lamprey feeding and
growth (Madenjian et al. 2003), where the surnmer-fall temperature regimes of GLO
and FLO lake trout in 1999-2000 resulted in only a 3% increase in total blood
consumption by sea lampreys feeding on GLO lake trout. However, sea lamprey
growth surges in fall (Bergstedt and Swink 1995) and a disproportionate amount of
blood consumption and mortality occurs then. The difference reported by Madenjian et
al. (2003) was not great, but the number of predicted deaths was 7% greater under the
GLO than under the FLO temperature regime. To explore how very low temperature
regimes (such as might characterize the siscowet, or deep-water, form of lake trout in
Lake Superior) might affect growth and mortality, the IBM was run not allowing
seasonal temperature to exceed 4°C. Such a constant low resulted in a substantial
decrease in predicted sea lampreys growth and blood consumption (Charles Madenjian,
USGS, Great Lakes Science Center, personal communication). Growth of sea lampreys
and predicted lake trout losses should be evaluated using the IBM and GLO and FLO

temperature regimes from Chapter Two.

Summary statistics describing habitat dimensions such as the temperatures and
depths occupied can obscure the tremendous individual variation and the potential for
habitat overlap with other members of the ﬁsh community. During late fall through
early spring lake trout were relegated to a narrow range of temperatures (Fig. 5). In the

winter, temperature in the upper 100 m, or more, of the water column ranges from 0 to 2

72

°C, due to the small change in density over that range and wind-driven mixing. During
the period of summer stratiﬁcation, individuals occupied temperatures ranging ﬁ'om 4 to
12 °C. The pattern for variability in depth was related, but opposite, since lake trout
during the winter could range widely in depth with little effect on temperature occupied.
The difference in median depths occupied among seasons was not as striking as with
temperature. The greatest variability in depth was recorded during winter, when
temperature varied by a maximum of about 2°C,'(Fig. 5). Unless lake trout moved deep
enough in winter to enter the warmer layer below the winter mixing layer (greater than

100 m; Bergstedt and O’Gorman 1989), temperature varied little.

Archival tags can be a cost-effective tool for collecting data on lake trout in lakes
and with species where sufﬁcient exploitation will provide for the return of tags. At a
return rate of 25%, our cost per returned tag was about $1,200 (U SD), including the
reward. Although that cost is higher than a single acoustic or radio tag encoded for
depth and temperature, the difference was offset by elimination of the need to keep a
crew on the lake for extended periods and by the data not being interrupted by gaps due
to weather or losses of contact. The cost also compares favorably with the
extraordinary cost that would be needed for vessel operations to generate comparable
data through netting. A key to encouraging the participation of sport and commercial
ﬁshers was the $100 reward offered. By offering a substantial amount, most people
encountering either the internal or external tag likely responded. In marine studies,
where distance and effort required to return a tag can be much greater, substantially

higher rewards have been offered (Block et al. 2005). Administrative approval of

73

suitably large rewards can be difficult to obtain, but when compared to the value of the
data, a reward that increases recovery rate by even a few percent is warranted. In our
case, most tags recovered within two years of deployment were redeployed, thereby

compensating for the cost of the reward.

The increase from 19 to 31% in tag recovery between 2002 and 2003 (Table 4),
likely reﬂected both the fact that in 2002, some ﬁsh tagged were captured in gill nets
and experienced more stress during capture and that some ﬁsh were held for a long
period (three weeks) to accommodate genotyping. The relatively small difference in
recovery rate between lake trout tagged and released on the lake and at the HBBS
suggests that the effect of capture and transport to the HBBS was minimal if the animals
tagged were released quickly (Table 4). If possible, however, tagging immediately at
the site of capture to eliminate as much stress from handling and transport as possible is

recommended.

There was an initial constraint imposed by the lack of availability of archival tags
small enough for lake trout. Tags of a suitable size only appeared in the late 19903. By
2005, tags were available that are small enough for most freshwater game and
commercial species and half the size Of the LTD-1110 tag used here. Those seeking to
collect similar data are encouraged to investigate the use of archival tags. I spent eight
of the ﬁrst ten years of my career working aboard a research vessel on Lake Ontario. I
am convinced that no technique that I saw is capable of generating the quality and

continuity of the data presented here at a similar cost.

74

Acknowledgments

I thank Mary K. Jones, Karen Slaght, Dave Keffer, and Erick Larson of the
Hammond Bay biological Station for their professionalism and support in the conduct
of this study, Mark Ebener for his help in coordinating ﬁsh collections from the tribal
ﬁsheries, and Jim Bence for assistance with statistical analyses. Funding for this study

was provided by the US. Geological Survey and Great Lakes Fishery Commission.

75

References

Bergstedt, R.A., and O'Gorman, R. 1989. Distribution of alewives in southeastern Lake
Ontario in autumn and winter: a clue to winter mortalities. Trans. Am. Fish. Soc.
118: 687 692.

Bergstedt, R.A., and Swink, W.D. 1995. Seasonal growth and duration of the parasitic
life stage of the landlocked sea lamprey (Petromyzon marinas). Can. J. Fish. Aquat.
Sci. 52: 1257—1264.

Block B.A., Teo, S.L., Walli, A., Boustany, A., Stokesbury, M.J., Farwell, C.J., Weng,
K.C., Dewar, H., and Williams, TD. 2005. Electronic tagging and population
structure of Atlantic blueﬁn tuna. Nature. 434: 1121-1127.

Brett, J .R., Shelbourn, J.E., and Shoop, CT. 1969. Growth rate and body composition
of ﬁngerling sockeye salmon, Oncorhynchus nerka, in relation to temperature and
ration size. J. Fish. Res. Board Can. 26: 2363-2394.

Christie, GO, and Regier, HA. 1988. Measures of optimum thermal habitat and their
relationship to yields for four commercial ﬁsh species. Can. J. Fish. Aquat. Sci. 45:
301-3 14.

Eaton, J.G., McCormick, J.H., Goodno, B.E., O’Brien, D.G. Stefany, H.G., Hondzo,
M., and Scheller, RM. 1995. A ﬁeld information-based system for estimating ﬁsh
temperature tolerances. Fisheries (Bethesda) 20(4):10—18.

Edsall, TA, and Cleland, J. 2000. Optimum temperature for growth and preferred
temperatures of age-0 lake trout. N. Am. J. Fish. Manage. 20: 804-809.

Elliott, J .M., and Hurley, M.A. 1999. A new energetics model for brown trout, Salmo
trutta. Freshw. Biol. 42: 235-246.

Elrod, J .H., O'Gorman, R., and Schneider, GP. 1996. Bathythermal distribution,
maturity, and growth of lake trout strains stocked in US. waters of Lake Ontario,
1978-1993. J. Great Lakes Res. 22: 722-743.

Fry, F.E.J. 1947. Effects of the environment on animal activity. Univ. Toronto Stud.
Biol. Ser. 55, Publ. Ont. Fish. Res. Lab. 68: 1-62.

Goddard, CL, and Tait, J .S. 1976. Preferred temperatures of F3 to F5 hybrids of
Salvelinusfontinalis X S. namaycush. J. Fish. Res. Board Can. 33: 197-202.

Grande, M., and Andersen, S. 1991. Critical thermal maxima for young salmonids. J.
Freshw. Ecol. 6: 275-279.

76

Hecky, R.E., Smith, R.E.H., Barton, D.R., Guildford, S.J., Taylor, W.D., Charlton,
M.N., and Howell, T. 2004. The nearshore phosphorus shunt: a consequence of
ecosystem engineering by dreissenids in the Laurentian Great Lakes. Can. J. Fish.
Aquat. Sci. 61: 1285—1293.

Ihssen, P., and Tait, J .S. 1974. Genetic differences in retention of swimbladder gas
between two populations of lake trout (Salvelinus namaycush). J. Fish. Res. Board
Can. 31: 1351-1354.

J obling, M. 1981. Temperature tolerance and the ﬁnal preferendum—rapid methods for
the assessment of optimum growth temperatures. J. Fish Biol. 19: 439-455.

Krueger, C.C., Marsden, J .E., Kincaid, H.L., and May, BL. 1989. Genetic
differentiation among lake trout strains stocked into Lake Ontario. Trans. Amer.
Fish. Soc. 118: 317-330.

Larsson, S., and Berglund, I. 1998. Growth and food consumption of 0+ Arctic char fed
pelleted or natural food at six different temperatures. J. Fish Biol. 52: 230-242.

Larsson, S. 2005. Thermal preference of Arctic char, Salvelinus alpinus, and brown
trout, Salmo trutta,—implications for their niche segregation. Environ. Biol. Fishes
73: 89-96.

Madenjian, C.P., Cochran, P.A., and Bergstedt, R.A. 2003. Seasonal patterns in growth,
blood consumption, and effects on hosts by parasitic-phase sea lampreys in the
Great Lakes: an individual-based model approach. J. Gt. Lakes Res. 29 (Suppl. 1):
332-346.

Madenjian, C.P., Bunnel, D.B, Desourcie, T.J., Holuszko, J .D., and Adams, J.V. 2005.
Status and trends of prey ﬁsh populations in Lake Michigan, 2004. Annual report to
the Great Lakes Fishery Commission, Lake Michigan Committee. Great Lakes
Science Center, Ann Arbor, Mich.

Mac, M.J. 1985. Effects of ration size on preferred temperature of lake charr Salvelinus
namaycush. Environ. Biol. Fishes 14: 227-231.

Magnuson, J .J ., Crowder, LB, and Medvick, PA. 1979. Temperature as an ecological
resource. Am. Zool. 19: 331-343.

McCauley, R.W., and Tait, J .W. 1970. Preferred temperature of yearling lake trout,
Salvelinus namaycush. J. Fish. Res. Board Can. 27: 1729-1733.

McCauley, R.W., and Casselrnan, J .M. 1981. The ﬁnal preferendum as an index of the
temperature for optimum growth in ﬁsh. Proceedings of the World Symposium on
Aquaculture in Heated Effluents and Recirculation Systems, Stavanger, Norway.
Vol. 2. pp 81-92.

77

Nalepa, T.F., Fanslow, D.L., Pothoven, S.A., Foley, A.J., and Lang, GA. 2007. Long-
term trends in benthic macroinvertebrate populations in Lake Huron over the past
four decades. J. Great Lakes Res. 33: 421—436.

O’Gorman, R., and Stewart, T]. 1999. Ascent, dominance and decline of the alewife in
the Great Lakes: food web interactions and management strategies. In Great Lakes
ﬁsheries policy and management, a binational perspective. Edited by W. Taylor, and
GP. Ferreri. Michigan State University Press, East Lansing, Mich. pp. 489-513.

Page, KC. 2001. Genetic diversity and interrelationships of wild and hatchery lake trout
in the upper Great Lakes: inferences for broodstock management and development
of restoration strategies. M.Sc. thesis, Department of Fisheries and Wildlife,
Michigan State University, East Lansing, MI.

Peterson, R.H., Sutterlin, A.M., and Metcalfe, J .L. 1979. Temperature preference of

several species of Salmo and Salvelinus and some of their hybrids. J. Fish. Res.
Board Can. 36: 1137-1140.

Riley, S.C., Roseman, E.F., Nichols, S.J., O’Brien, T.P., Kiley, CS, and Schaeffer, J .S.
in press. Deepwater demersal ﬁsh community collapse in Lake Huron. Trans. Am.
Fish. Soc.

SAS Institute Inc. 2001. SAS release 8.02 ed. SAS Institute Inc., Cary, NC.

Schaeffer, J.S., Bowen, A., Thomas, M., French, J .R.P., H1, and Curtis, G.L. 2005.
Invasion history, proliferation, and Offshore diet of the round goby Neogobius
melanostomus in western Lake Huron, USA. J. Great Lakes Res. 31: 414-425.

Warner, D.M., O’Brien, T.P., Faul, CS, and Stickel, R.G. 2004. Status of pelagic prey
ﬁsh in Lake Huron in 1997 and 2004. Annual report to the Great Lakes Fishery
Commission, Lake Huron Committee. USGS, Great Lakes Science Center, Ann
Arbor, MI.

78

CHAPTER THREE

Seasonal and diel depth and temperature distributions of lake Whiteﬁsh and of Great

Lakes- and Finger Lakes-ori gin lake trout relative to exploitation and potential bycatch.

79

Abstract

Depth and temperature data recorded during 2003 -2005 in Lake Huron by archival
tags implanted in lake Whiteﬁsh (Coregonus clupeaformis) and two strains of lake trout
(Salvelinus namaycush) were used to compare their seasonal depth and temperature
distributions and to examine whether seasonal differences in depths could be exploited
to reduce bycatch of lake trout in commercial gill net ﬁsheries for lake Whiteﬁsh. Both
Great Lakes origin (GLO) and Finger Lakes origin (FLO) lake trout occupied during
both daylight and dark signiﬁcantly deeper and colder waters seasonally than lake
Whiteﬁsh. The temperature differences were greatest during periods of stratiﬁcation
when behavioral thermoregulation was possible. The greatest depth and temperature
separation occurred between Finger Lakes lake trout and lake Whiteﬁsh. Based on a
comparison of the seasonal depth distributions of the two species, if the maximum depth
of gill nets could be regulated seasonally from 25 to 35 m, one-half or more of the lake
Whiteﬁsh population would be vulnerable while avoiding most of the lake trout
population. Percentages of lake trout targeted were lowest in late July (GLO = 8% and
FLO = 11%) and early August (GLO = 4% and FLO = 7%). Feasibility of this
approach depends on the shallower lake Whiteﬁsh being available to the bottom-set gill
nets and not suspended pelagically. Modal depths of commercial gill net effort for lake
Whiteﬁsh in Lake Huron exceeded modes from the lake Whiteﬁsh archival tag data and
suggested a portion of the lake Whiteﬁsh population was pelagic and suspended far
enough above bottom to not be vulnerable to the gill nets. Use of suspended gill nets to

target those ﬁsh could increase their vulnerability and further avoid lake trout.

80

Introduction

The structure of the ﬁsh community in Lake Huron has changed dramatically,
inﬂuenced by ﬁshing and the effects of invasive species over the past century. After
decades of exploitation and combined with the effects of the parasitic sea lamprey
(Petromyzon marinas), the abundance of the commercially important lake Whiteﬁsh
(Coregonus clupeaformis) was greatly reduced while the lake trout (Salvelinus

namaycush) became nearly extinct (Berst and Spangler 1972; Hansen 1999).

A common requirement for restoration of both species was control of the sea
lamprey. Beginning in the 19703, Lake Huron’s tributaries were treated with
larnpricides to reduce sea lamprey abundance (Smith and Tibbles 1980). These initial
efforts reduced sea lamprey abundance substantially, but not to levels thought necessary
for lake trout restoration. Stocked lake trout survived initially for the ﬁrst few years Of
their lives, but did not survive to maturity (age 6+ years) in adequate numbers. It was
not until a strategy to control a large remaining population of sea lamprey larvae in the
outﬂow from Lake Superior was implemented in the late 1990s (Schleen et al 2003)
that sea lamprey numbers approached levels thought to permit rehabilitation of the lake

trout population.

In contrast to lake trout, initial sea lamprey control efforts were adequate for lake

Whiteﬁsh restoration, as harvest in the 1990s was the highest on record (Ebener 1997;

Brown et al. 1999). When other hosts such as lake trout and Chinook salmon

81

(Oncorhynchus tshanytscha) were abundant, lake Whiteﬁsh became a secondary host to
the sea lamprey. More important than sea lamprey control, lake Whiteﬁsh restoration
was likely spurred by a number of changes in commercial ﬁshing regulations and the
ﬁshing gear used. In the 19708, the State of Michigan banned use of large-mesh gill
nets and promoted adoption of trap nets as the preferred gear (Brown et al 1999). Trap
nets, because of their capture of live ﬁsh, allow more ﬂexible regulations. Bycatch in
trap nets can be regulated to allow the release of non-target ﬁsh including small lake
Whiteﬁsh and species such as lake trout. Greater cooperation among State, Provincial,
and Tribal ﬁshery management agencies, and the adoption of quota management and
strategies for setting total allowable catch restrictions have also aided in the lake

Whiteﬁsh recovery.

Management of the lake Whiteﬁsh ﬁshery affects lake trout restoration because the
principal bycatch, regardless of the gear used, is lake trout (McNeil and deLaplante
1989; Johnson et al. 2004a). Bycatch consists of both the retained and discarded non-
target organisms, including unobserved mortalities of ﬁsh that are released (Crowder
and Murawski 1988). The observed portion of lake trout bycatch mortality are the ﬁsh
harvested under quotas or discarded by the gill net ﬁshery, and to a lesser-extent, gilled
in the leads and pots of trap nets (Johnson et al. 2004b). Unobserved mortality includes
delayed deaths caused by injury, stress, disease, or predation (Chopin et al. 1996).
Johnson et al. (2004b) estimated the survival of lake trout after being released alive
from the pots of trap nets as 98% (using in part data from Chapters One and Two). In

contrast, the survival of live lake trout taken from gill nets was estimated at 72%

82

(Gallinat et al. 1997). Recommendations to reduce bycatch losses of lake trout included
limiting lake Whiteﬁsh harvest to trap nets and reducing effort in summer when the ratio

of lake trout to lake Whiteﬁsh caught in either gear was highest (Johnson et al. 2004b).

Fishery managers desire to reduce losses of lake trout due to bycatch so as to
promote population rehabilitation. Less known and appreciated, is that commercial
ﬁshers also have an interest in the reduction of bycatch. Both gill net and trap net
ﬁshers found that as the lake trout population increased in the 20008, sorting, handling,
and releasing lake trout bycatch from their gear was a major task. Any viable strategy
that maintained their access to lake Whiteﬁsh, while reducing lake trout bycatch, would

be welcomed.

One approach to reducing bycatch would be to exploit differences in the spatial
distribution of species. In 1998 and 1999, a multi-agency effort occurred to determine
if sufﬁcient spatial separation existed at any seasonal period to permit a gill net ﬁshery
in areas Of Lake Huron designated for lake trout rehabilitation (Johnson et al. 2004b).
From October 1998 through December 1999, members of the Bay Mills Indian
Community in Michigan made 260 monitored gill net lifts. The number of lake trout
killed per lift decreased seasonally from 17.2 per lift during March-June, to 7.4 per lift
during J uly-Oct 7, to 4.9 per lift during the remainder of October and November. Based
on those results, a tribal small-boat gill net ﬁshery in Hammond Bay has been allowed

from October 1 to November 6 each year 2000-2007. Considering the change in lake

83

trout depth distributions since 1999 (Chapter 2), a reexamination of the seasonal overlap

between lake trout and lake Whiteﬁsh using more recent data is warranted.

The objectives Of this chapter were to use data from archival tags implanted in lake
Whiteﬁsh and lake trout of two genetic origins to (1) describe the seasonal temperature
and depth distributions of lake Whiteﬁsh, (2) compare the seasonal and day-night
temperature and depth distributions of lake Whiteﬁsh and lake trout on a ﬁner time scale
and broader geographical scale, and (3) to estimate the likely relative effects on lake
trout bycatch if the ﬁshery was managed to have seasonal depth restrictions on gear
placement. The key questions were: did the overlaps in depth distributions between the
two species change by season or by day versus night and might any of the seasonal

overlaps be exploitable to affect bycatch in the commercial ﬁshery.

Methods

Tags and tagging

Archival tags, model LTD-1110 manufactured by Lotek Wireless Inc., Newmarket,
Ontario, Canada, were surgically implanted in lake trout and lake Whiteﬁsh released
into Lake Huron. The tags were cylindrical—11 mm in diameter by 32 mm long and
rounded on one end. Weight was about ﬁve g in air and 2 g in water. Each tag
contained a thermistor and pressure sensor, and recorded temperature and depth.

Recording continued until tag recovery and download, or until the battery expired

84

(battery life was two to three years). Observations of date, time, temperature, and depth
were recorded at 15-second intervals until the memory was ﬁrll. Software then replaced
previous observations in a pattern that maintained a complete set of observations from
initialization to download, but with gradually increasing intervals between Observations.
Nearly 33,000 observations were recorded with average intervals increasing from 15

seconds by about 1.3 minutes for each month deployed and 16 minutes for each year.

Tag speciﬁcations were for 0.2°C resolution and i 0.3 °C accuracy for temperature
and 0.4% of the current depth scale with i1% accuracy. The potential depth range was
0 to 500 m, but, to improve resolution and accuracy, software initially set the range to
one-quarter of that scale (0 to 125 m). If ﬁsh moved deeper than the initial scale, the
range would progressively shift (permanently) to one-half scale (0 to 250 m) and then to
full scale (0 to 500 111). Resolution and accuracy (in parentheses) at those scales were

0.5 (i 1.25) m, 1.0 (:1: 2.5) m, and 2.0 (:1: 5.0) m, respectively.

Recovery of archival tags depended on the recapture of ﬁsh by sport and
commercial ﬁsheries and return of the tags. To promote returns, a reward of$100
(U SD) per tag was offered. Tagged ﬁsh were also marked externally with orange dart
tags. The external tags bore identifying serial numbers, reward advertisement, a contact
telephone number, and instructions to not freeze the tag. The internal tags also bore a
reward notice and contact information. Information posters describing the study and

reward were displayed in tackle and convenience stores, at launch sites, and were

85

provided to commercial and charter ﬁsheries. The Michigan Department of Natural

Resources and the Ontario Ministry of Natural Resources also publicized the study.

All lake Whiteﬁsh tagged were obtained from commercial trap nets (Table 11). In
2002, lake Whiteﬁsh were transported to the Hammond Bay Biological Station (HBBS)
for tagging. Unlike the lake trout, the lake Whiteﬁsh appeared stressed after transport
and tagging. Because of this observation, and the recovery of only one ﬁsh during
summer 2002, the remainder of lake Whiteﬁsh tagged in 2003 and 2004 were tagged
and released immediately on removal from the trap nets to eliminate transport stress.
Lake Whiteﬁsh tagged in November 2003 were ﬁom shallow experimental trap nets set
by the Chippewa Ottawa Resource Authority to collect ﬁsh for another tagging

experiment and were in spawning condition.

Archival tags were inserted through a 1.5-cm incision made slightly to the side of
the ventral rrridline anterior to the pelvic ﬁns. The incision was closed with two slow-
dissolving monoﬁlament sutures, a veterinary-grade, isocyanate adhesive was applied to
the knots, and an external tag inserted on each side of the dorsal ﬁn. The procedure
took about 3 minutes, during which the gills were irrigated. Lake Whiteﬁsh tagged at
the HBBS in 2002 were anesthetized, weighed, and measured before tagging. After
tagging, the ﬁsh were transferred to a recovery tank and released the next day. When
tagging was done in the ﬁeld (2003 and 2004), ﬁsh were moved directly from the trap
or gill net to a ﬂoating pen. After surgery, ﬁsh were placed in a recovery pen and

released when they regained equilibrium.

86

There were two genetic origins of lake trout used in this study. One was of upper
Great Lakes origin (GLO) and the other of Finger Lakes, New York (FLO) origin
(Krueger et al. 1989; Elrod et al. (1996a, 1996b); Chapter One). The source, handling,

and tagging of lake trout were identical to that described in Chapter Two.

Statistical analyses

Temperature and depth data were summarized by half-month seasonal intervals as
described in Chapter One. The ﬁrst half of each month included the ﬁrst 15 days and
the second half the remaining days in the month. The arithmetic mean temperature (°C)
in each half-month interval was used to best represent the average number of degree-
days experienced by a ﬁsh. For depth, the median depth occupied by each ﬁsh in each
interval was used. Because only one of the lake Whiteﬁsh tagged in 2002 was
recovered, only data from 2003-2005 from both lake trout and lake Whiteﬁsh were

included in the analyses.

With each ﬁsh, there was a recovery period after release during which the ﬁsh were
likely still recovering from surgery and the data were suspected to not represent normal
behavior. Each plot of temperature was inspected and a subjective judgment was made
as to the date when behavior appeared consistent with the remainder of the data. For
ﬁsh captured when the tag was still actively recording, the time of capture in the ﬁshing
gear was evident from a cessation of temperature and depth variation. Temperature

Observations before behavior appeared normal and after the ﬁsh was caught were not

87

used in the analyses. Because of the expanding interval of observations over time and
the greater number of Observations per hour for ﬁsh recaptured sooner after release, and
to give ﬁsh equal weight (the experimental unit in this study was the individual ﬁsh),
subsequent analyses were based only on the mean temperatures and median depths for
individual ﬁsh by half-month interval and considering the repeated nature of those
measures. Mean temperature was chosen for the reason stated above, and their
likelihood of use in bioenergetics models. The median depth was chosen as best
representing the most common depth frequented. However, in assessing exposure to

capture by bottom-set gill nets, the maximum depths observed were also summarized.

The temperatures analyzed and reported in this study were only those recorded
inside the abdominal cavity. At any instant, these measures could vary from the
temperature of the surrounding water. Based on laboratory studies (HBBS, unpublished
data), internal temperature after a rapid change in external temperature would move
50% toward the new external temperature within about ﬁve rrrinutes, and 90% toward
the new temperature within 15 minutes. Over the temporal scale of a half-month
interval, the mean internal temperature should closely approximate the average

temperature of the water occupied by a ﬁsh.

To describe the variability in the observations, percent frequency distributions for
temperature and depth were compiled by group for lake trout and lake Whiteﬁsh.
Because the time between release and recapture varied with individual ﬁsh, and because

of the gradually increasing intervals between observations, and hence the changing

88

number of observations per time interval, it was necessary to create a summary statistic
to ensure that all ﬁsh contributed equally to the distribution. Therefore, the average
temperature, median depth, and maximum depth occupied were calculated for
combinations of group, individual ﬁsh, half-month interval, and hour and used to

describe those statistics along with the 10th, 25m, 75‘“, 90th, and 90th percentiles by half-

month interval.

Because catch and bycatch in gill nets occurs primarily at night, a variable was also
added to account for light conditions. Times of sunrise, sunset, and twilight were
downloaded ﬁ'om the US. Naval Observatory (http://aa.usno.navy.mil/data) for the
span of the study. These data were used to classify each observation to one of three
light conditions—daylight (sunrise to sunset), twilight (the morning and evening
intervals when the sun is between 0° and 6° below the horizon, also termed civil
twilight), and dark (when the sun is more than 6° below the horizon). Observations
classiﬁed as being during twilight were eliminated in analyses that considered light
conditions. The distributions of temperatures and depths occupied by lake Whiteﬁsh
and each lake trout genetic origin were determined by day and night. The cumulative
percent frequency for depth was also determined by 3-m intervals and half-month

period.

By combining the lake Whiteﬁsh and lake trout data, three groups of ﬁsh were used

in the analyses—GLO lake trout, FLO lake trout, and lake Whiteﬁsh. The overall effect

89

Of group on the temperatures and depths occupied was evaluated with a mixed-model

ANOVA (SAS 2001). The underlying model for the analysis was

Yb,s,y,f 0rDh,s,y,f : :11 +ah +ﬂs +7h,s +6h,y +8h,s,y,f (l)

in which T h, S, y f or Dh,s,y,f is a half-month mean for temperature or depth and h, s, y,

and f denote the half-month interval, group, year, and individual ﬁsh. The ﬁxed effects

were the overall mean (u), the seasonal or half-month adjustment (or), the group

For the combined lake Whiteﬁsh and lake trout data, the best treatment of the
repeated nature of the data in the mixed-model ANOVA comparing depths and
temperatures occupied, was a ﬁrst-order, autoregressive structure allowed to differ for
ﬁsh within group ( 12). For both temperature and depth, the coefﬁcient for the year
by half-month variation (5) was best treated as a random effect allowed to differ
between groups (Table 12). This result was consistent with previous analyses of the
lake trout data (Chapter Two). Because gill nets catches were primarily made at night,

these analyses were also repeated separately by light condition (daylight or dark).

Results

Recoveries of lake whiteﬁsh came entirely from the commercial ﬁshery, because few

lake Whiteﬁsh are caught by Lake Huron sport ﬁsheries. Thirty-six of the total 361

90

tagged ﬁsh (10%) were recovered through 2005 (Table 11). Only one of 73 ﬁsh (1%)
tagged after transport to the HBBS was recovered and eight of 109 ﬁsh (7%) tagged in
November 2003 while in spawning condition were recovered. In comparison, 25 of 179
(14%) lake Whiteﬁsh tagged in May or June on the lake and immediately released were

recovered. Returns of tagged lake trout were described in Chapter 2.

Temperatures and depths occupied

Temperature

The temperatures occupied seasonally differed between lake Whiteﬁsh and each group
of lake trout (mixed model ANOVA, 112214], P < 0.0001). The seasonal pattern of least-
square mean (LSM) temperatures for lake Whiteﬁsh was most similar to GLO lake trout
(Fig. 10). Although the differences with GLO lake trout were relatively small, the LSM
temperatures for lake Whiteﬁsh were higher in 22 of 24 half-month intervals with one
tie (Table 13). However, only in late July and early August, at the peak of summer
stratiﬁcation, did the difference in LSMs between GLO lake trout and lake Whiteﬁsh
exceed 2°C. In only three additional half-month intervals, October and early November,
did the difference exceed even 1°C. The difference between the LSM temperatures
occupied by lake Whiteﬁsh and FLO lake trout was larger than with GLO lake trout and
more consistent (Table 13; Fig. 10). In all 24 half-month intervals, lake Whiteﬁsh
occupied higher temperatures than FLO lake trout (Table 13). From July through early
November, lake Whiteﬁsh were at temperatures from 1.9 to 3.7°C higher than FLO lake
trout. The greatest difference was in late July, when the LSM for lake Whiteﬁsh was

10.2 °C but only 6.5 °C for FLO lake trout.

91

90 Sam 83 ea 88 ease 65.. .000 220 268880 .6005 8:3 88220 80 eeoaaam a

 

 

one 2 0 mm 2 0m Gm E5

“No nm N 3 N #2 Dev .5 30H

em: 2 w E a 3 an: :0 a3 38 a: wee
wnmm n w n w 02 nab EA P3 mooN >02 NE:
03 S N N N ”N 02 000 mmmm .5 SON :3 2%
>8 o N m D To. .2 mm 000 mmmm HA moON .82 2% A
wmn m 3 2 S E 03 0.80 mu ”3?

as w mm om am an we as: :0 .5 88 a: 2-8
or _ a m 2 a mm :8. 5 S 88 80 28
o o o o o ow 00.: mmmm m3 NooN 5:. N
2a m E 2 2 S 2: 98 80: S 88 :2. 83:
%N m _ m ~ mm 00: mmm: ES

02 N e 2 R a R a8 8% S 88 a: one
m00N-mooN 03.3000 mummﬁﬁa 000.8 0008. 0000200 A093 c0580; momoomm 0.0wa“ 3qu
00:55:00 am 00m: E 00m: -0008 -0000." HEB: Z -005 00000 0

800 “Sodom emu mo 308$ 000802 0050m warm

00 995 000802 mammal

 

own: 00 when 180.. 05 @8305 05 5 00m: emu 00b 28 £83000 05 E 03: ﬁn mo 0000000 000 0008:: 05 00500002 0000000 000
0380: 05 080300 0008:: 0% .0058 £00.80— .momooem .0303 0000 E womwﬁ 0:00 003 m0 008:: 05 $00205 485$ 00:3
85 0080—8 000 mwﬁ 080800800. 00>an 08:30:: £3393 £5, nowwﬁ 5503:; 003 0:0 50: 00—2 .00 ggm .2 020,—.

92

Table 12. Results of model selection for a mixed-model ANOVA predicting the least-
squares mean temperature and depth occupied by lake Whiteﬁsh and lake trout of Great
Lakes origin and Finger lakes, NY origin in Lake Huron during 2003-2005. The AIC
values are for two potential treatments of the repeated nature of the observations crossed
with ﬁve potential treatments of the random components. The A AIC values are based on
the lowest AIC values (in hold) for analyses of temperature and depth.

 

 

 

 

Repeated statement Random statement AIC A AIC
Temperature
Type=ar(1) None 14396.5 4296.9
bimonth*year 14206.9 4107.3
bimonth*year/ type=ar(l) 14208.0 4108.4
bimonth*year/ group=group 14054.7 3955.1
bimonth*year/ type=ar(1) group=group - -
Type=ar(1).
sub=ﬁsh(group) None 10154.6 55.0
bimonth*year 10108.2 8.6
bimonth*year/ type=ar(1) - -
bimonth*year/ group=group 10099.6 0
bimonth*year/ type=ar(l) group=group - -
Depth
Type=ar( 1) None 25512.7 4582.2
bimonth*year - -
bimonth*year/ type=ar(1) 25498.6 4568.1
bimonth*year/ group=group 25458.1 4527.6
bimonth*year/ type=ar(1) group=group - -
Type=ar(1),
sub=ﬁsh(group) None 20938.6 8.1
bimonth*year 20933.7 3.2
bimonth*year/ type=ar(l) - -
bimonth*year/ group=group 20930.5 0
bimonth*year/ type=ar(1) ﬂit—group - -
Depth

The depths occupied seasonally also differed signiﬁcantly among lake Whiteﬁsh and

the two groups of lake trout (mixed model ANOVA, 1:12.183], P < 0.031). Because small

93

differences in depth can result in substantial differences in temperature if ﬁsh are located
near the therrnocline during summer stratiﬁcation, the differences were less consistent
than with temperature (Table 14). Judging ﬁom the relative size of the P values, the
underlying data were also considerably more variable. The largest differences between
lake Whiteﬁsh and FLO lake trout occurred in late July, August, and early November, and

with GLO lake trout in late July, early August, and early November (Table 14).

Daylight versus dark

In mixed model ANOVAS conducted separately by daylight and dark , temperatures
occupied by GLO or FLO lake trout and lake Whiteﬁsh differed in all comparisons (P <
0.001) (Table 15). The interaction of group with seasonal period was signiﬁcant (P =
0.063) only in the comparison of temperature between lake Whiteﬁsh and FLO lake trout
at night. The interaction terms in all other comparisons of temperature were not

signiﬁcant (P > 0.67, Table 15).

In mixed model ANOVAS conducted separately by daylight and dark, comparisons of
depths occupied between GLO or FLO lake trout and lake Whiteﬁsh were all different (P
< 0.01 to P < 0.001) (Table 15). The interaction of group with seasonal period was not

signiﬁcant in any of the depth comparisons (P > 0.67, Table 15).

94

Fig. 10. Frequency of hourly mean temperatures and hourly median depths by
half-month interval during daylight (sunrise to sunset) and dark (excludes twilight) for
Great Lakes origin lake trout (white ﬁll), Finger Lakes origin lake trout (light-gray ﬁll),
and lake Whiteﬁsh (dark- -gray ﬁll) 1n Lake Huron during 2002- 2005. In the box plots, the
horizontal line indicates the mean or median, the box the 25th and 75th percentiles, and the
whiskers the 10lh and 90th percentiles. Horizontal lines in panels B and D show modes for
depths of tribal gill net effort 1n 2003- 2005 (solid) and 2006- 2007 (dashed). Effort data
was provided by the Chippewa Ottawa Resource Authority, Sault Ste. Marie, Michigan.

 

6
14 A: Daylight- Temperature i

—=:>—
—r:=r:)—
—1:=:)—
—-—
—¢:D—
—¢:n—

Temperature (C)
8 :3 "

0010480900
-—

 

 

 

ll ,,111111 11,181,1th

B: Daylight — Depth 1

 

I

 

10

Emil III II
311 1:1 111 1111

 

l
111
I

80 I 1 1 | 1 l

 

 

.1...
ON

LL11)

 

 

Temperature (C)

 

 

 

oomemm

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Potential for reduced bycatch through depth restrictions of net gear

Seasonal differences in the distributions of lake Whiteﬁsh and lake trout exist in Lake
Huron, but are those differences large enough to spatially separate the species and be
useful in reducing bycatch? In the context of this question, differences in means become
less important than the range of depths occupied. Comparison of the depth distributions
(Fig. 10) indicated speciﬁc seasonal intervals existed when separation in depth could be
potentially useﬁil to ﬁshery regulation. The most pronounced separation occurred during

dark from mid-July through mid-September and again in November (Fig. 10).

The cumulative percent frequency of observations with depth indicated the proportion
of lake Whiteﬁsh, and of GLO and FLO lake trout, potentially available to gill nets
relative to a maximum depth limit for net sets (Table 16). From early July through
November, ﬁshing to a depth of 25 to 35 m (indicated by the horizontal lines in Table 16)
could target half the lake Whiteﬁsh population. At the same time, relatively low
percentages of lake trout would be targeted—most notably in late July (GLO = 8% and
FLO = 11%) and early August (GLO = 4% and FLO = 7%) (Table 16). Even at depths
where 75% of the lake Whiteﬁsh distribution is included in July—early September, only

about 25% of lake trout would be targeted in most cases.

These data (Fig. 10-1 1; Table 16), suggest that depth regulations might be used to

reduce lake trout bycatch by ﬁshing depths that would target the median depth occupied

by lake Whiteﬁsh while avoiding 75% or more of the lake trout. This approach would

96

 

$6 :6- 2 _._- 36 EN. _2 m4- and Ed. 2 N6. vi m.m _.m

 

 

 

2d 2d- 2 I- «dd 2... d2 2- odd 2d : md 5 3. dd 83
ad ad. 2 dd. add 2.2. E 2- odd 2d- : md- dd 2d 3
Rd 3d- 2 E. Sod 2d. 2: 2. Ed Ed. : dd. 3 3 dd 52
”Nd 2.2- 2 E- dev dad. 32 dd. 2d 2d. : 2- 2: dd 3
8d 3d. 2. 2. dev and. 2: 3. 3d 82- : 2-. ”.2 md dd so
dmd odd 2 2 dev 2.5.. n: 3. 8d 8d- 2 3. ed 2: we
ad ad. 2 d? dev Ed- 5 _.m- 3d dam- 2 3. dd 3 ed %m
2d 3d. 2 d. dev 23.. m2 2. odd add. 2 3. 3 3 dd
:d 22. 2 dd- :ddv add- 5 2. Rd ”S. 2 E- 2 2 3 2E
dd Rd- 2 3.. dev Be. a: 2. Rd 2:- 2 2- 2: ed 3
Ned odd. 2 ed- Edd Gd- m: 3. Rd 22- 2 E- dd ed dd .3
8d 2d. 2 ed. 8d SN- 2: m..- dmd 2d. 2 d- dd 2 3
3d 8.9 2 dd «md 22- 2: ed- 2d 2d. 2 ed- 3 dd 3 5:
mod mod- 2 3. 5d end. 22 3. 8d 8d- 2 dd dd 3. 3.
2d Rd- 2 d. 2d Ed- .3: 3. Ed ad 2 ad dd 3 ad 22
Ed and- 2 ed- 2d 2d- 22 Ed. 2d 8d 3 dd dd dd 3
8d 2d. 2 ed. Rd 2... 22 ed. 2d 8d. 2 _d- dd 2 2 a<
2d Ed. 2 ed. dd 8.7 2: ed- odd odd 2 dd 2 ed 3
2d 2d. 2 S- Rd 22. d: 3. Mad 8d. 2 dd S d d a:
2d 2d- 2 ad. end :2- 22 2d. 8d :d 2 _d .2 d d
2d dd. 2 d2- de 22- 22 ad. ad Ed 2 do E d d E
and odd- 2 d. 2d 22. 22 ad. ad Rd. 2 ad. 3 3 d2
mdd Sd. 2 ed. ”dd 32. m2 2. Ed 2d. 2 dd- 2 Z 8 :2
E a an 23 : 2 ed 23 E . do 23 1:5 020 one 585
Am 4 Am 4 Am 4 be:
”55 - 0.6 E: - 0.: 0.5 - o: Ge 23

 

.3225 some 553» memmto 50253 383 >835 803883 8:

ea do 232 2: 2a <>oz< 385-858 u add 3235 585.22 3 gear 93 a Edd @223 93 3 Ba .8: 83 52.6 52.5
3223 Swan 98 AOAOV Ewto 8di 3on 3 33:80 6% BBSoQEB .28 282: can 3:239: mo Gamay meeoE 0.83333 .3 035.

97

Fig. 11. Frequency of hourly maximum depths by half-month interval during daylight
(sunrise to sunset) and dark (excludes twilight) for Great Lakes origin lake trout (white
ﬁll), Finger Lakes origin lake trout (light-gray ﬁll), and lake Whiteﬁsh (dark-gray ﬁll) in
Lake Huron during 2002-2005. In the box plots, the horizontal line indicates the mean or
median, the box the 25'h and 75‘h percentiles, and the whiskers the 10th and 90th
percentiles. Horizontal lines in panels B and D show modes for tribal gill net effort in
2003-2005 (solid) and 2006—2007 (dashed). Effort data was provided by the Chippewa
Ottawa Resource Authority, Sault Ste. Marie, Michigan.

 

o
10 7 A: Daylight - Maximum Depth

 

 

 

 

 

 

 

 

 

appear to particularly protect FLO lake trout. If observations from the archival tags
could be directly converted to estimates of vulnerability, as in Table 16, restrictions
variable by month specifying the maximum depth gear could be ﬁshed would be a
valuable tool for management to reduce bycatch of the commercial ﬁshery. The
distribution of hourly median depths from tagged ﬁsh did not match the depth-effort data
provided by the Chippewa-Ottawa Resource Authority (Fig. 10). During daylight in June

through November, less than 25 percent of lake Whiteﬁsh were as deep as the

98

 

$6 and N2 w.~ mod K..— Nom I. and $6 mu mé Ndm aém mdm

 

 

 

Rd Rd 2: 2.” 8d 22 SR dd 2d Rd 2. E ddm ddm SR 83
2d 2: :_ 3. 8d Rd Rn Rd Rd Rd S 3 ”R ER _.R
8d Rd E 2: add Rd don _.2 Rd Rd- E 3- 3: 9R QR 32
Rd odd _: 3 add R2 20 E :d Rd R M: QR SN 2%.
23 and. d2 3. 2d Rd Rm dd odd Rd S 3 SN Sm Sm so
8d Rd. R. 3.. 2d 2: Rm 2 3d ddd. 2_ Nd ddm ddm Sm
Rd 8d 2: dd 8d d: 2% ad 3d Rd d2 3 ddm 9R d.R 3m
Rd Rd R. dd. 23 3d 2m 3 2d Rd .2 Rd Rdm QR NR
ddd NE ”2 2. Rod Rd m2, 3: Rd Rd 2_ ed wdm nR RR 22
8d Rd 2 2: add 2% ”2. q: Rd 8d :2 _d _.R QR d.R
mod dd. .2 2. Rd «S 3% 3, 2d Rd 02 3 «R 2d ddm .2
Rd dd ”2 ad Rod R: 0:. 2 2d :2 22 Rd .3 mdm ”R
Rd Rd. N: 3- 2d R2 Rd 3, Rd Rd 2: ad NR ”.2” SR :3
Rd Nd- R_ d. dwd Rd at, ad 8d 3d d: 2 SR QR 9R
Rd ad. 2: 2,. Ed Rd dR m..- dmd 8d N2 3 «R RR _.R 32
Rd Rd- 8m Rd- Rd Rd Rm 3 Rd Rd 22 dd RR ddm ER
Rd 8d Rd S Rd 8... d8 3. Rd 22 _2 3, RR ”R odd .3
8d 8d- Rm dd 2d Rd Rm 3 Rd R2 v2 3 RR NR ddv
dad 2d RN dd 2d 82 Rm dd Rd N2 82 3 NR QR mdv 3:
Rd 8d SN 3 2d Rd 0R Rd Rd odd § dd 3m QR mdv
8d 2d gm 3 2d 2: NR 3 2d Rd 8_ ed ”.3 RR Rdv 8;
Rd 8d RN R 2d 3: RR R Rd Rd 2: 3 «R OR a?
Rd Gd Rm 3 Rd 2: Rm 3 Rd 3d 2 3. ER NR Ev .5
E a d 23 E a do 23 E a dd Em.— 25 0.6 o: 585
Am < Am < Am < an:
23 made 0.6 “35 SEE 0d 0.6 358 0: Ed 23

 

.329? 8: 0.83 mcozooboo Eotomeom €285 some 553» mecO
5033 A282 REPS 289388 8.38m mo 232: Ba 28 <>OZ< 388.35.: a God EBBE 5:08.22 ,3 :05: 8.3 E ﬁaBEB
83 E 98 Bob 83 6.7: Ewto 8di Swarm 28 529 Ewto mod—3 805 3 83:80 25 23% BEBE do Rama 2223-382 .3 035.

99

Table 15. Results of mixed-model AN OVAs assessing differences by light condition
(daylight or dark) in the mean temperatures (Temp) and median depths (Depth) occupied
in Lake Huron during 2002-2005 by lake trout of two genetic origins, upper Great Lakes
origin (GLO) and Finger Lakes, New York origin (FLO), and lake Whiteﬁsh LWF. Data
were recorded with archival tags. The main effects were: Group (GLO vs. LWF or FLO
vs. LWF), Period (half-month seasonal periods), and the interaction of Group and Period.

 

 

Light Depen-
Condi- dent
tion variable Fixed effect (if F P > F
GLO lake trout vs. lake Whiteﬁsh
Daylight Temp Group 1, 63.2 34.12 <.001
Period 23, 62.5 21.77 <.001
Group*Period 23, 62.5 0.55 0.94
Depth Group 1, 30.9 7.38 0.01
Period 23, 30.9 2.14 0.03
GrouptPeriod 23, 30.9 0.56 0.71
Dark Temp Group 1, 50.6 45.26 <0.001
Period 23, 49.5 24.57 <0.001
Group*Period 23, 49.5 0.86 0.67
Depth Group 1, 30.4 15.83 <0.001
Period 23, 30.4 2.56 0.01
Group*Period 23, 30.4 0.82 0.68
FLO lake trout vs. lake Whiteﬁsh
Daylight Temp Group 1, 51.9 57.07 <0.001
Period 23, 51.6 22.11 <0.001
Group*Period 23, 51.6 0.84 0.67
Depth Group 1, 73.6 30.55 <0.001
Period 23, 73.2 2.05 0.01
Group*Period 23, 73.2 0.43 0.99
Dark Temp Group 1, 52.6 83.08 <0.001
Period 23, 52.3 24.75 <0.001
Group*Period 23, 52.3 1.67 0.06
Depth Group 1, 71.2 45.56 <0.001
Period 23, 70.8 1.94 0.02
Group*Period 23, 70.8 0.57 0.93

 

100

modal depth of effort in 2003-2005. Hence, it would appear that gear was set deeper than
most of the occupied depths recorded by the tags in lake Whiteﬁsh. At night the
percentage was even smaller, approaching 10 percent during some seasonal intervals. In

2006-2007, the modal depth for gill net effort was even deeper than in 2003-2005 and

deeper even than the 90th percentile for hourly median depths of lake Whiteﬁsh (Fig. 10).

The distributions of the hourly maximum depths occupied by lake whiteﬁsh (Fig. 11)
differ only slightly from the distributions of hourly median depths (Fig. 10) and also

failed to match the reported depths of gill net effort.

The hourly distribution of maximum depths in early August by 5-m depth interval
(Fig. 12) showed a pattern of diel movement which might explain the difference between
the modal depth of gill net effort and the depth distribution of lake Whitefish based on the
archival tag data (Fig. 10-11). During daylight, a higher frequency of observations
occurred at the outer range of the lake Whiteﬁsh depth distribution than at night (Fig. 12).
During dark, 1.9% of observations were in the 46- to SO-m depth range, compared to
7.3% during daylight. The lower percentage at night indicates that a group of ﬁsh must
be moving, either horizontally or vertically, in and out of deep waters in a diel pattern.
The archival tags lacked a means to establish horizontal position, so it was not possible to

say which direction of movement was responsible for the pattern.

101

the study and the following two years.

uring

Percents add to 100% within each hour. The area between the lines at 05:30 and 20:30 is

the approximate period of daylight. Arrows show the modal depths for tribal gill net

Fig. 12. Percent frequency of observatrons of maxrmum depth for combmatrons of
individual lake Whiteﬁsh, day, and hour during the ﬁrst half of august 2003-2005.
effort d

 

 

 

 

 

 

 

       

 

 

 

 

slim...
.J
4].
.1.‘ Esﬁémﬁgﬁﬂuﬂ
or. ,. gx§n\
, Q .. .iﬁxﬂmﬁg... m
1.4 o m
1 M...

505050
2211...

55:09“. “condom

ISCUSSIOI‘I

D

Ecological factors affecting depth and temperature distributions

Lake Whiteﬁsh and lake trout are both native residents of the Great Lakes, yet seasonally

show clear differences in depth and temperature distributions (Fig. 10). From a review of

the literature, Christie and Regier (1988) listed ranges for laboratory studies of

102

 

2: QB c2 02 o9 2: QB o3 oo— mZA

 

 

 

 

 

 

dd dd dd dd dd dd dd dd dd d:
dd dd dd dd dd dd dd dd dd 82 dd_
dd dd dd dd dd dd dd dd dd dd dd_
dd dd dd dd dd dd dd dd dd dd dd
dd dd dd dd dd dd ddd dd dd dd dd dd
dd dd dd dd dd dd dd dd dd dd dd dd
dd dd dd 2d 2d dd dd dd dd dd dd dd
dd dd 82 dd dd dd dd dd 2d _d dd dd dd
ddd dd dd dd dd dd dd dd dd dd dd dd dd dd
dd dd dd dd dd dd dd_ dd dd dd dd dd dd_ dd dd dd
dd dd dd dd dd dd dd dd dd dd dd dd dd dd dd dd
dd dd dd dd dd dd dd dd dd dd dd dd dd dd dd dd
dd dd dd dd dd dd dd dd dd dd dd dd dd dd dd dd
dd dd dd dd dd dd dd dd dd dd dd dd dd dd dd dd
dd dd _d dd dd dd dd 2 dd dd dd dd dd dd dd dd
dd 2 d dd d2 dd dd 2 dd dd 2 2 2d d_ dd dd
dd 2 d dd 2 dd d d : dd : d dd : dd dd
dd d d d d d dd d d dd d d dd d dd dd
d d d d d d dd d d d_ _ d : d d_ dd
d 2 d d d d dd 2 d : d d d d dd 2
d d d d d d d d d d d d d d d d_
d d d d d d d d d d d d d d d d
”:5 o: 0.5 “E: o: 0.5 $5 0.: 0.6 :5 0d 0.6 $2 0.: 0.5 Add
ddnEodmom 38m ddswsd. 83 ddsmsd. Edam bad. 83 b3. dram 53n—

 

doomdoom wad So: cad :dm 3323?: do 3235988 do.“ dwﬁ 3358 53>
33030 dﬁddd Edgxda no B23 .2025 83 5 £31: swung”? 8:: Ed .285 83 829 Ewuo $di 380 .Sod. 83 8.5V
Emto deadw— ddwﬁm Hod 29.5% 69% Ed do din: bag: 98 EZBE 5:08.22 .3 5:263 308% 0323830 .3 93:.

103

 

o3 o2 oo_ oo— 2: GS 00— mSA

 

 

 

 

 

 

 

 

 

 

dd dd dd dd dd dd d2 dd d:
dd dd dd dd ddd dd dd _ dd dd d2
dd dd dd dd dd dd dd dd dd d2
dd dd dd dd d2 dd dd dd dd dd dd
d2 dd dd dd d2 d2 dd dd dd dd dd dd dd dd
dd dd dd dd dd dd dd dd dd dd dd dd dd dd
dd dd dd dd dd dd 2d dd dd dd dd dd 5 dd
dd 5 2d dd dd dd dd dd dd dd dd d2 dd dd dd
dd dd dd dd dd dd dd dd dd dd dd dd dd dd dd
dd dd dd d2 dd dd dd dd dd dd dd 2d dd dd dd dd
dd dd dd dd dd dd 2d dd dd dd dd dd dd dd dd dd
dd dd dd dd dd dd dd dd dd dd dd dd dd dd dd dd
dd 2d dd dd dd dd dd dd dd 5 dd dd dd dd dd dd
dd dd dd 2d dd dd dd 2d dd dd dd dd dd dd dd dd
dd 2d dd dd dd dd dd dd dd dd dd dd dd dd dd dd
dd dd dd dd _d _d _d dd dd dd dd dd dd dd dd dd
dd 2 2 dd dd _d dd dd 2 dd dd _d dd dd 2 dd
dd 2 2 dd 2 2 dd 2 2 dd 2 2 dd 2 2 dd
2 2 2 dd 2 d dd 2 2 dd d d_ d d d dd
d d d dd d d dd d d d d : d d d 2
d d d dd _ _ 2 _ _ d d d d d d 2
d d d d d d d d d d _ d d d d d
.25 0.2 0.6 25 0.: So “55 02 30 ”SE 02 So 25 0.: 0.5 Add
822302 023 828302 Edam donodoo Bag 39900 Edam donEodmom 8d..— 595

 

6035.80 .3 033—.

104

temperature preference of lake Whiteﬁsh of 12.7 to 13.5°C (ignoring one value of 14 °C
they reported from a study of larval lake Whiteﬁsh) and optimal temperatures for grth
of 12 to 15.1°C. For lake trout, the values listed were 10.8 to 12.7 °C and 10 to 16 °C.
As discussed in Chapter Two, lake trout were rarely found at those temperaturds and
were likely responding to other variables. Similarly, lake whiteﬁsh were rarely found
between 12.7 and 135°C at any time of year (Fig. 10). In July-September, when strong
thermal stratiﬁcation is in place in Lake Huron and behavioral thermoregulation is
possible, the mean temperatures occupied by lake Whiteﬁsh ranged ﬁ'om 8.7 to 10.2 °C,
about two degrees lower than laboratory values. The most likely causw of a separation
in depth or temperature are related to differences in diet and separation of prey items,
and more speculative issues related to life history and evolution of temperature _

preferences.

Diet and food availability might be the most important variawa currently
controlling depth distributions of lake trout (Chapter Two) and lake Whiteﬁsh. Despite
the importance of lake Whiteﬁsh as a commercial specids in the Great Lakes for more
than a century, the best diet information is recent and related to changes in their diet in
Lake Michigan——believed to be the result of changes in the food web caused by the
invasion of Dreissenid mussels (Pothoven 2005). Similar changes in food webs have
occurred in Lake Huron (N alepa et al. 2007) and elsewhere in the Great Lakes (Hecky
et al. 2004); hence, the data of Pothoven (2005) are likely indicative of current diets in
Lake Huron. Except for one group of unusually large lake Whiteﬁsh sampled in

northwestern Lake Michigan in 2002-2005, ﬁsh were a rare item in the diet of lake

105

Whiteﬁsh (Pothoven 2005). Lake whiteﬁsh were predominantly benthivores, with
Dreissenid mussels recently becoming a prominent part of their diet. By comparison,
lake trout in Lake Huron are primarily piscivores (Fig. 8 in Chapter Two). Differences
in diet undoubtedly have affected, and continue to affect, the relative depth and

temperature distributions of lake trout and lake Whiteﬁsh.

The food web has changed throughout most of the Great Lakes due to effects of
phosphorus reductions and interception of nutrients by invasive Dreisenid mussels
(Hecky et al. 2004). The disappearance of Diporeia spp., formerly a key food item of
lake Whiteﬁsh has been notable in several lakes (N alepa et al. 2007). The recent
offshore movement of the commercial ﬁshery for lake Whiteﬁsh was thought to relate to
these changes in the food web (Mohr and Ebener 2005). Increasing dependence on
Dreissenid mussels would shift the lake Whiteﬁsh distribution closer to the depth of
greatest density of quagga mussels (Dreissenia bugensis), the more abundant and
deeper ranging of the Dreissenids. Quagga mussels were most abundant in 2003 at
depths of 31-50 m (N alepa 2007), consistent with modes of gill net effort in the tribal
ﬁsheries (Fig. 10-11). In this study, the least-square mean depths occupied by lake
Whiteﬁsh were generally around 30 m at the lower end of that depth range (Table 14).
Key changes in the prey base of lake trout in Lake Huron (Chapter Two) that might also
have inﬂuenced their depth distribution were the reduced importance of rainbow smelt
(Osmerus mordax) in the diet, the disappearance of alewife (Alosa pseudoharengus) a
former important diet item, and the appearance of a new prey item, the invasive and

demersal round goby (Neogobius melanostomus). During the period of this study, the

106

depth distribution of round gobies peaked at 37 m (Edward Roseman, USGS, Great
Lakes Science Center, unpublished data), similar to the least-square mean depth for lake
trout in late summer (Chapter Two; Table 16). The depths of both lake Whiteﬁsh and
lake trout are consistent with recent measures of the distributions of their major prey
items, lending support to a hypothesis that prey availability and distribution is an

important variable affecting lake whiteﬁsh and lake trout depth distributions.

Innate differences arising from post-glacial evolution undoubtedly exist in the
behavior of lake Whiteﬁsh to ﬁll a niche as a short-lived planktivore/benthivore, and of
lake trout to ﬁll a niche (primarily) as a long-lived piscivore. As suggested by Larsson
(2005), species regularly encountering low prey availability might evolve to occupy
lower temperatures and maximize efﬁciency of using food resources. As a long-lived
predator facing variation in both year-class strength and variability in prey availability
in the Great Lakes, lake trout might favor low temperaturds and deep water more than a
lake Whiteﬁsh, which feeds on a more consistent diet of benthos and zooplankton
(Pothoven 2005). Lake Whiteﬁsh can feed either on plankton or benthos, but shiﬁ
mainly to benthos if lake herring (C. artedi) are present (Carl and McGuiness 2006).
Lake herring historically supported the largest commercial ﬁshery catch in Lake Huron,
with lake Whiteﬁsh making up only 35 % of the total catch. Lake herring have been
decreasing in abundance in the main basin of Lake Huron since 1970 (Ebener and Mohr
2005) and currently are uncommon. Since the disappearance of lake herring, plankton
could be an increased part of the lake Whiteﬁsh in Lake Huron diet. But because

plankton abundance is low in the hypolirnnion, feeding on plankton in metalimnion or

107

epilimnion would be expected to result in a decrease in depth and increase in

temperature, which is contrary to the results presented in this chapter.

Diel movement and availability

Availability is one of the key variables governing catch per effort. Marr (1951)
deﬁned availability as “variations in catch not associated with abundance or ﬁshing
intensity, but rather with variations in environmental factors.” The effect of those
environmental factors is to change the distribution of a population over time, such that
the proportion located in areas where they are vulnerable to the ﬁshing gear changes.
Because his deﬁnition focuses on variation in catch, the concepts of spatial availability
are connected to vulnerability. To be spatially available to gill nets, lake Whiteﬁsh must
be within the range of bottom depths ﬁshed and within a distance from bottom covered
by the height of the net. The distributions of lake Whiteﬁsh based on archival tag data
(Fig. 10-11; Table 16) suggested that larger catches of lake whiteﬁsh and less bycatch
of lake trout could be achieved by ﬁshing shallower bottom depths than are currently
targeted by tribal gill net ﬁsheries. This would be true if all lake Whiteﬁsh were closely
associated with bottom. The distribution of gill net effort realized in the ﬁshery is a
result of trial and error by the ﬁshers and it seems highly unlikely that they would ﬁsh
in deep waters, and bypass a high density of ﬁsh if they could be caught closer to shore
in shallower waters. A more likely situation is that the population of lake Whiteﬁsh is
dispersed horizontally over a broad range of depths with a large portion suspended at
some distance above the bottom. Considering the small portion of the distribution at

depths consistent with the bottom depth being ﬁshed with gill nets, most of that

108

population might be suspended. In that case, the potential beneﬁts for decreasing
bycatch in Table 16 would be exaggerated unless an approach to ﬁshing gill nets at
controlled depths above the bottom were employed. The potential cause for the lack of
discovery of this distribution by the commercial ﬁshery could be due to the difﬁculty in

setting suspended nets well off the bottom.

The apparent pattern of diel movement ﬁ'om shallow to deep waters in daylight (Fig.
12) might also partially explain the depth range being ﬁshed. To be caught in passive
gear like a gill net, a ﬁsh must move to intersect the gear. Because no information
exists on geographical position of the tagged ﬁsh, it is unclear whether the movement
observed (Fig. 12) was a local descent to the bottom by a group of ﬁsh remaining in the
same general location or an oblique movement up and down the slope along bottom by
a group of ﬁsh that remains within the height of the gill nets. In the case of vertical
movement, random foraging movements would result in capture during the time spent
near bottom, which is during daylight, inconsistent with the general belief that gill nets
primarily catch at night. In the case of movement along the slope, movement would be
directional between depths, and near the bottom, where ﬁsh might be intercepted by gill
nets. Intercepting such a movement could be particularly effective if nets were not
oriented directly down the slope. Catch in this case could occur in twilight or darkness,
when ﬁsh are leaving ﬁom, or returning to, shallower depths. In either case, based on
the analysis presented here, a relatively small proportion of the population was using the

depths being ﬁshed and providing the catch to the ﬁshery (Fig. 12).

109

Day and night bottom and midwater trawl catches made along contour during 2004
and 2005 at depths of 30, 60, and 120 m in the Apostle Islands area of Lake Superior
provide evidence that diel movements up and down the slope occur (Yule et al. 2008).
Bottom trawls ﬁshed at 30 m during the night caught a wide range of lengths of lake
Whiteﬁsh. At the same depth during the day only a few small and no large lake
Whiteﬁsh were caught (Yule et al. 2008). At 60 m, trawls caught primarily large lake
Whiteﬁsh (> 300 mm) at night in contrast to primarily small lake whiteﬁsh during the
day. The implication is that movement between those depths occurs between day and
night. Using the Lake Huron archival tag data presented here, vertical versus horizontal
movement could not be ruled out. The relative catches in day and night bottom trawls
in Lake Superior (Yule et al. 2008) and previous nighttime acoustic and midwater trawl
surveys at the same sites and years (reported in greater detail by Yule et al. 2007),
showed that few lake Whiteﬁsh were pelagic (Yule et al. 2008). This suggests that lake
whiteﬁsh moved up and down the slope and were associated with bottom. Yule et al.
(2008), however, did not comment on the possibility of suspension of lake Whiteﬁsh
during the day, as suggested by the archival tag data of the present study (Fig. 12). The
slope in the Apostle Islands area of Lake Superior is steeper and the distances
associated with a change in depth less than is typical in northern Lake Huron and these
may be variables that could cause differences in the lake Whiteﬁsh behavior between the

two lakes.

Management recommendations

110

A number of management actions have been adopted in Lake Huron to regulate gear
type and use to restrict bycatch or reduce mortality of lake trout (Brown et al 1999).
Since the late 19703, the State of Michigan has banned large-mesh gill nets and has only
permitted the use of trap nets (Brege and Kevem 1978). Under the year 2000 consent
decree governing ﬁshing in 1836 treaty waters, the tribes agreed to ﬁsh only trap nets
south of the Hammond Bay Reﬁrge Harbor, which includes Hammond Bay (U .S.
District Court 2000). All lake trout caught in trap nets must be released. Dead lake
trout not retained by the gill-net ﬁshery would be wasted, and a quota was set for
retention by the tribal gill-net ﬁshery. Because lake trout are the principal bycatch in
gill nets set to capture lake Whiteﬁsh (McNeil and deLaplante 1989; Johnson et al.
2004a), managers in Lake Huron conducted studies to examined both the survival of

ﬁsh released after capture and ways to minimize bycatch (Johnson et al. 2004b).

The number retained or discarded dead was measured in 260 monitored gill net and
96 trap net lifts in the Hammond Bay area in 1998 and 1999 (Johnson et al. 2004b). In
both gear types, the ratio of lake trout to legal lake Whiteﬁsh was highest in summer and
lowest in fall. Bycatch mortality expressed as lake trout killed per lift was highest in
spring for gill nets and in the summer for trap nets. The fates of lake trout collected
from Lake Huron trap nets for use in Chapter One were used to assess delayed mortality
of lake trout released alive from the Lake Huron trap net ﬁsheries. Only three of 186
lake trout taken from trap nets and transported to the HBBS, implanted with archival
tags, and held for a day before release did not survive the transport and handling

resulting in an estimate of only 1.6% delayed mortality for ﬁsh taken alive from the trap

11]

net pot (Johnson et al 2004b). Even though percent mortality of lake trout released
from trap nets was low, the number caught can be seasonally high and hence the catch
in gill nets and trap nets can make an important contribution to total mortality. Using
rates for losses from gill nets and trap nets in the study, Johnson et al. (2004b) estimated
that 10 150 kg of lake trout would be killed in a gill net ﬁshery catching 100 000 kg of
lake Whiteﬁsh and 4 900 kg in a 345 00 kg trap net ﬁshery. Either loss was deemed

substantial.

One management option to reduce losses to bycatch would be to switch entirely to
trap nets. However, tribal ﬁshers are afforded the right to use gill nets under the year
2000 consent decree (US. District Court 2000) and therefore this strategy in not viable
unless tribal ﬁshers choose independently to do so. A second option is to reduce ﬁshing
effort, but the lake Whiteﬁsh p0pu1ation in Lake Huron is strong and restricting effort is
not needed to protect stocks (Ebener 1997). The most palatable means to reduce
bycatch was to redirect effort toward seasons and depths where ﬁshing is economically
feasible and the overlap of lake Whiteﬁsh and lake trout distributions is small. Based on
results from Johnson et al. (2004b), a limited small-boat gill net ﬁshery was permitted
October 1 through November 8, and again in December in Hammond Bay south of the

regular boundary for gill nets.

The data presented here and in Chapter 2 suggest that depth distributions of lake

trout and, most likely, lake Whiteﬁsh have changed since 1999 and that the beneﬁts of

seasonal and depth restrictions should be reexamined. From early July through

112

November, ﬁshing at a depth between 25 to 35 m could target 50% of the lake Whiteﬁsh
population while exposing less than 25% of the lake trout populations to the ﬁshery
(Fig. 10-11; Table 16). However, concern remains that the measures of depths ﬁ'om the
archival tags may not have distinguished between ﬁsh that are suspended versus those
closely associated with bottom and vulnerable to gill nets. For lake trout, this issue is
not a concern, because ﬁsh located deeper than gill net sets would not be available to
those nets regardless of their vertical distribution. Viability of the depth and season
regulations suggested above could be determined through an experimental ﬁshery to
answer whether lake Whiteﬁsh could be caught in economically viable numbers at those

depths, while avoiding lake trout.

Based on archival tag data (Fig. 10-12), a considerable portion of the lake Whiteﬁsh
population was located shallower in summer than the depths currently being ﬁshed with
bottom-set gill nets, suggesting that lake Whiteﬁsh may be predominately suspended. If
lake Whiteﬁsh are more pelagic in Lake Huron than previously thought, new
opportunities for exploitation with suspended gill nets could exist—if a logistically
feasible method of achieving that could be found. Suspended gill nets have been used
in research to target pelagic coregonines (e. g., Milne et al. 2005). In the Great Lakes,
the lakewide assessment plan for Lake Michigan ﬁsh communities (Schneeberger et al.;
available on the Great Lakes Fishery commission web site,
http://www.glfc.org/pubs/SpecialPubs/lwassesO1.pdf), gives speciﬁcations for the
suspended gill nets being used to sample Chinook salmon and rainbow trout, although it

is noted that the suspended nets are bulky, require a net reel, and may not be adaptable

113

to some commercial vessels. On suitable vessels, commercial use of suspended gill nets
can be viable, as evidenced by their use in the late-fall ﬁshery for herring roe on Lake

Superior (Stockwell et al. in review).

Feasibility of this suggestion rests both on the depth distribution of lake Whiteﬁsh
and on their geographical distribution, which was not measured in this study. If pelagic
lake Whiteﬁsh are dispersed broadly, densities may be too low to support a ﬁshery.
Additional research would be required to determine whether there are concentrations of

suspended lake whiteﬁ sh spatially dense enough to support a commercial ﬁshery.

Use of archival tags

The continuity of data collected with archival tags led to production of new typed of
information, such as the plots of the distributions of individuals in Fig. 10-12 and Table
16. Because records are not interrupted by ﬁsh going too deep or because of weather,
as with conventional telemetry, complete data sets were available from release to
recapture. The continual recording eliminates the uncertainties that exist in net
captures, where the knowledge is limited to the presence of a ﬁsh in the area of a single
trawl tow or over the duration of a gill—net set. Within the limitations of archival tag
capabilities versus information needs, archival tags can provide much more complete

and unambiguous descriptions of behavior than net surveys.

114

A further suggested use of these data is to improve models of ﬁsh behavior, growth,
or food consumption into the more realistic form of an “Individual Based Model” or
IBM. MacKay (1992) ﬁrst used an IBM approach to improve the Kitchell and Breck
(1980) and Kitchell (1990) models of sea lamprey feeding and host mortality. This
model was further improved when Madenjian et al. (2003) incorporated more speciﬁc
information on individual variability in sea lamprey growth (Bergstedt and Swink 1995)
and seasonal temperatures occupied by lake trout hosts (Bergstedt et al. 2003). Similar
improvements in models of ﬁsh growth and energy or prey requirements are possible
using the data from Chapters Two and Three. Temperature is an important variable
affecting metabolism and food conversion (Brett et al. 1969) and summary statistics can
mask broad variation in thermal behavior (Fig. 10-11). It is now possible to create
IBMs of growth and prey needs of lake whiteﬁsh and lake trout that incorporate the
documented behavior from tagged individuals and relate that variation to documented
weight, sex, and maturity at tagging and capture. Measures of observed growth of
individual ﬁsh combined with a record of the temperatures they occupied while at large
in the Great Lakes have not previously been available. The data in this dissertation

should be exploited for this purpose.

Lake Whiteﬁsh, in this study, were more susceptible to the stress of handling and
tagging than lake trout. The poor return of lake Whiteﬁsh tagged at the HBBS in 2002
suggested that lake Whiteﬁsh were not capable of surviving the combined stresses of
capture, transportation, surgical procedures, and release. As with lake trout (Chapter 2)

the best approach was to tag them at the site of the trap net. This approach limited

115

tagging to days with light winds, but also minimized stress to lake Whiteﬁsh caused by
implantation of the tags. Minimizing stress will be critical to success with future lake

Whiteﬁsh studies using archival tags.

Acknowledgments

I thank Mary K. Jones, Karen Slaght, Dave Keffer, and Erick Larson of the
Hammond Bay biological Station for their professionalism and support in the conduct
of this study, Mark Ebener for his help in coordinating ﬁsh collections from the tribal
ﬁsheries and helping to focus my thinking regarding lake Whiteﬁsh management, and
Jim Bence for assistance with statistical analyses. Funding for this study was provided
by the US. Geological Survey, the US. Fish and Wildlife Service through the Great

Lakes Fishery Trust, and Great Lakds Fishery Commission.

116

 

References

Bergstedt, R.A., and Swink, W.D. 1995. Seasonal growth and duration of the parasitic
life stage of the landlocked sea lamprey (Petromyzon marinus). Can. J. Fish. Aquat.
Sci. 52: 1257-1264.

Bergstedt, R.A., Argyle, R L., Seelye, J .G., Scribner, K.T., and Curtis, G.L. 2003. In
situ determination of the annual thermal habitat use by lake trout (Salvelinus
namaycush) in Lake Huron. J. Great Lakes Res. 29(Suppl. 1): 347-361.

Berst, AH. and Spangler, GR. 1972. Lake Huron: Effects of exploitation,
introductions, and eutrophication on the salmonid community. J. Fish. Res. Bd.
Canada 29: 877-887.

Brege, DA, and Kevem, NR. 1978. Michigan commercial ﬁshing regulations: a
summary of public acts and Conservation Commission orders, 1865-1975. Michigan
Sea Grant College Program, Report MICHU-SG-78-605, Michigan State
University, East Lansing, Mich.

Brett, J.R., Shelbourn, J .E., and Shoop, CT. 1969. Growth rate and body composition
of ﬁngerling sockeye salmon, Oncorhynchus nerka, in relation to temperature and
ration size. J. Fish. Res. Board Can. 26: 2363-2394.

Brown, R.W., Ebener, M., and Gorenﬂo, T. 1999. Great Lakes commercial ﬁsheries:
historical overview and prognosis for the future. In Great Lakes ﬁsheries policy and
management, a binational perspective. Edited by W. Taylor, and CR Ferreri.
Michigan State University Press, East Lansing, Mich. pp. 307-3 54.

Carl, LM., and McGuiness, F. 2006. lake Whiteﬁsh and lake herring population
structure and niche in ten south-central Ontario lakes. Environ. Biol. Fishes 75:
315-323.

Chopin, F., Inoue, Y., and Arimoto, T. 1996. Development of a catch mortality model.
Fisheries Research 25: 377-382.

Christie, GO, and Regier, HA. 1988. Measures of optimum thermal habitat and their
relationship to yields for four commercial ﬁsh species. Can. J. Fish. Aquat. Sci. 45:
301-314.

Crowder, L. B., and Murawski, SA. 1998. Fisheries bycatch: implications for
management. Fisheries 23: 8-17.

Ebener, M. P. 1997. Recovery of lake Whiteﬁsh populations in the Great Lakes.
Fisheries. 22(7): 18-20.

117

Ebener, M.P., and Mohr, LC. 2005. The coregonine community. In The state of Lake
Huron in 1999. Edited by MP. Ebener. Great Lakes Fishery Commission Special
Publication, Ann Arbor, Mich. pp. 47-64.

Elrod, J .H., O’Gorman, R., and Schneider, OR, and Schaner, T. 1996a. Geographical
distributions of lake trout strains stocked in Lake Ontario, 1978-1993. J. Great
Lakes Res. 22: 871-883.

Elrod, J .H., O’Gorman, R., and Schneider, C.P. 1996b. Bathythermal distribution,
maturity, and growth of lake trout strains stocked in U. S. waters of Lake Ontario,
1978-1993. J. Great Lakes Res. 22: 722-743.

Gallinat, M. P., Ngu, H.H., and Shively, U.D. 1997. Short-term survival of lake trout
released from commercial gill nets in Lake Superior. N. Am. J. Fish. Manag. 17:
136-140.

Hansen, M.J. 1999. Lake trout in the Great Lakes: basinwide stock collapse and
binational restoration. In Great Lakes ﬁsheries policy and management, a binational
perspective. Edited by W. Taylor, and GP. Ferreri. Michigan State University Press,
East Lansing, Mich. pp. 417-453.

Hecky, R.E., Smith, R.E.H., Barton, D.R., Guildford, S.J., Taylor, W.D., Charlton,
M.N., and Howell, T. 2004. The nearshore phosphorus shunt: a consequence of
ecosystem engineering by dreissenids in the Laurentian Great Lakes. Can. J. Fish.
Aquat. Sci. 61: 1285-1293.

Johnson, J .E., Jonas, J ., and Peck, J .P. 2004a. The problem of bycatch management,
with emphasis on lake trout bycatch in 1836 Treaty waters of the upper Great Lakes.
Mich. Dept. of Nat. Res., Fish. Res. Rep. 2070, Ann Arbor, Mich.

Johnson, J .E., Ebener, M.P., Gebhardt, K., and Bergstedt, R.A. 2004b. Comparison of
catch and lake trout bycatch in commercial trap nets and gill nets targeting lake
Whiteﬁsh in northern Lake Huron. Michigan Department of Natural Resources,
Fisheries Research Report 2071, Ann Arbor, Mich.

Kitchell, J .F. 1990. The scope for mortality caused by sea lamprey. Trans. Am. Fish.
Soc. 119: 642-648.

Kitchell, J .F., and Breck, J .E. 1980. Bioenergetics model and foraging hypothesis for
sea lamprey (Petromyzon marinas). Can. J. Fish. Aquat. Sci. 37: 2159-2168.

Krueger, C.C., Marsden, J .E., Kincaid, H.L., May, BL. 1989. Genetic differentiation
among lake trout strains stocked into Lake Ontario. Trans. Amer. Fish. Soc. 118:
317-330.

118

 

Larsson, S. 2005. Thermal preference of Arctic char, Salvelinus alpinus, and brown
trout, Salmo trutta,——implications for their niche segregation. Environ. Biol. Fishes
73: 89-96.

MacKay, NA. 1992. Evaluating the size effects of lampreys and their hosts: application
of an individual based model. In Individual-based models and approaches in
ecology: populations, communities, and ecosystems, eds. D. DeAngelis and L.
Gross, pp. 278 294. New York: Chapman & Hall.

Madenjian, C.P., Cochran, P.A., and Bergstedt, R.A. 2003. Seasonal patterns in growth,
blood consumption, and effects on hosts by parasitic-phase sea lampreys in the
Great Lakes: an individual-based model approach. J. Gt. Lakes Res. 29 (Suppl. 1):
332-346.

Marr, J .C. 1951. On the use of the terms abundance, availability and apparent
abundance in ﬁshery biology. Copeia, 2: 163-169.

McNeil, PI, and deLaplante, E]. 1989. Estimated catches of rainbow trout, brown
trout, lake trout, and Paciﬁc salmon in commercial gillnets, Lake Huron, 1988.
Ontario Ministry of Natural Resources, Lake Huron Fisheries Assessment Unit,
report 3-89, Owen Sound, Ontario.

Milne, S.W., Shuter, B.J., and Sprules, W.G. 2005. The schooling and foraging ecology
of lake herring (Coregonus artedi ) in Lake Opeongo, Ontario, Canada. Can. J. Fish.
Aquat. Sci. 62: 1210-1218 (2005)

Mohr, LC, and Ebener, MP. 2005. Status of Lake Whiteﬁsh (Coregonus
clupeaformis) in Lake Huron. Gt. Lakes Fish. Comm. Tech. Rep. No. 66: 105-125.

Nalepa, T.F., Fanslow, D.L., Pothoven, S.A., Foley, A.J., and Lang, GA. 2007. Long-
term trends in benthic macroinvertebrate populations in Lake Huron over the past
four decades. J. Great Lakes Res. 33: 421—436.

Pothoven, SA. 2005. Changes in lake Whiteﬁsh diet in Lake Michigan, 1998-2001. Gt.
Lakes Fish. Comm. Tech. Rep. No. 66:127—140.

SAS Institute Inc. 2001. SAS release 8.02 ed. SAS Institute Inc., Cary, NC.

Schleen, L.P., Christie, G.C., Heinrich, J .W., Bergstedt, R.A., Young, R.J., Morse, T.J.,
Lavis, D.S., Bills, T.D., Johnson, J .E., and Ebener, MP. 2003. Development and
implementation of an integrated program for control of sea lampreys in the St.
Marys River. J. Great Lakes Res. 29(Suppl. 1): 677-693.

Smith, B.R., and Tibbles, J .J . 1980. Sea lamprey (Petromyzon marinas) in Lakes

Huron, Michigan, and Superior: history of invasion and control, 1936-78. Can. J.
Fish. Aquat. Sci. 37: 1780-1801.

119

Stockwell, J .D., Ebener, M.P., Black, J .A., Gorman, O.T., Hrabik, T.R., Kinnunen,
R.E., Mattes, W.P., Oyadomari, J .K., Schram, S.T., Schreiner, D.R., Seider, M.J.,
Sitar, SP, and Yule, D.L. In review. A synthesis of cisco recovery in Lake
Superior: implications for native ﬁsh rehabilitation in the Laurentian Great Lakes.
N. Am. J. Fish. Manag.

United States District Court. 2000. United States, Bay Mills Indian Community, Sault
Ste. Marie Tribe of Chippewa Indians, Grand Traverse Band of Ottawa and
Chippewa Indians, Little River Band of Ottawa Indians, and Little Traverse Bands
of Odawa Indians v. State of Michigan. Consent Decree, Case Number 2:73 CV26,
Western District of Michigan, Southern Division, Kalamazoo, Mich.

Yule, D.L., Adams, J .V., Stockwell, J .D., and Gorman, OT. 2007. Using multiple gears
to assess acoustic detectability and biomass of ﬁsh species in Lake Superior. N. Am.
J. Fish. Manag. 27 : 106-126

Yule, D.L., Adams, J .V., Stockwell, J .D., and Gorman, OT. 2008. Factors affecting

bottom trawl catches: implications for monitoring the ﬁshes of Lake Superior. N.
Am. J. Fish. Manag. 28: 109-122.

120