INFLUENCE OF LIFE HISTORY TRAITS ON ACCUMULATION OF POLYBROMINATED
DIPHENYL ETHERS AND POLYCHLORINATED BIPHENYLS IN THREE LAKE TROUT
POPULATIONS FROM LAKE SUPERIOR

By
Laura Christine Claus

A THESIS
Submitted to
Michigan State University
In partial fulfillment of the requirements for the degree of
MASTER OF SCIENCE

Fisheries and Wildlife
2011

ABSTRACT
INFLUENCE OF LIFE HISTORY TRAITS ON ACCUMULATION OF POLYBROMINATED
DIPHENYL ETHERS AND POLYCHLORINATED BIPHENYLS IN THREE LAKE TROUT
POPULATIONS FROM LAKE SUPERIOR
By
Laura Christine Claus
Bioaccumulation of lipophilic contaminants can be influenced by many factors, including
life history. Life history traits, such as age at maturity and reproductive schedules evolve to
optimize the allocation of available energy to reproduction and growth, which may have
implications for the distribution and effects of lipophilic contaminants. The goal of this study
was to understand how life history traits can influence the distribution of lipophilic contaminants
in three distinct lake trout (Salvelinus namaycush) populations from Lake Superior: siscowet and
lean morphotypes from Marquette and Michipicoten Island leans. The siscowet morphotype is
characterized by higher body fat content, slower growth and lower fecundity than the lean. The
Michipicoten Island lean is a slow growing lean population with high reproductive investment at
older ages. The concentrations of PCBs and PBDEs in the muscle, ovary and liver were
measured and compared to population relevant metrics such as fecundity, muscle fat, length and
age. As predicted, based on higher body fat, siscowets had higher concentration of contaminants
in the muscle tissue when adjusted for exposure. The Michipicoten leans had relatively higher
concentrations of contaminants in the ovary tissue compared to siscowets of similar ages, which
is consistent with their higher fecundities and gonad weights. The patterns of contaminant
bioaccumulation between the three populations suggest that it is necessary to consider life
history traits when monitoring for lipophilic contaminants because such patterns allow for the
inference of the effects of contaminants on the general health of the fish population.

ACKNOWLEDGMENTS

I begin by thanking Cheryl Murphy who spent countless hours working with me on this
project and providing insight and guidance throughout this process. I also thank Dan Jones,
whose help and support with the GC/MS analyses was invaluable. I thank Mohamed Fasial who
provided me with valuable insight to this project. I thank the Michigan Department of Natural
Resources and Environment for providing necessary equipment and guidance with fish sampling
and age data, especially Shawn Sitar. I also thank Rick Goetz and Dan Rosauer for their
assistance with sampling. I thank the Ontario Ministry of the Natural Resources for sampling
near Michipicoten Island. I thank Beverly Chamberlin at the Michigan State University Mass
Spectrometry Facility for her endless support with the use of the GC/MS equipment. I thank Ben
Felt for his trips into Canada and the U.P. as part of sampling trips for this research and his work
on the fecundity data. I thank Anna Boegehold for her work in the lab. I thank Lindsey Hartrey
for her work with the lipid extractions. I thank Ben Schmitt, Dane Larsen and the other Murphy
lab members who have assisted with this research. I also thank the members of the MIRTH lab
who provided support and friendship throughout this process. Finally, I thank my parents, my
sister, Jenni and my fiancé, Matt for their constant love and support throughout all of my
education.
Funding for this research was provided through a grant from the Center for Water
Sciences at Michigan State University. Additional funding was provided by the Center for
Integrated Toxicology at Michigan State University. Funding for travel and additional analyses
was provided through the Dr. Howard A. Tanner Fisheries Excellence Fellowship.

iii

TABLE OF CONTENTS

LIST OF TABLES…………………………………………………………………....…...v
LIST OF FIGURES………………………………………………………………………vi
LIST OF SYMBOLS OR ABBREVIATIONS………………………….………...…..…xi
INTRODUCTION…………………………………………………….…………………..1
METHODS…………………………………………………………………..……………9
Field Sampling……………………………………………………….………........9
Contaminant Analysis……………………………………………………....……13
Column Construction…………………………………………………….13
Contaminant Extraction………………………………………………….14
Quality Control…………………………………………………………..16
Contaminant Quantification……………………………………………...19
Lipid Extraction.........................................................................................22
Statistical Analysis.................................................................................................24
ANCOVA..................................................................................................24
Multiple Regression...................................................................................25
RESULTS………………………………………………………………………………..29
Life History Results...............................................................................................29
Contaminant Results..............................................................................................42
DISCUSSION……………………………………………………………..……………..57
Life History Influences..........................................................................................58
Liver Concentration Patterns.................................................................................61
Muscle Concentration Patterns..............................................................................65
Ovary Concentration Patterns................................................................................66
Habitat Specific Influences....................................................................................67
Health Effect of Contaminants...............................................................................70
Assumptions, Limitations and Future directions...................................................72
Implications for Monitoring...................................................................................73
Implications for Management: Multiple Stressors.................................................76
Conclusions............................................................................................................79
LITERATURE CITED…………………………………………………………………..81

iv

LIST OF TABLES

Table 1. Standards and congeners used for analysis…………….……………………………….18
Table 2. Amount of contaminants found in blanks and the average percent recoveries for the
PAR samples……………………………………………………………………………........19
Table 3. Detection Limits and Limits of Quantification…………………………………………21
Table 4. List of statistical tests used……………………………………………………………..28
Table 5. Differences in life history characteristics based on ANCOVA………………………...32
Table 6. Best models based on AIC regression method…………………………………………43
Table 7. Differences in PBDE and PCB concentrations in the liver tissue based on ANCOVA
results……………………………………………………………………………………...…45
Table 8. Differences in PBDE and PCB concentration in the muscle tissue based on ANCOVA
results…………………………………………………………………………………...……49
Table 9. Differences in PBDE and PCB concentrations in the ovary tissue based on ANCOVA
results………………………………………………………………………………………...54

v

LIST OF FIGURES

Figure 1. The phenotypic differences between lean and siscowet lake trout (courtesy Shawn
Sitar)……………………………………………………………………………………….......7
Figure 2. A) Map of Lake Superior. Large circles indicate sampling locations (map courtesy of
mspmag.com). B) Location of two sites of gill net sets off the coast of Marquette. The
numbers on the map indicate the depth in meters at each site (map courtesy of Ralph Tingley
III). C) Locations of gill nets set for lake trout sampling near Michipicoten Island (map
courtesy of the Ontario Ministry of Natural Resources)…………………..…………………10
Figure 3. Location of the four fatmeter readings taken along right lateral body. Readings were
then calibrated with lipid extraction data to generate muscle fat percentages for lake trout.
(Photograph courtesy of Sara Smith)……………………………………………….……......11
Figure 4. Histogram of the age ranges (years) of each population. The vertical axis contains the
frequency counts for each age range. Marquette leans (n=9), Marquette siscowets (n=12) and
Michipicoten Island leans (n=10)………………………….……………...............................29
Figure 5. The marginal mean fork length (FL) and FL-at-age relationships were compared
between the Marquette leans (n=9), siscowets (n=12) and Michipicoten leans (n=10). A)
Comparison of the marginal mean fork lengths (mm) between lake trout populations. Error
bars indicate the 95% confidence interval of the population marginal means. Letters indicate
populations that are significantly different based on the results of an ANCOVA. B)
Relationship between fork length (mm) and age (years) of the three lake trout populations.
2

Regression lines are included for each population (Michipicoten leans: β=0.02, R =0.88;
2

2

Marquette leans: β=0.02, R =0.21and siscowets: β=0, R =0). Letters indicate populations
that have significantly different slopes of the regression line based on the results of an
ANCOVA.....…………………………………………………………………………...……31
Figure 6. The marginal mean body weights and the relationships between weight and age were
compared between the populations: Marquette leans (n=9), siscowets (n=12) and
Michipicoten leans (n=10). A) Comparison of population marginal mean body weights (g)
between lake trout populations. Error bars indicate the 95% confidence interval of the
population marginal means. Letters indicate populations that are significantly different based
on the results of an ANCOVA. B) Relationship between body weight (1000 g) and age
(years) between the Marquette lean, siscowet and Michipicoten lean lake trout populations.
2

Regression lines are included for each population (Michipicoten leans: β=0.07, R =0.80;
2

2

Marquette leans: β= 0.04, R =0.14 and siscowets: β=0, R =0). Letters indicate populations

vi

that have significantly different slopes of the regression line based on an ANCOVA. There
were no significant differences between slopes….....................................……….………34
Figure 7. The marginal mean fecundities and the relationships between fecundity and age were
compared between the populations: Marquette leans (n=9), siscowets (n=12) and
Michipicoten leans (n=11). A) Comparison of population marginal mean fecundities
(number of eggs) between lake trout populations including data from Michipicoten lake trout
that was not included in this study. Error bars indicate the 95% confidence interval of the
population marginal means. Letters indicate populations that are significantly different to an
alpha value of 0.05 based on the results of an ANCOVA. B) Relationship between total
fecundity (1000 eggs) and age (years) between lean, siscowet and Michipicoten lake trout
populations. Regression lines are included for each population (Michipicoten leans: β=0.33,
2
2
2
R =0.03; Marquette lean: β=0.12, R =0.27; siscowet: β=0, R =0). There were no significant
differences between slopes…………………………………….…………..................……...36
Figure 8. The marginal mean gonad weights and the relationships between gonad weight and age
were compared between the populations: Marquette leans (n=9), siscowets (n=12) and
Michipicoten leans (n=11). A) Comparison of population marginal mean gonad weights (g)
between lake trout populations including data from Michipicoten lake trout that was not
included in this study. Error bars indicate the 95% confidence interval of the population
marginal means. Letters indicate populations that are significantly different to an alpha value
of 0.05 based on the results of an ANCOVA. B) Relationship between gonad weight (g) and
age (years) between the lake trout populations. Regression lines are included for each
2
2
population (Michipicoten leans: β=0.01, R =0.01; Marquette leans: β=0.11, R =0.27,
2

siscowets: β= -0.01, R =0.01). There were no significant differences between
slopes...................................................................................................………………...…..…37
Figure 9. The marginal mean fecundities adjusted for body weight and the relationships between
adjusted fecundity and age were compared between the populations: Marquette leans (n=9),
siscowets (n=12) and Michipicoten leans (n=11). A) Comparison of population marginal
mean fecundity adjusted for body weight (total eggs/g) between lake trout populations
including data from Michipicoten lake trout that was not included in this study. Error bars
indicate the 95% confidence interval of the population marginal means. Letters indicate
populations that are significantly different to an alpha value of 0.05 based on the results of an
ANCOVA. B) Relationship between fecundity adjusted for body weight (total eggs/g) and
age (years) between the lake trout populations. Regression lines are included for each
2
2
population (Michipicoten leans: β= -0.03, R =0.29; Marquette leans: β=0.08, R =0.26;
2

siscowets: β=0.01, R =0.02)………………………………………......………………..……38
Figure 10. The marginal mean gonadosomatic indices (GSI) and the relationships between GSI
and age were compared between the populations: Marquette leans (n=9), siscowets (n=12)
and Michipicoten leans (n=11). A) Comparison of population marginal mean GSIs between
lake trout populations including data from Michipicoten lake trout that was not included in
this study. Error bars indicate the 95% confidence interval of the population marginal means.
Letters indicate populations that are significantly different to an alpha value of 0.05 based on
the results of an ANCOVA. B) Relationship between GSI and age (years) between the lake
vii

trout populations. Regression lines are included for each population (Michipicoten leans: β=
2
2
2
-0.39, R =0.26, Marquette leans: β=0.75, R =0.28, siscowets: β= -0.09, R =0.06). There
were no significant differences between slopes.........…………………………………..…... 39
Figure 11. The marginal mean lipid concentrations and the relationships between lipid
concentrations and age were compared between the populations: Marquette leans (n=9),
siscowets (n=12) and Michipicoten leans (n=10). A) Comparison of population marginal
mean lipid concentrations between lake trout populations. The lipid concentrations were
taken from muscle tissue and reported as a percentage in wet weight. Error bars indicate the
95% confidence interval of the population marginal means. Letters indicate populations that
are significantly different to an alpha value of 0.05 based on the results of an ANCOVA. B)
Relationship between lipid concentration (% lipid wt/w) and age (years) between the lake
trout populations. Regression lines are included for each population (Michipicoten leans:
2
2
2
β=0.07, R =0.01; Marquette leans: β= -1.26, R =0.67; siscowets: β= -0.39, R =0.04). There
were no significant differences between slopes. ...............……………..............................…41
Figure 12. The marginal means of PBDE and PCB concentrations in the liver were compared
between the populations along with the relationship between PBDE and PCB concentrations
in the liver and age. A) Comparison of population marginal mean PBDE liver concentrations
(ng/g) between lake trout populations. B) Comparison of the population marginal mean PCB
liver concentrations. Error bars indicate the 95% confidence interval of the population
marginal means. C) Relationship between PBDE liver concentrations (ng/g) and age (years)
between the lake trout populations. Regression lines are included for each population
2
2
(Michipicoten leans: β=0.20, R =0.87, n=9; Marquette leans: β=0.11, R =0.05, n=9;
2

siscowets: β=0, R =0, n=12). D) Relationship between PCB liver concentration (ng/g) and
2

age (years) between the lake trout populations (Michipicoten leans: β=0.13, R =0.91, n=10;
2

2

Marquette leans: β=-0.04, R =0.05, n=9; siscowets: β=0, R =0, n-12). Letters indicate
populations that have significantly different slopes of the regression line or population
marginal means based on the results of an ANCOVA. ..........................……………………46
Figure 13. The marginal means of PBDE and PCB concentrations in the liver were compared
between the populations along with the relationship between PBDE and PCB concentrations
in the liver and age at particular age ranges. A) Relationship between PBDE liver
concentration (ng/g) and age (years) for ages less than or equal to 14. Regression lines are
2
included for each population (Michipicoten leans: β=0.20, R =0.39, n=4; Marquette leans:
2

β=0.25, n=, R =0.17, n=9). B) Relationship between PCB liver concentration (ng/g) and age
(years) for ages less than or equal to age 14. Regression lines are included for each
2
2
population (Michipicoten leans: β=0.17, R =0.72, n=9; Marquette leans: β= -0.08, R 0.06,
n=4). C) Comparison of the marginal mean PBDE liver concentrations for lake trout at or
above age 20: siscowets (n=8), Michipicoten leans (n=5). D) Comparison of the marginal
mean PCB liver concentrations for lake trout at or above age 20. Error bars indicate the
standard errors of the population marginal mean. Letters indicate populations that have
significantly different slopes of the regression line or population marginal means based on
the results of an ANCOVA..............………………………………………..……………..…47
viii

Figure 14. The relationship between PBDE concentrations in the muscle tissue and age were
compared between the populations: Marquette leans (n=9), siscowets (n=12) and
Michipicoten leans (n=10). Muscle PBDE concentrations were also compared between
populations at three ages. Comparing three ages are necessary for a strong comparison
because of the differences in the slopes of the regression lines for each population. A)
Relationship between PBDE muscle concentration (ng/g) and age (years). Regression lines
2
are included for each population (Michipicoten leans: β=0.15 R =0.76; Marquette leans:
2

2

β=0.04, R =0.02; siscowets: β= -0.01, R =0). Letters indicate populations that have
significantly different slopes of the regression line to an alpha value of 0.05 based on the
results of an ANCOVA. B) Comparison of PBDE muscle concentrations (ng/g) at age 17
and 22 years. Error bars indicate the standard errors of the population marginal mean.
Comparison of population marginal mean PBDE muscle concentrations (ng/g) between lake
trout populations for ages 12 years showed no significant differences between populations.
Letters indicate the significant difference between marginal means. Error bars indicate the
standard errors of the population marginal mean...........……………………………..……...50
Figure 15. The marginal mean PCB concentrations in the muscle were compared between the
populations along with the relationship between PBDE and PCB concentrations in the
muscle: Marquette leans (n=9), siscowets (n=12) and Michipicoten leans (n=10). A)
Relationship between PCB muscle concentration (ng/g) and age (years). Regression lines are
2
included for each population (Michipicoten leans: β=0.14, R =0.76, Marquette leans: β=0.02,
2

2

R =0; siscowets: β=0.04, R =0.11). Letters indicate populations that have significantly
different slopes of the regression line based on the results of an ANCOVA. B) Comparison
of the population marginal means PCB muscle concentrations with two outliers removed
(ng/g). Error bars indicate the standard errors of the population marginal mean. Letters
indicate population marginal means across that are significantly different to an alpha value of
0.05 based on the results of an ANCOVA........................................................…………...…51
Figure 16. The marginal means of PBDE and PCB concentrations in the muscle adjusted for liver
concentrations were compared between the populations along with the relationship between
PBDE and PCB adjusted muscle concentrations and age: Marquette leans (n=9), siscowets
(n=12) and Michipicoten leans (n=10). A) Comparison of PBDE muscle concentrations
(ng/g) that have been adjusted for liver PBDE concentrations (ng/g). Error bars indicate the
95% confidence interval of the population marginal means. B) Relationship between the
PBDE muscle concentrations (ng/g) adjusted for PBDE liver concentrations (ng/g) and age
2
(years) between the lake trout populations (Michipicoten leans: β= -0.20, R =0.34;
2

2

Marquette leans: β=0.09, R =0; siscowets: β= -0.08, R =0.01). C) Comparison of PCB
muscle concentrations (ng/g) that have been adjusted for liver PBDE concentrations (ng/g).
Error bars indicate the 95% CIs of the population marginal means. D) Relationship between
PCB muscle concentrations (ng/g) adjusted for PCB liver concentrations (ng/g) and age
2
(years) between the lake trout populations (Michipicoten leans: β=0.04, R =0.11; Marquette
2

2

leans: β=0.05, R =0.07; siscowets: β=0.03, R =0.01). Letters indicate populations that have
significantly different slopes or population marginal means based on the results of an
ANCOVA.....…………………………………………………………………………….…..52
ix

Figure 17. The marginal means of PBDE and PCB concentrations in the ovary were compared
between the populations along with the relationship between PBDE and PCB concentrations
in the ovary and age. A) Comparison of population marginal mean PBDE ovary
concentrations (ng/g). Error bars indicate the 95% CI of the population marginal mean. B)
Comparison of population marginal mean PCB ovary concentrations (ng/g). Error bars
indicate the 95% CI of the population marginal mean. C) Relationship between PBDE ovary
concentrations (ng/g) and age (years) between the lake trout populations. Regression lines
2
are included for each population (Michipicoten leans: β=0.12, R =0.91, n=10; Marquette
2

2

leans: β=0.13, R =0.61, n=9; siscowets: β=0.02, R =0.02, n=10). D) Relationship between
PCB ovary concentration (ng/g) and age (years) between the lake trout populations
2
2
(Michipicoten leans: β=0.09, R =0.78, n=10; Marquette leans: β=0.09, n=9, R =0.24;
2

siscowets: β=0.04, R =0.23, n=12). Letters indicate populations that have significantly
different slopes of the regression line or population marginal means based on the results of an
ANCOVA.………...…………...............................................................................................55
Figure 18. The marginal means of PBDE and PCB concentrations in the ovary were compared
between the populations along with the relationship between PBDE and PCB concentrations
in the ovary and age: Marquette siscowets (n=8) and Michipicoten leans (n=5). A)
Relationship between PBDE ovary concentration (ng/g) and age (years) for ages greater than
2
2
age 20 (Michipicoten leans: β= -0.04, R =0.11; siscowets: β=0.07, R =0.24). B)
Relationship between PCB ovary concentration (ng/g) and age (years) for ages greater than
2
2
20 (Michipicoten leans: β=0.01, R =0; siscowets: β=0.06, R =0.24). C) Comparison of the
marginal mean PBDE ovary concentration for ages ≥20 years. Error bars indicate the 95%
confidence interval of the population marginal mean. D) Comparison of the marginal mean
PCB ovary concentrations for ages ≥20 years. Letters indicate populations that have
significantly different slopes of the regression line or population marginal means to an alpha
value of 0.05 based on the results of an ANCOVA......................................………………...56

x

KEY TO SYMBOLS OR ABBREVIATIONS

AIC

Akaike information criterion

Concen.

concentration

CI

confidence interval

GC/MS

Gas chromatography / mass spectrometry

GSI

Gonadosomatic index

g

gram

µL

microliter

mL

milliliter

ng/g

nanogram / gram

Kow

octanol-water partition coefficient

OMOE

Ontario Ministry of the Environment

PBDE

polybrominated diphenyl ether

PCB

polychlorinated biphenyl

Pop.

Population

β

slope of the regression line

TCDD

2,3,7,8-tetrachlorodibenzo-p-dioxin

v/v

volume / volume

w/w

weight / weight

wt/w

wet weight

xi

Introduction
Lipophilic contaminants, such as polybrominated diphenyl ethers (PBDEs) and
polychlorinated biphenyls (PCBs), are usually transferred throughout the body of a fish by
binding to lipoproteins and triglycerides. The movement and distribution of triglycerides and
lipoproteins as a result of metabolism, growth and reproduction, constantly reposition the
lipophilic contaminants, often leading to uneven accumulation patterns in different tissues
(Dulfer et al., 1998). Growth and reproductive processes are influenced heavily by life history
(Roff, 2002) and in this study, I will explore how the life history traits of three different
populations of lake trout (Salvelinus namaycush) can potentially influence PBDE and PCB
accumulation in different tissues. The potential link between life history traits and lipophilic
contaminant accumulation patterns could have implications for management and monitoring of
lake trout in the Great Lakes.
Environmental challenges faced by populations will dictate birth and death rates and
therefore will influence population structure including size at age (growth), age structure, age at
maturity, fecundity and survival rates (Roff, 2002). Life history traits are heritable
characteristics that are theoretically a result of evolution and optimize survival and reproduction
and include traits such as age and size at maturity, growth rate and reproductive effort with age.
These traits directly influence population structure. Populations that face different
environmental challenges could select for traits that allow for different tradeoffs between
survival and reproduction, and such tradeoffs could lead to different population growth
trajectories and reproductive schedules. For example, a population that experiences high adult
mortality may reach sexual maturity sooner, exhibit increased growth rates and show high
reproductive investment at younger ages when compared to a longer-lived population (Roff,

1

2002). These differences in growth rates and reproductive outputs between populations could
also lead to different patterns of lipophilic contaminant bioaccumulation.
Lipids are an energy source that can either be stored or mobilized for use in growth or
reproductive processes. In lake trout, the dominant storage lipids are triacylglyercols (DiGiulio
& Hinton, 2008). Stored triacylglycerols provide a moderate degree of independence from a
constant forage food sources (DiGiulio & Hinton, 2008). Additionally, lipid storage in some
morphotypes compensates for vertical lift and drag by reducing an individual’s specific gravity
(Henderson & Anderson, 2002). Lipids can be used for egg production and are transferred into
the developing egg mass or incorporated into vitellogenin, the yolk-precursor protein (Kleinow
et al., 2008). Environmental pressures on populations will dictate whether lipids are stored, used
for growth or reproduction at particular ages. Lake trout that live at a greater depth and store
more lipids as muscle fat expend less energy to migrate through the water column because
energy is no longer required to use a swim bladder (Henderson & Anderson, 2002). Inversely,
lake trout that experience higher rates of adult mortality are likely to allocate lipids towards
growth and reproduction earlier in life. The trade-offs between energy allocations to growth or
reproduction is meant to optimize survival of the population, but has implications for the
distribution of lipophilic contaminants that bioaccumulate in the body of a fish.
Lipophilic contaminants are moved through the body by binding with lipoproteins, such
as vitellogenin (Ungerer & Thomas, 1996). The movement of lipophilic contaminants within the
body begins at the time of exposure. When PCBs and PBDEs enter the body, they are
metabolized by the cytochrome P450 (CYP) enzyme system once they enter the liver (Schlenk et
al., 2008). The cytochrome P450 enzyme system is the most dominant system in the phase I
biotransformation oxidation processes (Schlenk et al., 2008). The CYP enzymes convert

2

lipophilic pollutants into water-soluble products that can be excreted (Schlenk et al., 2008). The
particular CYP isozyme involved depends on the congener. PCBs with coplanar structures
interact with the Ah receptor and induce CYP1A (James, 2001). CYP2B isozymes are induced
by ortho-substituted non-planar PCBs (Arnold & Feeley, 2002). Oxygen is introduced into the
PCB via the cytochrome P450 enzyme system and the hydroxylated metabolite may be
conjugated with a glucuronide or sulfate with UDP-glucuronosyltransferase or PAPSsulfotransferase as catalysts (James, 2001). The metabolite is then excreted through the bile;
however, when organisms are exposed to lipophilic pollutants at a higher rate than the
contaminants are metabolized, the contaminants begin to bioaccumulate in the organism.
Lipophilic contaminants can bioaccumulate disproportionately throughout the body in a
pattern related to the lipid concentration of the organs. Additionally, the accumulation
efficiencies of tissues vary (Karjalainen et al., 2006). For example, when brook trout (Salvelinus
fontinalis) were exposed to 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD), the total assimilation
through dietary exposure for females was 84 % (Tietge et al., 1998). The ovaries only retained
6% of the total body burden of TCDD and 43% of muscle concentrations(Tietge et al., 1998;
Walker et al., 1994). In previous studies, it was found that patterns of lipophilic contaminant
accumulation follows lipid concentrations throughout the body, including the blood, liver, ovary,
muscle and eggs (Tietge et al., 1998).
The ease at which lipophilic contaminants bind to lipoproteins and triglycerides is a
process that is determined by the contaminant’s octanol-water partition coefficient (Kow) (Streit
et al., 1991). Kow is the ratio of the concentration of a chemical in octanol and water at
equilibrium (United Geological Survey, 2010). The octanol is considered a surrogate for lipid
matter, and examining the Kow allows for an understanding of the fate of chemicals in the
3

environment. A chemical with a high Kow, ranging between 2 and 7, will bioaccumulate at
greater rates than a chemical with a lower Kow (Streets et al., 2006).
The lipophilic contaminants of interest in this study include two industrial persistent
contaminants with high Kow, PCBs and PBDEs. PCBs, in the past, were used in a variety of
commercial and industrial applications including fluorescent light ballasts, electrical
transformers, plasticizers in plastic and rubber, and cable insulation. The manufacture and use of
PCBs was prohibited in 1976, however the release of PCBs into the environment continues due
to existing equipment and processes (United Nations Environment Programme, 2002). PCBs are
highly lipid soluble; the logarithms of Kow values for tetra, penta, hexa and octa PCBs range
from 5.46 to 7.67 (Hawker & Connell, 1988). PBDEs are a class of brominated flame retardants
used in items such as furniture foam, plastics and draperies. PBDEs are also highly lipid soluble;
the logarithmic Kow values for pentaBDE-99 range from 6.5 to 8.5 (US EPA, 2008b).
Production of these compounds are not banned in the United States, however the manufacture of
pentaBDE and octaBDE ceased starting in 2005 (US EPA, 2011). Both PCBs and PBDEs can be
transported atmospherically from heavily polluted areas and deposited in more remote areas
(Datta et al., 1999 and Schmide et al., 2007). The atmospheric deposition of these contaminants
complicates management of these contaminants in the Great Lakes ecosystem.
PCBs and PBDEs are ubiquitous in the Great Lakes and in fish because of their
widespread usage and chemical composition and may contribute to adverse health effects within
aquatic ecosystems. As top predators in the Great Lakes ecosystem, lake trout have historically
been known to bioaccumulate lipophilic contaminants. PCB levels in whole lake trout from the
Great Lakes have declined since the 1970s, although recently the levels have stabilized and
4

remain above levels requiring consumption warnings in some locations (US EPA & Environment
Canada, 2009). Concentrations of PBDEs in lake trout eggs from the Great Lakes have
increased in the last 20 years (Valters et al., 2005). The PBDE levels in whole lake trout from all
five of the Great Lakes was increasing through the early 2000s, but more recently we are
beginning to see a decline (US EPA & Environment Canada, 2009). Concentrations of these
contaminants in particular tissues have health implications for humans that consume lake trout
(Devault et al., 1996) and the health of the fish populations (Maule et al., 2005; Olsson et al.,
1999).
PBDEs and PCBs induce sublethal toxic effects, which can lead to stress on human and
fish populations. PCBs have been found to affect the immune system in fish and can be
embyotoxic (Maule et al., 2005; Olsson et al., 1999). PCBs are endocrine disruptors that reduce
the uptake of vitellogenin by oocytes and reduce fecundity (Halden et al., 2010; Rose et al.,
2003). Additionally, the planar PCB congeners produce dioxin-like effects (Tillitt et al., 2008).
PBDEs, as endocrine disruptors, show many structural similarities to dioxins, which
suggest that they may have toxic properties that are similar to dioxins (US EPA, 2010b). The
structural similarities between PBDEs and thyroid hormones suggest disruption of the immune
system (Luthe et al., 2008; Birchmeier et al., 2005).
The siscowet and lean morphotypes from Lake Superior along with the Michipicoten
Island lean population were chosen for this study because of the unique life history traits
exhibited for each population. Populations of lake trout from Lake Superior can be grouped into
two ecologically distinct morphotypes (Figure 1). Lean and siscowet lake trout morphotypes are
estimated to have diverged from each other around the time of the Wisconsin Glacial Episode,
although debate exists as to whether the morphotypes diverged before or after recolonization

5

(Page et al., 2004; Eshenroder, 2008). The bathymetric distribution of siscowets and leans result
in key life history differences, most noticeable is the relative lipid content (Sitar et al., 2008).
Populations of siscowets are found in large abundances at depths greater than 100 meters and are
robust with higher body fat content, whereas lean lake trout populations are most abundant in
depths less than 50 meters and are slender with low body fat content (Eschmeyer & Phillips,
1965; Hansen et al., 1995). The higher lipid content in siscowets is hypothesized to overcome
drag and reduce energy expenditures during vertical migration (Henderson & Anderson, 2002).
Whether these morphological differences result from phenotypic plasticity or genetic differences
is still under investigation, but studies indicate a genetic basis for differences in lipid levels
between leans and siscowets (Goetz et al., 2010, Eschmeyer and Phillips, 1965).
In addition to differences in lipid accumulation and distribution patterns, siscowets and
leans also show differences in growth and fecundity, which may have implications for
contaminant distribution. Lean lake trout grow faster than siscowets and have a higher relative
fecundity (Hansen et al., 1995). The Michipicoten Island lake trout is believed to be a separate
genetic strain of lean lake trout (Meredith, 2008). Michipicoten leans grow slower than the
Marquette leans and live in north-eastern Lake Superior where they are best able to persist
because fishing pressure and sea lamprey (Petromyzon marinus) parasitism rates are low (Bronte
et al., 2007; Ebener, 1998). The growth rate of the individual is important to the accumulation of
lipophilic contaminants because of the effect of biodilution: a faster growing individual will have
lower concentrations of lipophilic contaminants at a specific length compared to a slower
growing individual (Simoneau et al., 2005). However, this theory of biodilution assumes
individuals with different growth rates experienced similar contaminant exposure rates.

6

Figure1. The phenotypic differences between lean and siscowet lake trout (courtesy Shawn
Sitar).
The interaction of life history traits of the populations and lipophilic contaminants like
PBDEs and PCBs causes unique concerns for human health and lake trout population
management. Accumulation of lipophilic contaminants into the muscle tissue of lake trout that
are being consumed by humans creates human health concerns. Siscowets are being considered
for use as high-grade fish oil dietary supplements because of their high population numbers and
lipid content. Human consumption of fish oils has been associated with many human health
benefits including prevention of cardiovascular diseases (Marik & Varon, 2009) and siscowet
lake trout have been found to contain very high levels of omega-3 fatty acids (Wang et al., 1990).
Fish oil extracted from siscowets could be used as nutritional supplements and used to produce
high-value fish meal. However, if siscowets store large amounts of lipids, and thus lipophilic
contaminants, the benefits of consumption of omega-3 fatty acids could be compromised by
increased exposure to halogenated contaminants. Life history traits can vary within species and I
suspect that life history traits should be taken into consideration when using lake trout data for
public health recommendations and when considering the response of a population to a stressor
such as PBDEs and PCBs.
7

Additionally, mobilization of lipophilic contaminants through the body allows for the
maternal transfer of contaminants into the ovaries. Accumulation of endocrine disrupting
contaminants in the ovaries, and thus the eggs, presents potential health concerns for the
developing embryo (Miller et al., 1993). The additional stress of lipophilic contaminants adds to
the multitude of stressors faced by lake trout in the Great Lakes and contributes to the decline of
naturally reproducing lean lake trout populations.
My study investigates the role of life history traits on patterns of lipophilic contaminant
accumulation in three lake trout populations. The energy trade-offs between growth and
reproduction for leans and siscowets have been reported in previous studies (Eschmeyer and
Philips, 1965; Hansen et al., 1995, Miller et al., 2000). My hypothesis is that the distribution of
PCBs and PBDEs will be unique to each population and will reflect life history trade-offs. I
predict that lean populations with a greater reproductive investment will have higher
concentrations of PCBs and PBDEs in the ovary tissue than siscowets. The siscowets, due to
larger stores of muscle fat are predicted to have higher concentrations of PCBs and PBDEs in the
muscle tissue. My research investigated this hypothesis by sampling these three lake trout
populations from Lake Superior and collecting life history data, such as age, fecundity, muscle
lipid concentration and weight. Analyses of muscle, ovary and liver tissue samples for selected
PCB and PBDE congener concentrations were performed using gas chromatography/mass
spectrometry methods. The life history and contaminant data were then analyzed in tandem to
understand the interaction between life history of a population and contaminant concentrations.
My study will improve our understanding of the relationship between life history and lipophilic
contaminants; a relationship that can have implications for the management and consumption of
lake trout.

8

Methods
Field Sampling

Sexually mature female lake trout were collected during September and October of 2008
and 2009 off the coast of Marquette, USA and Michipicoten Island, Canada; both in Lake
Superior (Figure 2A). Lean and siscowet lake trout were collected offshore of Marquette as part
of the Michigan Department of Natural Resources project sampling. The Marquette sampling
used bottom gill nets with individual nylon, mutifilament panels (91.4 m x 1.8 m) and mesh sizes
including 10.2, 11.4, 12.7 and 14.0 cm. The nets were set at two sites for twenty-four hours
before lifting. The first site, that targeted siscowets, was between 375 meters and 385 meters
deep (latitude 46°37.10’ and longitude 87°16.20’). The second site was selected to capture
leans, and was between 109 and 145 meters (latitude 46°33.17’ and longitude 87°18.90’)
(Figure 2B). The mesh sizes were selected in order to target mature lake trout (greater than 550
mm). The minimum length requirement was based on current knowledge of average mature lake
trout lengths (Goetz et al., in review). The Marquette sampling resulted in the capture of thirtyfour female lake trout, but only nine leans and twelve siscowets were mature females. The lake
trout sampled from Michipicoten Island were part of the Ontario Ministry of Natural Resource’s
effort to stock their hatchery programs (Felt, unpublished). Lake trout were caught by gill nets
with 50 and 100 foot panels of 2 and 2.5 inch monofilament. The nets were set for between 3
and 70 hours before lifting at seven main sites (Figure 2C). The sampling trip occurred over the
course of about 20 days in both 2008 and 2009. A total of 73 lake trout were caught in 2008, but
only twenty mature females were caught near Michipicoten Island between 2008 and 2009. Ten

9

lake trout from the Michipicoten population were included in this study.
A.

Michipicoten Island

Marquette

B.
375
109

North
N

Marquette
Munising

C.

Michipicoten
Island

N
Figure 2. A) Map of Lake Superior. Large circles indicate sampling locations (map courtesy of
mspmag.com). B) Location of two sites of gill net sets off the coast of Marquette. The numbers
on the map indicate the depth in meters at each site (map courtesy of Ralph Tingley III). C)
Locations of gill nets set for lake trout sampling near Michipicoten Island (map courtesy of the
Ontario Ministry of Natural Resources).
10

On site, various measurements were made including morphology metrics, length and fork
length, weight and fatmeter readings. Lean and siscowet lake trout were identified based on their
morphometry that included fin size, eye location, eye size and visceral fat (Bronte & Moore,
2007; Eschmeyer and Philips, 1965). Length and fork length were measured onsite to the nearest
milimeter. Weight was measured to the nearest gram and ovary and liver mass was taken to 0.1
gram. Muscle fat measurements were taken along the outer body cavity with a fish fatmeter
(Model FM-692, Distell Inc.,West Lothian, Scotland). Fish were then filleted with the skin on
and the muscle tissue, liver and ovaries were removed. The tissue samples were stored
separately in aluminum foil and frozen at -80°C until analysis.
The fatmeter readings were taken at consistent locations along the right side of the outer
body (Figure 3). The fatmeter is a noninvasive microwave sensor that emits a low frequency
microwave beam (2 GHz ±20 MHz) that efficiently heats water dielectrically and provides a
reading of water content of the tissue. Fatmeter readings were converted to lipid percentages
based on a relationship of lipid to water content in fish, calibrated using lipid extraction data (see
Lipid Extraction section).

Figure 3. Location of the four fatmeter readings taken along right lateral body. Readings were
then calibrated with lipid extraction data to generate muscle fat percentages for lake trout.
(Photograph courtesy of Sara Smith)

11

Additional information was collected offsite after collections. Ages were determined by
counting the annuli on the otoliths. Fecundity was later determined by extracting four fresh
ovary samples approximately 15 grams, which were weighed and all eggs were counted in each
sample. The number of eggs per gram of each sample was averaged and multiplied by gonad
mass to determine fecundity. Photos of approximately 25 eggs from each fish were taken on a
Leica MZ6 dissecting scope. The pictures were then imported into Axiovision Rel. 4.7 where
the numbers of pixels along the diameter of the egg were measured. The pixel number of a
known length was used to convert the pixel measurements into egg diameter measurements.

12

Contaminant Analysis
PBDEs and PCBs were quantified using a contaminant extraction procedure and
quantified using gas chromatography/mass spectrometry (GC/MS) methods. This section details
the column construction, contaminant extraction, quality control and contaminant quantification.

Column Construction
Two chromatography columns were constructed using mixtures of packing materials and
used for contaminant extraction (see Contaminant Extraction). Before the chromatography
columns were constructed, packing materials were prepared. First, anhydrous sodium sulfate
and silica gel with a mesh of 70/230 were activated by placing in an oven at 215°C for at least
70 hours before use. Once the silica gel was activated, it was mixed with a variety of chemicals
in order to create packing materials used in the column construction. The 44% (weight/weight)
sulfuric acid/silica packing material was made by mixing 56 grams of activated silica with 25
mL of concentrated sulfuric acid until the packing was free flowing. The 10% (w/w) silver
nitrate/silica packing material was made by dissolving 6 grams of silver nitrate in 23 mL of
HPLC grade water. The silver nitrate/water mixture was then combined with 53 grams of
activated silica and shaken until the packing was free flowing. The 33% (w/w) 1 M sodium
hydroxide/silica packing material was prepared by weighing 1.3 grams of sodium hydroxide and
dissolving it into 33 mL of HPLC grade water. The water/sodium hydroxide mixture was
combined with 67 grams of activated silica and shaken until the packing was free flowing. All
packing materials were stored in desiccators until use.
The extraction column was prepared by plugging a 250 mL cylindrical funnel first with
glass wool and then with 25 g of 44% sulfuric acid/silica and finally with 8 g of anhydrous

13

sodium sulfate. The amount of 44% sulfuric-acid/silica in the extraction column was increased
for more samples with higher concentrations of lipids, which was indicated by the color change
of the silica gel mixtures.
The acid-base column was prepared by plugging a chromatography column (14 x 3.5 cm
i.d.) with glass wool and adding, in order from bottom to top: 1.5 g 10% silver nitrate/silica
packing, 1.0 g activated silica packing, 2.0 g 33% sodium hydroxide/silica packing, 1.0 g
activated silica packing, 4.0 g 44% sulfuric acid/silica packing, 2.0 g activated silica packing and
2.0 g anhydrous sodium sulfate.
All glassware was rinsed three times with acetone, three times with toluene and twice
with dichloromethane between each sample.

Contaminant Extraction
All digestion, extraction and GC/MS procedures were modified from the Ontario
Ministry of the Environment’s method (OMOE, 2004).
Approximately 5 grams of muscle, ovary and liver tissue was homogenized before
contaminant extraction. Muscle tissue was homogenized with dry ice in a commercial grade
blender. Ovary tissue and liver tissue was thawed and manually homogenized using a sharp
blade. Each sample was digested overnight in 50 mL of concentrated hydrochloric acid. The
hydrochloric acid was pre-extracted before contact with the sample using a 3:1:1 ratio of
hydrochloric acid, dichloromethane and hexane to prevent contamination of the sample. The
first column used during the extraction process was the extraction column (see Column
Construction). The extraction column was pre-rinsed with 50 mL of hexane and the eluate was
discarded. The digested tissue solution (also referred to as the acid-layer) was extracted with 40

14

mL of hexane by mixing in a separatory funnel for 2 minutes. The hexane and acid layers were
then allowed to separate and then the hexane extract was run through the extraction column to
remove water, excess lipids and other polar compounds. The acid-layer was returned to the
original flask and the extraction was repeated twice and added the hexane from each extraction
to the extraction column and at the end, discarded the remaining acid-layer. To ensure
maximization of extraction, the original flask and the separatory funnel were rinsed twice with
20 mL of hexane, which was added to the extraction column. Eluate from extractions was
collected in a round bottom flask and placed on a rotary-evaporator at 45°C until only about 2
mL of the eluate remained. In between individual sample extractions, the rotary evaporator was
rinsed with 150-200 mL of hexane.
Each sample was subjected to chromatographic cleanup with the acid-base column,
which was made with silica gel mixtures (see Column Construction). The acid-base column was
pre-rinsed with 50 mL of hexane and the eluate was discarded. Then each sample extract was
added to the acid-base column. The original round-bottom flask containing the sample was
rinsed three times with 2 mL hexane and added to the acid-base column. Then, 100 mL of
hexane was added to the acid-base column followed by 80 mL of 50:50 dichloromethane and
hexane mixture. The acid-base column was allowed to drain to bed level between each addition.
Once the column was well drained, the round bottom flask containing the eluate was placed on
the rotary evaporator until approximately 1 mL was remaining in the flask. The concentrated
sample extract was then transferred to a clean glass vial and the round bottom flask rinsed twice
with hexane and then twice with dichloromethane with each rinse was added to the glass vial.
The glass vial was stored at -80°C until evaporated using a nitrogen evaporator. The sample was

15

not completely evaporated and the remaining sample was transferred into an amber glass
autosampler vial with a 100 µL insert.

Quality Control
The native standards were purchased from Wellington Laboratories (Guelph, Ontario)
and were spiked in the precision and recovery (PAR) sample between every 20 tissue samples in
order to understand recovery of native PBDEs. The native standard includes BDE 28, 47, 66, 99,
100, 153 and 154 (Table 1).
13

The internal standard [ C12]- 2,2’,4,4’-TeBDE (47L) was also purchased from
Wellington Laboratories (Guelph, Ontario). A spiked procedure blank (SPB) was analyzed
between every 10 samples and included a known concentration of internal standard and about
50mL of pre-extracted hydrochloric acid (extracted as described below). The internal standard
was added to ever12y sample at a known concentration in order to determine recovery rates, but
concentrations were not corrected based on recovery.
Calibration standards were purchased from Wellington Laboratories (Guelph, Ontario)
13

and included [ C12]- 2,2’,4,4’-TeBDE (47L), 2,4,4’-TrBDE (28), 2,2’,4,4’-TeBDE (47),
2,3’,4,4’-TeBDE (66), 2,2’,4,4’,5-PeBDE (99), 2,2’,4,4’,6-PeBDE (100), 2,2’,4,4’,5,5’-HxBDE
(153) and 2,2’,4,4’,5,6’-HxBDE (154) (Table 1). This set of calibration standards generated a
2

calibration curve, which included data from 4 different concentrations. The R values of the
calibration curves were all greater than 0.90 suggesting a strong relationship between
concentration and area under the curve of the peak generated by the GC/MS.

16

The PCB calibration standards were purchased from Cambridge Laboratories (Andover,
MA). The calibration standards generated a calibration curve with data at 6 different
concentrations and included 2’,3,4,4’,5-PeCB (123), 2,3,3’,4,4’,5-HxCB (156), and
2,3’,4,4’,5,5’-HxCB (167) (Table 1). The internal standard BDE 47L was added to each
2

calibration standard at a known concentration. The R values of the calibration curves were all
above 0.88.

17

Table 1. Standards and congeners used for analysis
Type of Standard
Precision and
Recovery (PAR)

Source
purchased
Wellington
Laboratories
(Guelph,
Ontario)

Congeners included in Standard

BDE 28
BDE 47
BDE 66
PBDE
Calibration
Standard
(continued)

PCB Calibration
Standard

[ C12]- 2,2’,4,4’-TeBDE
2,4,4’-TrBDE
2,2’,4,4’-TeBDE
2,3’,4,4’-TeBDE

BDE 99
BDE 100
BDE 153
BDE 154

PBDE
Calibration
Standard

Wellington
Laboratories
(Guelph,
Ontario)
Wellington
Laboratories
(Guelph,
Ontario)

2,4,4’-TrBDE
2,2’,4,4’-TeBDE
2,3’,4,4’-TeBDE
2,3’,4,4’-TeBDE
2,2’,4,4’,6-PeBDE
2,2’,4,4’,5,5’-HxBDE
2,2’,4,4’,5,6’-HxBDE

BDE 47L

Internal Standard

BDE 28
BDE 47
BDE 66
BDE 99
BDE 100
BDE 153
BDE 154
BDE 47L

2,3’,4,4’-TeBDE
2,2’,4,4’,6-PeBDE
2,2’,4,4’,5,5’-HxBDE
2,2’,4,4’,5,6’-HxBDE

PCB 123
PCB 156
PCB 167
BDE 47L

2’,3,4,4’,5-PeCB
2,3,3’,4,4’,5-HxCB
2,3’,4,4’,5,5’-HxCB
13
[ C12]- 2,2’,4,4’-TeBDE

13

[ C12]- 2,2’,4,4’-TeBDE

13

Wellington
Laboratories
(Guelph,
Ontario)

Cambridge
Laboratories
(Andover, MA)

18

All solvents were either analytical reagent grade or HPLC grade. All samples were
analyzed in duplicate.
The average percent recoveries for the PAR samples were reported along with the
amount of each contaminant found in the blanks (Table 2).

Table 2. Amount of contaminants found in blanks and the average percent recoveries for the
PAR samples
Congener number
BDE 28
BDE 47
BDE 66
BDE 99
BDE 100
BDE 153
BDE 154
PCB tetra unknown
PCB 123
PCB penta unknown #1
PCB penta unknown #2
PCB 156
PCB 167
PCB hexa unknown #1
PCB hexa unknown #2

Greatest amount found in blank
(ng)
0.18
2.44
0.78
0.52
49.15
2.56
2.15
44.56
31.02
163.2
25.44
1.45
5.81
58.24
49.04

Average % recovery for the
Precision and Recovery Samples
62.82
95.9
101.42
111.3
91.22
268.23
249.97
PCB standards were not included
in PAR

Contaminant Quantification
The samples were analyzed using an Agilent 6890N gas chromatograph coupled to an
Agilent 5973 mass spectrometer with helium as the carrier gas (Michigan State University Mass
Spectrometry Facility). The samples were injected as 1 µL in splitless mode with a 2 minute
purge time and the injection port was held at 270°C. The GC column used was a 30 m x 0.25
mm (i.d.) capillary tube coated with DB-5MS and 0.25 µm film thickness (J&W Scientific,
Folsom CA). The GC temperature program was as follows: 110°C for 1 min., 45°C/min to

19

200°C, 6°C/min to 240°C, 4°C/min to 325°C and held at 325°C for 5 min. The MS transfer
line was held at 280°C. Selected ion monitoring was used to detect two masses for the tri-,
penta- and hexa-BDEs and the tri-, penta- and hexa-PCBs. The selected ion monitoring reported
only the two ion masses determined to correspond to the molecular ion of the selected congeners
of PBDEs and PCBs. Samples with low signal-to-noise ratios were reinjected with 2 µL of
extract. A signal to noise ratio less than 10 was considered not quantifiable and less than 3 was
considered not detectable. The detection limits and limits of quantification were used to replace
contaminant readings with low signal to noise ratios with a random number (Table 3). A sample
with a signal to noise ratio less than 3 for a particular congener was assigned a random number
between the zero and the detection limit. A sample with a signal to noise ratio less than 10 for a
particular congener was assigned a random number between the detection limit and the limit of
quantification. The use of random numbers was based on my assumption that the overall
concentrations for a given tissue against the frequency of fish with that concentration would form
a bell-curve. However, the limits detection of the GC/MS machine would truncate this curve at
the detection limit and limit of quantification. Therefore, samples with low signal-to-noise ratios
were assigned random numbers in place of zeros in order to provide a more normal distribution
of PBDE and PCB concentrations.

20

Table 3. Detection Limits and Limits of Quantification
Congener number
BDE 28
BDE 47
BDE 66
BDE 99
BDE 100
BDE 153
BDE 154
PCB tetra unknown
PCB 123
PCB penta unknown #1
PCB penta unknown #2
PCB 156
PCB 167
PCB hexa unknown #1
PCB hexa unknown #2

Detection limit (ng)
0.34
0.42
0.17
0.75
0.36
0.71
0.24
64.5
86.2
534
120
0.01
5.35
0.01
n/a

Limit of Quantification (ng)
1.22
n/a
1.05
5.76
3.16
3.73
1.67
125
96.1
958
979
34.3
13.6
90.7
n/a

Calibration curves for both PBDEs and PCBs were generated in order to relate the output
of the GC/MS data to output from known concentrations. A calibration curve for PBDEs was
generated before and after every set of samples run through the GC/MS. The PCB calibration
curve was generated once for all samples injected at 1 µL and once for samples injected at 2 µL.
The isotopic ratio between primary and secondary ion masses was within ±10% of theoretical
values. The detection of a second ion mass allowed me to correctly determine that the
compound was a PCB or PBDE; additionally the use of the secondary ion mass allowed for the
inclusion of several PCB congeners that were not identified in the standard. One unknown tetraPCB, two unknown penta-PCBs and two unknown hexa-PCBs were detected with enough
frequency throughout the samples to add to the analysis. These unknown PCB congeners were
identified based the known molecular mass of a primary and secondary ion and an isotopic ratio
within ±10% of theoretical values.
The concentrations were determined using MassLynx V4.1 and QuanLynx software
(Waters Corp.), after conversion of the ChemStation data files to MassLynx file format.

21

Lipid Extraction
Fatmeter readings were calibrated by extracting lipids from muscle tissue scanned by the
2

fat meter. Muscle samples (approximately 5 cm ) were collected from areas on the fish where
the fatmeter readings were taken (Figure 3) and stored in -20°C freezer. At the time of lipid
extraction, the tissue was thawed and placed in a drying oven at 75°C until fully dehydrated (i.e.
the mass of the sample did not change). The percent water in the muscle tissue was calculated as
the dried tissue mass divided by the wet tissue mass; percent water was later used to estimate a
wet mass of the dried sample used for the lipid extraction.
The dehydrated tissue was homogenized and lipid content was estimated using a
procedure described by Bligh and Dyer (1959). Approximately 1.5 g of dried tissue was
weighed and the sample was hydrated by adding 12 mL of distilled water and allowed to settle.
After 2 minutes, 45 mL of a 1:2 (v/v) mixture of chloroform and methanol was added to the
hydrated sample. A small amount of Whatman ashless powder was added to the sample and the
sample was homogenized for an additional 3 minutes. An additional 15 mL of chloroform was
added and the mixture was homogenized with the sample for 30 seconds. Then 15 mL of
distilled water was added and the mixture was homogenized for 30 more seconds. The sample
was then filtered through a Buchner funnel using Whatman #1 filter paper. The filter paper was
transferred back into the homogenizer and homogenized for 3 minutes with 50 mL of
chloroform. The sample was then filtered with a new filter paper through the Buchner funnel.
To ensure the entire sample was captured, the homogenizer was rinsed with about 25 mL of
chloroform and the rinsate was filtered. Filtrate was transferred to a graduated cylinder and
allowed to sit for 5 hours to allow the methanol/water and chloroform/lipid layers to separate.

22

The upper layer of methanol/water was removed and the remaining chloroform/lipid layer was
allowed to stand for 12 hours in a fume hood to evaporate the chloroform. The remaining lipid
layer was then dried in a drying oven for 30 minutes at 100°C. The percent lipid was calculated
using wet weight as the amount of lipid, adjusted for the volume removed with the
methanol/water layer, and reported as a percentage. Each tissue was analyzed as duplicate
subsamples. The log-transformed percentages of lipid obtained for each sample were associated
with their respective log-transformed fatmeter reading using a linear equation of best fit.
The data included in this equation (lipid concentration from muscle tissue) was collected
over several years and included fatmeter readings from three separate fatmeters along with
muscle samples and fatmeter readings from lake trout from several additional stocks beyond the
three stocks examined in this study.

23

Statistical Analysis
ANCOVA
Life history traits that include fork length, body weight, fecundity, gonad weight,
gonadosomatic indices and lipid concentration were analyzed using an ANCOVA with age as a
covariate. Fecundity was further adjusted for body weight (Qunice et al., 2008) and an
ANCOVA was conducted with age as a covariate. Gonadosomatic indices (GSIs) were
determined as the gonad weight divided by the body weight, multiplied by 100 and were further
analyzed using an ANCOVA with age as a covariate. All variables were tested for normality and
all life history traits except for adjusted fecundity, GSI and lipid concentration, which were
originally normally distributed, were log-transformed to achieve normality. The ANCOVAs
were tested for homogeneity of slopes. If the slopes of the regression lines within each
population were significantly different, then an interaction term between population type and age
was included in the model and reported. Additionally, if the slopes were significantly different,
the populations’ marginal means were compared at three ages in order to obtain an adequate
understanding of the population differences.
There was a lack of fecundity and gonad weight data from the Michipicoten population
because of the remote location of the island and the delay in arrival of fresh ovaries to the
laboratory. Most of the Michipicoten lake trout tissue that were included in the contaminant
analyses did not have associated fecundity and gonad weight data. However, there were twenty
additional lake trout collected from Michipicoten Island. Life history data from these additional
twenty lake trout were included in the analyses to help describe the population. The addition of
these data allowed for a stronger comparison of the fecundity and gonad weight of Michipicoten
population against the lean and siscowet populations.

24

Differences in PBDE and PCB concentrations of the muscle, ovary and liver tissues
between each population were examined within each tissue type using an ANCOVA. The PBDE
and PCB concentrations were all log-transformed to achieve normality. Age was included as the
covariate and the homogeneity of the slopes was tested using an interaction term between age
and population type. If the slopes of each population’s regression line were significantly
different, the interaction term was reported. A box-plot analysis was used to ensure there were
no outliers and consequently two outliers were removed from the PCB muscle analysis and the
ANCOVA was rerun.
The different lake trout populations exhibited extremely different age structures such that
Marquette lean and siscowet age ranges did not overlap; thereby making comparisons between
populations problematic. Therefore, in certain situations, lake trout were compared within a
certain age group to reduce the influence of age. Ovary and liver concentrations of PCB and
PBDES were compared between siscowets and Michipicotens age 20 years or older using an
ANCOVA. Although the ANCOVA included age as a covariate, truncating the ages in this way
allowed for a stronger comparison between fish of the same age. Liver concentrations were
compared with an ANCOVA between Michipicoten and lean lake trout age 14 year and younger
in order to examine the homogeneity of the regression slopes of lake trout of similar ages.
Multiple Regression
A multiple regression was used to determine which variables best predicted PCB and
PBDE concentrations in the muscle tissue. The regression analysis used the Akaike information
criterion (AIC) in order to understand how contaminant concentrations for a specific tissue can
be predicted by demographic variables and contaminant concentrations from other tissues.

25

Because concentrations of PCBs were an order of magnitude higher than PBDEs, in order to
simultaneously compare the contaminants, the muscle concentrations of contaminants were
standardized by dividing by the mean contaminant concentration. PBDE and PCB
concentrations of the liver were standardized in a similar way using the mean of the
concentrations. The variables included in the AIC analysis for the muscle tissue included age,
population type, liver concentration, lipid concentration and fork length. All PBDE and PCB
concentrations used in the AIC analyses were log-transformed to achieve normality. All
variables were tested for correlation and removed if correlation was high between variables.
The AIC revealed that the best regression models that predicted muscle PBDE and PCB
concentrations included the liver concentration of the contaminant. Liver concentrations could
be representing the current exposure rate. Therefore, to correct for potential differences in
exposure rate and thus isolate life history differences, the muscle concentrations of PBDEs and
PCBs were divided by corresponding liver concentrations.
Lake trout are mainly exposed to PCBs and PBDEs through ingestion of prey and the
liver, because its main function is the metabolism of lipophilic compounds (Hinton et al., 2008),
is one of the first organs to be exposed to lipophilic contaminants. PCBs taken in by organisms
are removed from the liver within 14 days (Karjalainen et al., 2006); whereas muscle tissues
accumulate lipophilic contaminants at a slower rate than other organs based on the internal
transfer of the contaminants and has been found to continue accumulating even after exposure
(Antunes et al., 2008). Therefore the accumulation of PCBs and PBDEs in the muscle tissue is
representative of a longer accumulation period, whereas the liver concentrations of PCBs and
PBDEs would be more indicative of the present exposure rate. Based on the residence times of
contaminants in both the liver and the muscle, I assume that liver concentrations of PCBs and
26

PBDEs can provide an indication of current exposure rates and I attempted to control for the
exposure variable in order to understand the role of life history variations without the
confounding effect of difference in exposure. These adjusted PBDE and PCB muscle
concentrations were then used in an ANCOVA analysis with age as a covariate to determine if
muscle concentrations are different between populations after correcting for potentially different
exposure rates. Since ovary concentrations were not predicted by liver concentrations in the
AIC, ovary concentrations were not adjusted for liver concentrations.
A multiple regression, using AIC, was used to determine which variables best predicted
PCB and PBDE concentrations in the ovary tissue and liver tissue. All contaminant
concentrations were log-transformed. Two outliers were identified, using a Cook’s distance test,
and removed in the regression analyses for the PCB ovary data. Two outliers were identified and
removed for the PBDE ovary analysis based on the residual and Cook’s distance test. One
outlier was removed in the PBDE liver analysis due to the Cook’s distance test. The variables
included to predict ovary concentrations were age, population type, liver and muscle
concentrations, lipid concentration, fork length, fecundity, egg diameter and gonad weight. The
variables included to predict liver concentrations were age, population type, muscle
concentration, lipid concentration and fork length. All statistical analyses were conducted using
SAS version 9.1 (Table 4).

27

Table 4. List of statistical tests used
Statistical analysis
ANCOVA (with age as covariate)
Fork length
Body weight
Fecundity
Gonad weight
Adjusted fecundity
Gonadosomatic Index

Additional information- transformations and
manipulations
log-transformed
log-transformed
log-transformed, included additional Michipicoten data
log-transformed, included additional Michipicoten data
divided by body weight, included additional Michipicoten data
Gonad weight divided by body weight and multiplied by 100,
included additional Michipicoten data

Muscle lipid concentrations
PBDE muscle concentration
PCB muscle concentration
PBDE ovary concentration
PCB ovary concentration
PBDE liver concentration

log-transformed
log-transformed, two outliers removed
log-transformed, additional test with only lake trout ≥ 20 years
log-transformed, additional test with only lake trout ≥ 20 years
log-transformed, additional test with only lake trout ≥ 20 years,
and third test with lake trout ≤ 14 years
PCB liver concentration
log-transformed, additional test with only lake trout ≥ 20 years
and third test with lake trout ≤ 14 years
Adjusted PBDE muscle concentration divided by liver concentrations for each individual
Adjusted PCB muscle concentration
divided by liver concentrations for each individual
Multiple regression using Akaike information criterion (AIC) to predict best model
PBDE muscle concentrations
Concentrations were standardized by dividing by the mean and
log-transformed
PCB muscle concentration
Concentrations were standardized by dividing by the mean and
log-transformed
PBDE ovary concentration
log-transformed, two outliers removed
PCB ovary concentration
log-transformed, two outliers removed
PBDE liver concentration
log-transformed, one outlier removed
PCB liver concentration
log-transformed

28

Results
Life History Results
On average, siscowets were older than Marquette leans, while Michipicoten leans had the
largest age range. Mature female siscowets’ ages ranged from 15 to 26 years (n=12), while
mature female Marquette leans only ranged from 6 to 14 years (n=9). Twenty mature females
were caught between 2008 and 2009 near Michipicoten Island, ON and ten Michipicotens were
included in the analysis. The Michipicoten Island leans’ ages spanned both the Marquette leans
and siscowets with an age range from 6 to 27 years (Figure 4).

7
6
Frequency

5
4

Marquette lean

3

Michipicoten lean

2

Marquette siscowet

1
0
5-10

11-15 16-20 21-25
Age Range (years)

25-30

Figure 4. Histogram of the age ranges (years) of each population. The vertical axis contains the
frequency counts for each age range. Marquette leans (n=9), Marquette siscowets (n=12) and
Michipicoten Island leans (n=10).

The growth rates of both Michipicoten and Marquette leans were much higher than
siscowets. The fork lengths (FL) -at-age of the siscowets were significantly lower than both the
Marquette leans and Michipicoten populations (ANCOVA, Table 5; Figure 5A). Michipicoten
leans had the highest population marginal mean fork length (means adjusted for confounding
variable of age), but were not significantly different than the Marquette leans. The interaction

29

between age and population was not significant in the FL-at-age model (similar slopes) and
therefore not included in the final ANCOVA model, however a t-test indicated that
Michipicotens had a significantly greater slope between age and FL than siscowets (t=-2.53,
p=0.0185; Figure 5B) indicating a higher growth rate. The Marquette leans did not appear to
have a significantly different growth rate than either the Michipicoten leans or siscowets.

30

A.

Fork Length (mm)

a

a

800

b

600
400
200
0
Michipicoten
Island lean

Marquette lean

Marquette
siscowet

B.
Michipicoten lean
Marquette lean
Marquette siscowet

Fork Length (mm)

1000

a

800
ab

600

b

400

5

15

25

Age (years)

Figure 5. The marginal mean fork length (FL) and FL-at-age relationships were compared
between the Marquette leans (n=9), siscowets (n=12) and Michipicoten leans (n=10). A)
Comparison of the marginal mean fork lengths (mm) between lake trout populations. Error bars
indicate the 95% confidence interval of the population marginal means. Letters indicate
populations that are significantly different based on the results of an ANCOVA. B) Relationship
between fork length (mm) and age (years) of the three lake trout populations. Regression lines
2

are included for each population (Michipicoten leans: β=0.02, R =0.88; Marquette leans: β=0.02,
2

2

R =0.21and siscowets: β=0, R =0). Letters indicate populations that have significantly different
slopes of the regression line based on the results of an ANCOVA.
31

Table 5. Differences in life history characteristics based on ANCOVA
Dependent
Variable
Fork Length
(mm)

Independent
Variables
+

Age
Population
Total Model

Degrees of
freedom

F-value

P-value

1

19.75

0.0001**

2

35.42

<0.0001**

3

25.72

<0.0001**

1

15.15

0.0006**

2

27.00

<0.0001**

3

20.76

<0.0001**

1

6.25

0.0181*

2

18.96

<0.0001**

3

13.88

<0.0001**

1

9.62

0.0042*

2

20.46

<0.0001**

3

15.36

<0.0001**

1

1.44

0.2408

2
3

1.12
1.43

0.3401
0.2562

1

4.15

0.0511

2

4.30

0.0236*

3

3.68

0.0237*

1

0.28

0.6025

2

4.77

0.0168*

3

3.77

0.0220*

Body Weights
(g)
+

Age
Population
Total Model
Fecundity
(total eggs)
+

Age
Population
Total Model
Gonad weight (g)
+

Age
Population
Total Model
Fecundity adjusted for body
weight (total eggs / g)
+
Age
Population
Total Model
Gonadosomatic
index
+
Age
Population
Total Model
Muscle Fat (%
lipid)
+

Age
Population
Total Model
+

Age was covariate for all ANCOVA models; *significant to <0.05; ** significant to <0.001
32

Similar to fork lengths, lake trout body weights were also significantly different among
the lake trout populations with siscowets smaller than the other populations (ANCOVA with age
as the covariate; Table 5). Siscowets had significantly lower body weights than both the lean
and Michipicoten populations (Figure 6A); however there were no significant difference between
the slopes of the regression lines between the populations (Figure 6B). Despite the lack of
differences in the slopes between each population, there appears to be different patterns in
growth (Figure 6B). Siscowets have a lower mean body weight despite being older than the
Marquette leans and the Michipicoten leans are also heavier on average than the siscowets.

33

Population Marginal
Mean Body Weight (g)

A.

9000
7500
6000
4500
3000
1500
0

a
a
b

Michipicoten Marquette lean
Island lean

Marquette
siscowet

B.
Michipicoten lean
Marquette lean
Marquette siscowet

Body Weight (1000 g)

12
10

a

8
6
a

4

a
2

5

15

25

Age (years)

Figure 6. The marginal mean body weights and the relationships between weight and age were
compared between the populations: Marquette leans (n=9), siscowets (n=12) and Michipicoten
leans (n=10). A) Comparison of population marginal mean body weights (g) between lake trout
populations. Error bars indicate the 95% confidence interval of the population marginal means.
Letters indicate populations that are significantly different based on the results of an ANCOVA.
B) Relationship between body weight (1000 g) and age (years) between the Marquette lean,
siscowet and Michipicoten lean lake trout populations. Regression lines are included for each
2

2

population (Michipicoten leans: β=0.07, R =0.80; Marquette leans: β= 0.04, R =0.14 and
2

siscowets: β=0, R =0). Letters indicate populations that have significantly different slopes of the
regression line based on an ANCOVA. There were no significant differences between slopes.
34

Fecundity and gonad weight were compared between populations to determine
differences in reproductive investment. Michipicoten leans had the largest fecundities and gonad
weights (ANCOVA, age as the covariate; Table 5; Figure 7 and 8). Fecundities, adjusted for
body weight, showed no significant difference between the populations (Figure 9A). The
interaction between age and population type for body weight adjusted fecundities was also not
significant, however there was a significant difference in the slopes of the regression lines
between Michipicoten leans and Marquette leans, with Marquette leans changing production of
eggs with age more rapidly than Michipicoten leans (t=2.25, p=0.03; Figure 9B). The bodyweight-adjusted fecundity decreased with age in the Michipicoten leans and increased with age
in the Marquette leans. Similarly, Michipicoten leans had a significantly higher marginal mean
GSI than the Marquette leans (Figure 10A), however there was no difference in slopes of the
regression lines (Figure 10B). Analysis of fecundity, gonad weight and GSI all suggest that
Michipicotens produce more eggs with age, but this does not reflect an increase in relative
reproductive investment; Michipicotens get much larger with age (Figure 5B) and as a result can
produce more eggs.

35

Population Marginal
Mean Total Fecundity
(eggs)

A.

12000
10000
8000
6000
4000
2000
0

a
b
b

Michipicoten
Island lean

Marquette
lean

Marquette
siscowet

Total Fecundity (1000 eggs)

B.
18

Michipicoten leans
Marquette lean
Marquette siscowet

14

a

10

a

6

a
2
5

15

25

Age (years)
Figure 7. The marginal mean fecundities and the relationships between fecundity and age were
compared between the populations: Marquette leans (n=9), siscowets (n=12) and Michipicoten
leans (n=11). A) Comparison of population marginal mean fecundities (number of eggs)
between lake trout populations including data from Michipicoten lake trout that was not included
in this study. Error bars indicate the 95% confidence interval of the population marginal means.
Letters indicate populations that are significantly different to an alpha value of 0.05 based on the
results of an ANCOVA. B) Relationship between total fecundity (1000 eggs) and age (years)
between lean, siscowet and Michipicoten lake trout populations. Regression lines are included
2
2
for each population (Michipicoten leans: β=0.33, R =0.03; Marquette lean: β=0.12, R =0.27;
2

siscowet: β=0, R =0). There were no significant differences between slopes.

36

A.
Gonad Weight (g)

1400

a

1200
1000
800

b

600

b

400
200
0
Michipicoten Island
lean

Marquette lean

Marquette siscowet

B.
Michipicoten lean
Marquette lean
Marquette siscowet

Gonad Weight (g)

2500

a
1500

a
500

a

5

15

25

Age (years)

Figure 8. The marginal mean gonad weights and the relationships between gonad weight and age
were compared between the populations: Marquette leans (n=9), siscowets (n=12) and
Michipicoten leans (n=11). A) Comparison of population marginal mean gonad weights (g)
between lake trout populations including data from Michipicoten lake trout that was not included
in this study. Error bars indicate the 95% confidence interval of the population marginal means.
Letters indicate populations that are significantly different to an alpha value of 0.05 based on the
results of an ANCOVA. B) Relationship between gonad weight (g) and age (years) between the
lake trout populations. Regression lines are included for each population (Michipicoten leans:
2
2
2
β=0.01, R =0.01; Marquette leans: β=0.11, R =0.27, siscowets: β= -0.01, R =0.01). There were
no significant differences between slopes.
37

Fecundity adjusted for
body weight (total eggs/g)

A.

2
a
a

Fecundity adjusted for body weight
(total eggs/g)

Marquette lean

Marquette
siscowet

1
0.5

0
Michipicoten
Island lean

B.

a

1.5

2.0
a
1.5

1.0

b

ab

0.5

Michipicoten lean
Marquette lean
Marquette siscowet
5

15

25

Age years)

Figure 9. The marginal mean fecundities adjusted for body weight and the relationships between
adjusted fecundity and age were compared between the populations: Marquette leans (n=9),
siscowets (n=12) and Michipicoten leans (n=11). A) Comparison of population marginal mean
fecundity adjusted for body weight (total eggs/g) between lake trout populations including data
from Michipicoten lake trout that was not included in this study. Error bars indicate the 95%
confidence interval of the population marginal means. Letters indicate populations that are
significantly different to an alpha value of 0.05 based on the results of an ANCOVA. B)
Relationship between fecundity adjusted for body weight (total eggs/g) and age (years) between
the lake trout populations. Regression lines are included for each population (Michipicoten
2
2
2
leans: β= -0.03, R =0.29; Marquette leans: β=0.08, R =0.26; siscowets: β=0.01, R =0.02).

38

A.
Gonadosomatic Index

20
ab

16

b

12
8
4
0
Michipicoten Island
lean

B.

Marquette lean

25

Marquette siscowet

Michipicoten lean
Marquette leans
Marquette sicowet

Gonadosomatic Index

20
a

15

a
a

10

5

5

15
Age (years)

25

Figure 10. The marginal mean gonadosomatic indices (GSI) and the relationships between GSI
and age were compared between the populations: Marquette leans (n=9), siscowets (n=12) and
Michipicoten leans (n=11). A) Comparison of population marginal mean GSIs between lake
trout populations including data from Michipicoten lake trout that was not included in this study.
Error bars indicate the 95% confidence interval of the population marginal means. Letters
indicate populations that are significantly different to an alpha value of 0.05 based on the results
of an ANCOVA. B) Relationship between GSI and age (years) between the lake trout
populations. Regression lines are included for each population (Michipicoten leans: β= -0.39,
2
2
2
R =0.26, Marquette leans: β=0.75, R =0.28, siscowets: β= -0.09, R =0.06). There were no
significant differences between slopes.

39

The percentage of lipid in the muscle tissue was significantly different between
populations (ANCOVA; Table 5). Fatmeter readings were converted to percent lipid using the
2

following equation (R = 0.6117).

;
Siscowets had significantly higher levels of muscle fat compared to Michipicoten leans,
but were not significantly different than Marquette leans (Figure 11A). There were no changes
with age in muscle lipid concentrations for all stocks (Figure 11B).

40

Muscle Fat (% wt/w lipid)

A.

25

b

20

a

ab

Michipicoten Island
lean

Marquette lean

15
10
5

0

Marquette siscowet

Muscle Lipid Concentration
wt/w (% lipid)

B.
Michipicoten lean
Marquette lean
Marquette siscowet

30

25
20
15

a

10

a

5

a
5

15
Age (years)

25

Figure 11. The marginal mean lipid concentrations and the relationships between lipid
concentrations and age were compared between the populations: Marquette leans (n=9),
siscowets (n=12) and Michipicoten leans (n=10). A) Comparison of population marginal mean
lipid concentrations between lake trout populations. The lipid concentrations were taken from
muscle tissue and reported as a percentage in wet weight. Error bars indicate the 95%
confidence interval of the population marginal means. Letters indicate populations that are
significantly different to an alpha value of 0.05 based on the results of an ANCOVA. B)
Relationship between lipid concentration (% lipid wt/w) and age (years) between the lake trout
populations. Regression lines are included for each population (Michipicoten leans: β=0.07,
2
2
2
R =0.01; Marquette leans: β= -1.26, R =0.67; siscowets: β= -0.39, R =0.04). There were no
significant differences between slopes.

41

Contaminant Results
According to the AIC method, empirical support for multiple linear regression models to
predict muscle contaminant concentrations based on age, population type, liver concentration,
lipid concentration and fork-length was strongest for the model that included population type,
liver contaminant concentration (PCB or PBDE) and fork-length (Table 6). The PBDE model
also included muscle lipid concentration (Table 6).
Variables important to predicting the contaminant concentrations in the ovary were
different between contaminants. Using the AIC method, PBDE concentrations in the ovary were
best predicted by age, population type, PBDE muscle concentration, gonad weight and egg
diameter (Table 6). The PCB concentrations in the ovary were best predicted by age, population
type, muscle PCB concentration, fecundity and egg diameter (Table 6).
PBDE and PCB concentrations in the liver were predicted by nearly the same variables.
Multiple regressions using the AIC method found that PBDE liver concentrations were best
predicted by the population type, PBDE muscle concentration, muscle lipid concentration and
fork length (Table 6). The PCB liver concentrations were best predicted by the population type,
PCB muscle concentration and muscle lipid concentration (Table 6).

42

Table 6. Best models based on AIC regression method
Modeled to predict
PBDE muscle concentration

Variables

AIC
-46.85

R-squared
0.85

-52.70

0.90

-42.81

0.83

-56.40

0.77

-29.08

0.80

-17.94

0.69

Population
PBDE liver concentration
Muscle lipid concentration
Fork length
PCB muscle concentration
Population
PCB liver concentration
Fork length
PBDE ovary concentration
Age
Population
PBDE muscle concentration
Gonad weight
Egg diameter
PCB ovary concentration
Age
Population
PCB muscle concentration
Egg diameter
Fecundity
PBDE liver concentration
Population
Muscle lipid concentration
PBDE muscle concentration
Fork length
PCB liver concentration
Population
PCB muscle concentration
Lipid concentration

43

Liver PBDE and PCB concentrations showed significant changes in bioaccumulation
rates with age between the populations (ANCOVA). For liver contaminant concentrations,
results indicated a significant interaction between age and population; the slopes of the
populations’ regression lines were significantly different (Table 7). There were no significant
differences between the marginal means of the populations for both PBDEs and PCBs (Figure
12A and 12B). However, there were significant differences between the slopes of Michipicotens
and Marquette leans for both PBDEs and PCBs with the rate of PBDE and PCB accumulation in
the livers increasing with age for the Michipicoten leans and show a declining trend with age for
the Marquette leans (Figure 12C and 12D).
Liver concentrations can be considered an indicator of exposure levels, and when
Michipicotens and Marquette leans younger than 15 were compared for PBDE and PCB
concentrations, there were no significant differences in the slopes of the regression lines (Table
7; Figure 13A and B), indicating that the lean morphotypes were similarly exposed to
contaminants. In contrast, when Michipicotens and siscowets of the same age (≥age 20) were
compared, there was a significant difference in the marginal mean PBDE and PCB liver
concentrations with siscowets having much lower concentrations in the liver (Table 7; Figure
13C and D); thereby indicating that the siscowet morphotypes is potentially exposed to lower
levels of contaminants.

44

Table 7. Differences in PBDE and PCB concentrations in the liver tissue based on ANCOVA
results
Dependent Variable

Independent
Variables

Degrees of
freedom

F-value

P-value

1

0.42

0.5209

2
2

0.98
4.07

0.3891
0.0301*

5

4.95

0.0030*

1

2.28

0.1440

2
2

1.66
4.12

0.2118
0.0289*

5

9.41

<0.0001**

1

0.10

0.7579

1
1
3

0.24
1.02
0.72

0.6389
0.3379
0.5633

1

0.76

0.4066

1
1
3

1.00
1.84
0.83

0.3429
0.2076
0.5079

1

1.62

0.2347

1

10.93

0.0092*

2

10.83

0.0040*

1

1.21

0.3000

2

7.89

0.0204*

2

7.88

0.0105*

PBDE liver concentration (ng/g)
+

Age
Population
Interaction term
Total Model
PCB liver concentrations (ng/g)
+

Age
Population
Interaction term
Total Model
PBDE liver concentrations for
lake trout ≤ age 14 (ng/g)
+

Age
Population
Interaction term
Total Model
PCB liver concentrations for
lake trout ≤ age 14 (ng/g)
+

Age
Population
Interaction Term
Total Model
PBDE liver concentrations for
lake trout ≥ age 20 (ng/g)
+

Age
Population
Total Model
PCB liver concentrations for
lake trout ≥ age 20 (ng/g)
+

Age
Population
Total Model
+

Age was covariate for all ANCOVA models

* significant to <0.05
** significant to <0.001
45

B.

a

80
60

a

a

40
20
0

C.

D.
PCB liver concen. (1000 ng/g)

Michipicoten lean
Marquette lean
Marquette siscowet

180

PBDE liver concen. (ng/g)

Population Marginal Mean
PCB liver concen. (ng/g)

Population Maringal Mean
PBDE liver concen. (ng/g)

A.

140
a
100
60

b

ab

20
5

1000
800
600
400
200
0

a
a
a

2
Michipicoten lean
Marquette lean
Marquette siscowet

1.5
a
1

ab
0.5
b
5

15
25
Age (years)

15
25
Age (years)

Figure 12. The marginal means of PBDE and PCB concentrations in the liver were compared
between the populations along with the relationship between PBDE and PCB concentrations in
the liver and age. A) Comparison of population marginal mean PBDE liver concentrations
(ng/g) between lake trout populations. B) Comparison of the population marginal mean PCB
liver concentrations. Error bars indicate the 95% confidence interval of the population marginal
means. C) Relationship between PBDE liver concentrations (ng/g) and age (years) between the
2
lake trout populations (Michipicoten leans: β=0.20, R =0.87, n=9; Marquette leans: β=0.11,
2

2

R =0.05, n=9; siscowets: β=0, R =0, n=12). D) Relationship between PCB liver concentration
2

(ng/g) and age (years) between the lake trout populations (Michipicoten leans: β=0.13, R =0.91,
2

2

n=10; Marquette leans: β=-0.04, R =0.05, n=9; siscowets: β=0, R =0, n-12). Letters indicate
populations that have significantly different slopes of the regression line or population marginal
means based on the results of an ANCOVA.
46

Marquette lean
Michipicoten lean

12
8
4

a
a

0
4

Population Marginal Mean
PBDE liver concen. (ng/g)

C.

6

8

10 12
Age (years)

PCB liver concen. (100 ng/g)

B.

16

a

2

a
4

D.

a
200
160

120
80

6

14

240

Marquette lean
Michipicoten lean

10

Population Marginal Mean
PCB liver concen. (ng/g)

PBDE liver concen. (10 ng/g)

A.

b

40
0
Michipicoten Marquette
Island lean siscowet

6

8

10 12 14
Age (years)

1000
a

b

500

0
Michipicoten Marquette
Island lean siscowet

Figure 13. The marginal means of PBDE and PCB concentrations in the liver were compared
between the populations along with the relationship between PBDE and PCB concentrations in
the liver and age at particular age ranges. A) Relationship between PBDE liver concentration
(ng/g) and age (years) for ages less than or equal to 14. Regression lines are included for each
2
2
population (Michipicoten leans: β=0.20, R =0.39, n=4; Marquette leans: β=0.25, n=, R =0.17,
n=9). B) Relationship between PCB liver concentration (ng/g) and age (years) for ages less than
or equal to age 14. Regression lines are included for each population (Michipicoten leans:
2
2
β=0.17, R =0.72, n=9; Marquette leans: β= -0.08, R =0.06, n=4). C) Comparison of the
marginal mean PBDE liver concentrations for lake trout at or above age 20: siscowets (n=8),
Michipicoten leans (n=5). D) Comparison of the marginal mean PCB liver concentrations for
lake trout at or above age 20. Error bars indicate the standard errors of the population marginal
mean. Letters indicate populations that have significantly different slopes of the regression line
or population marginal means based on the results of an ANCOVA.
47

PBDE concentrations in the muscle tissue accumulated in significantly higher
concentrations in siscowets than in leans (ANCOVA, Table 8; Figure 14A; intercepts
significantly different (t=-2.37, p=0.03)). The difference in the rate of accumulation of muscle
PBDE concentrations with age across different populations made it difficult to compare muscle
PBDE concentrations directly. Therefore, I compared PBDE muscle concentrations at the mean
age 17, age 12 and age 22. At age 17 and 22, Michipicoten lake trout had significantly lower
concentrations of muscle PBDEs than siscowets; however, there were no significant differences
between the populations at age 12 (Figure 14B).
PCB muscle concentrations were significantly higher in siscowets than in Michipicoten
and Marquette leans. The statistical analyses were performed with the two outliers removed
from the Michipicoten population (see Methods), and when muscle PCB concentration was
compared with age across the three stocks, slopes were not significant, but the siscowet
intercepts were significantly higher than the other stocks (Table 8, Figure 15A). Marginal means
of PCB muscle concentrations for each stock also showed that siscowets had significantly higher
concentrations (Figure 15B).
To tease out life history effects on the muscle concentrations of contaminants, I corrected
for exposure by dividing the muscle concentration of contaminant by liver concentration of
contaminant; this procedure is justified because of the multiple regression (Table 6) indicated
that PCB and PBDE muscle concentrations could be predicted by their respective liver
concentrations. The PBDE muscle concentrations when adjusted for liver concentrations showed
that Michipicotens had significantly lower concentrations compared to siscowets, but siscowets
and Marquette leans were not significantly different from each other (Figure 16A and 16B).
When muscle PCB concentration was adjusted for liver concentrations, there were also

48

significant differences between populations (Table 8). Michipicotens and Marquette leans had
significantly lower PCB muscle concentrations than siscowets (Figure 16C and 16D).

Table 8. Differences in PBDE and PCB concentration in the muscle tissue based on ANCOVA
results
Dependent
Variable
PBDE muscle
concentration
(ng/g)

Independent
Variables

Degrees of
freedom

F-value

P-value

Age
Population

1

2.89

0.1016

2

5.50

Interaction term
Total Model

2
5

3.22
11.50

0.0105*
0.0570
<0.0001**

1

10.78

0.0030*

2

6.65

0.0048*

3

24.34

<0.0001**

1

0.93

0.3440

2

5.25

0.0122*

3

3.74

0.0232*

1

1.10

0.3035

2

9.06

0.0010**

3

14.33

<0.0001**

+

PCB muscle
concentrations
(ng/g)
+

Age
Population
Total Model
PBDE muscle
concentrations
adjusted for
liver
concentrations
+

Age
Population
Total Model
PCB muscle
concentrations
adjusted for
liver
concentrations
+

Age
Population
Total Model
+

Age was covariate for all ANCOVA models; * significant to <0.05; ** significant to <0.001

49

B.
Michipicoten lean
Marquette lean
Marquette siscowet

500

300

a
b

100

400

b

350
300
250
200
150

a

100
50
0

ab

5

Pop. Marginal Mean PBDE
muscle concen. (ng/g)

PBDE muscle concen. (ng/g)

A.

15
Age (years)

Michipicoten Marquette
Island lean siscowet

25

Figure 14. The relationship between PBDE concentrations in the muscle tissue and age were
compared between the populations: Marquette leans (n=9), siscowets (n=12) and Michipicoten
leans (n=10). Muscle PBDE concentrations were also compared between populations at three
ages. Comparing three ages are necessary for a strong comparison because of the differences in
the slopes of the regression lines for each population. A) Relationship between PBDE muscle
concentration (ng/g) and age (years). Regression lines are included for each population
2
2
(Michipicoten leans: β=0.15 R =0.76; Marquette leans: β=0.04, R =0.02; siscowets: β= -0.01,
2

R =0). Letters indicate populations that have significantly different slopes of the regression line
to an alpha value of 0.05 based on the results of an ANCOVA. B) Comparison of PBDE
muscle concentrations (ng/g) at age 17 and 22 years. Error bars indicate the standard errors of
the population marginal mean. Comparison of population marginal mean PBDE muscle
concentrations (ng/g) between lake trout populations for ages 12 years showed no significant
differences between populations. Letters indicate the significant difference between marginal
means. Error bars indicate the standard errors of the population marginal mean.

50

B.
5

Michipicoten lean
Marquette lean
Marquette siscowet

4
3

a
2
a

1

Pop. Maringal Mean PCB
Muscle Concen. (ng/g)

PCB muscle concen. (1000 ng/g)

A.

2100
1800
1500
1200
900
600
300
0

b
a

a

a
5

15
Age (years)

25

Figure 15. The marginal mean PCB concentrations in the muscle were compared between the
populations along with the relationship between PBDE and PCB concentrations in the muscle:
Marquette leans (n=9), siscowets (n=12) and Michipicoten leans (n=10). A) Relationship
between PCB muscle concentration (ng/g) and age (years). Regression lines are included for
2
2
each population (Michipicoten leans: β=0.14, R =0.76, Marquette leans: β=0.02, R =0;
2

siscowets: β=0.04, R =0.11). Letters indicate populations that have significantly different slopes
of the regression line based on the results of an ANCOVA. B) Comparison of the population
marginal means PCB muscle concentrations with two outliers removed (ng/g). Error bars
indicate the standard errors of the population marginal mean. Letters indicate population
marginal means across that are significantly different to an alpha value of 0.05 based on the
results of an ANCOVA.

51

B.
b

10
8
6
4
2
0

a

Adjusted PBDE muscle value

Adjusted pop. marginal mean
PBDE muscle values (muscle
concen. / liver concen.)

A.
ab

Michipicoten lean
Marquette lean
Marquette siscowet

14
10
a

6

2

a
5

5
4
3
2
1
0

Adjusted PCB muscle value

Adjusted pop. marginal mean
PCB muscle values (muscle
concen. / liver concen.)

C.
b

a

a

a
15
Age (years)

25

D.
5
4

a

3
2
1

a
a
5

15
Age (years)

25

Figure 16. The marginal means of PBDE and PCB concentrations in the muscle adjusted for liver
concentrations were compared between the populations along with the relationship between
PBDE and PCB adjusted muscle concentrations and age: Marquette leans (n=9), siscowets
(n=12) and Michipicoten leans (n=10). A) Comparison of PBDE muscle concentrations (ng/g)
that have been adjusted for liver PBDE concentrations (ng/g). Error bars indicate the 95%
confidence interval of the population marginal means. B) Relationship between the PBDE
muscle concentrations (ng/g) adjusted for PBDE liver concentrations (ng/g) and age (years)
2
between the lake trout populations (Michipicoten leans: β= -0.20, R =0.34; Marquette leans:
2

2

β=0.09, R =0; siscowets: β= -0.08, R =0.01). C) Comparison of PCB muscle concentrations
(ng/g) that have been adjusted for liver PBDE concentrations (ng/g). Error bars indicate the 95%
CIs of the population marginal means. D) Relationship between PCB muscle concentrations
(ng/g) adjusted for PCB liver concentrations (ng/g) and age (years) between the lake trout
2
2
populations (Michipicoten leans: β=0.04, R =0.11; Marquette leans: β=0.05, R =0.07;
2

siscowets: β=0.03, R =0.01). Letters indicate populations that have significantly different slopes
or population marginal means based on the results of an ANCOVA.
52

PBDE and PCB ovary concentrations did not differ significantly between populations
(Table 9). There were also no significant differences between the marginal mean PBDE or PCB
concentrations of the populations (Figure 17A and 17B). Siscowets had a significantly different
slope between PBDE concentration in the ovary and age compared to both the Marquette leans
and the Michipicoten leans (Figure 17C). The concentration of PBDE in the ovary tissue was
decreasing with age in the siscowets, but increasing with age for both lean populations. There
were no differences in rates of increase in ovary PCB concentration with age between
populations (Figure 17D).
However, despite the lack of significant differences between the marginal means, the
difference in slopes of the regression lines suggest that the Marquette and Michipicoten lean
populations are accumulating PBDEs into the ovary tissue at different rates based on age (Figure
17C). To further investigate this, the ages were truncated and siscowets were compared to
Michipicoten leans age twenty and older. Due to lack of Marquette leans older than age twenty,
Marquette leans were not included in this analysis. When the older ages were compared,
siscowets and Michipicotens did not have significantly different slopes between age and PBDE
or PCB concentrations (Figure 18A and 18B). However, siscowets older than 20 have a lower
marginal mean concentration of PBDEs and PCBs in their ovaries than Michipicotens in the
same age range (Figure 18C and 18D).

53

Table 9. Differences in PBDE and PCB concentrations in the ovary tissue based on ANCOVA
results
Dependent
Variable
PBDE ovary
concentrations
(ng/g)

Independent
Variables

+

Age
Population
Interaction term
Total Model

Degrees of
freedom

F-value

P-value

1

10.17

2
2

1.08
3.73

0.0038*
0.3553
0.0382*

5

7.89

0.0001**

1

34.64

2
3

0.72
18.79

<0.0001**
0.4981
<0.0001**

1

1.26

0.2884

1

22.09

0.0008**

2

17.81

0.0005**

1

1.05

0.3304

1

5.12

0.0471*

2

5.24

0.0277*

PCB ovary
concentrations
(ng/g)
+

Age
Population
Total Model
PBDE ovary
concentrations
for lake trout
≥ age 20 (ng/g)
+

Age
Population
Total Model
PCB ovary
concentrations
for lake trout
≥ age 20 (ng/g)
+

Age
Population
Total Model
+

Age was covariate for all ANCOVA models

* significant to <0.05
** significant to <0.001

54

B.
a

a

Michipicoten lean
Marquette lean
Marquette siscowet

C.
PBDE ovary concen. (10 ng/g)

Pop. Marginal Mean
PCB ovary concen. (ng/g)

a

50
40
30
20
10
0

D.

18

14
10
a

6

2

PCB ovary concen. (100 ng/g)

Pop. marginal mean PBDE
ovary concen. (ng/g)

A.

400

a

a

a

300
200

100
0

Michipicoten lean
Marquette lean
Marquette siscowet

8

a

6
a
4
2

a

b

a
5

15
Age (years)

25

5

15
25
Age (years)

Figure 17. The marginal means of PBDE and PCB concentrations in the ovary were compared
between the populations along with the relationship between PBDE and PCB concentrations in
the ovary and age. A) Comparison of population marginal mean PBDE ovary concentrations
(ng/g). Error bars indicate the 95% CI of the population marginal mean. B) Comparison of
population marginal mean PCB ovary concentrations (ng/g). C) Relationship between PBDE
ovary concentrations (ng/g) and age (years) between the lake trout populations (Michipicoten
2
2
leans: β=0.12, R =0.91, n=10; Marquette leans: β=0.13, R =0.61, n=9; siscowets: β=0.02,
2

R =0.02, n=10). D) Relationship between PCB ovary concentration (ng/g) and age (years)
2

between the lake trout populations (Michipicoten leans: β=0.09, R =0.78, n=10; Marquette
2

2

leans: β=0.09, n=9, R =0.24; siscowets: β=0.04, R =0.23, n=12). Letters indicate populations
that have significantly different slopes of the regression line or population marginal means based
on the results of an ANCOVA.
55

B.

Michipicoten lean
Marquette siscowet

80
60

a

40
a
20

0

20

22

24
26
Age (years)

Michipicoten lean
Marquette siscowet
8
a
6

a

4

2
0
20

28

C.

22

24
26
Age (years)

28

D.
70

a

Pop. Marginal Mean PCB
ovary concen. (ng/g) ≥ age 20

Pop. Marginal Mean PBDE
ovary concen. (ng/g) ≥ age 20

PCB ovary concen. (100 ng/g)

PBDE ovary concen. (ng/g)

A.

60
50
40
b

30
20
10
0

2000

a

1500

1000

b

500
0

Michipicoten Marquette
Island lean
siscowet
Figure 18. The marginal means of PBDE and PCB concentrations in the ovary were compared
between the populations along with the relationship between PBDE and PCB concentrations in
the ovary and age: Marquette siscowets (n=8) and Michipicoten leans (n=5). A) Relationship
between PBDE ovary concentration (ng/g) and age (years) for ages greater than age 20.
2
Regression lines are included for each population (Michipicoten leans: β= -0.04, R =0.11;
2

siscowets: β=0.07, R =0.24). B) Relationship between PCB ovary concentration (ng/g) and age
2

2

(years) for ages greater than 20 (Michipicoten leans: β=0.01, R =0; siscowets: β=0.06, R =0.24).
C) Comparison of the marginal mean PBDE ovary concentration for ages ≥20 years. Error bars
indicate the 95% confidence interval of the population marginal mean. D) Comparison of the
marginal mean PCB ovary concentrations for ages ≥20 years. Letters indicate populations that
have significantly different slopes of the regression line or population marginal means to an
alpha value of 0.05 based on the results of an ANCOVA.
56

Discussion
The results of this study indicate that there may be a potential link between lipophilic
contaminants, such as PCBs and PBDEs and the rate of bioaccumulation in different areas of
female lake trout’s body (muscle, liver and ovary) in a manner that is population specific (Table
6). The most striking differences in bioaccumulation patterns reveled in this study is that
Michipicoten lean lake trout older than age 20 have higher concentrations in the ovary tissue
compared to siscowets of the similar ages (Figure 18C and 18D). This finding has implications
for reproductive success and offspring survival of lean lake trout. Additionally, siscowets
accumulate higher concentrations of PCBs and PBDEs in the muscle (Figure 16A and 16C),
which is consistent with the higher concentrations of lipids in the muscle tissue and has been
supported in other studies (Miller et al., 2000). Aspects of the population that could contribute to
the specific bioaccumulation patterns found in this study include life history traits or habitat
specific environmental variables. The role of life history traits and habitat on PBDE and PCB
bioaccumulation patterns will be discussed along with how bioaccumulation patterns can cause
unique issues based on the area of accumulation.
It is not surprising that PBDEs and PCBs are transported and deposited throughout the
body of the fish; the chemical composition of PBDEs and PCBs allow them to be transported
with triglycerides and lipoproteins (Dulfer et al., 1998). Lipophilic contaminants can be
incorporated into bile salt micelles in the liver and then be transported throughout the body by
lipoproteins (Dulfer et al., 1998). Due to the high solubility of lipophilic contaminants by
lipoproteins, lipoproteins serve as the main transport of contaminants through the plasma (Dulfer
et al., 1998). More specifically, lipophilic contaminants can be transported into the ovaries by
binding with lipoproteins. Based on in vitro binding assays, it is believed that the interaction

57

between vitellogenin, a yolk-precursor protein, and a common lipophilic contaminant is a nonspecific association between the lipophilic contaminant and the large lipid moiety of the
vitellogenin protein (Monteverdi & Giulio, 2000). Lipoproteins such as vitellogenin can then
transfer lipophilic contaminant such as PBDEs and PCBs in to the eggs (Monteverdi & Giulio,
2000).

Life History Influence
Population was a consistent predictor of PBDE and PCB tissue concentrations (Table 6).
This suggests that life history traits could be influencing PBDE and PCB accumulation patterns
in lake trout. The difference in life history traits between lean and siscowet populations has been
previously reported (Eschmeyer & Phillips, 1965; Hansen et al., 1995; Bronte et al., 2007;
Ebener, 1998) and my study adds additional support to those findings. Below I discuss key life
differences before I describe how these life history patterns affect distribution of contaminants in
various tissues.
There were several key life history traits differences between the populations. The fork
length and body weight data from my study suggest that leans grow much faster than siscowets.
Across all age, both the fork lengths and body weights of siscowets were significantly lower than
the two populations of leans (Figure 5A and 6A). In addition, the Michipicotens have a faster
growth rate compared to siscowets and are maturing at a younger age than siscowets (Figure 5B).
This higher growth rate, shown by both lean populations, may lead to biodilution of
contaminants as contaminants are dispersed throughout the growing tissues (Lavigne et al., 2010;
Hammar et al., 1993) and may explain the higher concentrations of PBDEs and PCBs in the
muscle tissue of siscowets. Increased growth rates have been shown to heavily influence

58

bioaccumulation of PCBs in the body of fish (Ng & Gray, 2009). Although, the theory of
biodilution relies on the assumption that populations are exposed to similar levels of
contaminants and that populations are metabolizing and excreting contaminants at similar rates.
Populations of Altantic salmon (Salmo salar L) and lake trout demonstrate different metabolic
rates based on life history traits, such as growth rates and lipid storage (Rossignol et al., 2011;
Henderson & Anderson, 2002).
Across all the ages of the Michipicoten leans and Marquette leans, the Michipicoten lake
trout had higher GSIs although siscowets were not significantly different from either population
of leans (Figure 10A). Additional research that compares Michipicoten leans to other genetic
strains of lean lake trout has found that Michipicotens have increased reproductive effort with
age and they invest more effort into reproduction at older ages than the other genetic strains
(Felt, unpublished). This rate of reproductive investment with age was determined in a common
garden experiment and suggests that high reproductive investment at older ages is heritable (Felt,
unpublished).
The Marquette leans had significantly greater rate of increase in fecundity, adjusted for
body weight, with age than the Michipicoten leans (Figure 9B). This suggests that the Marquette
leans are producing more eggs as they age. However, the Marquette leans only ranged in age
from 6 to 15 and this lack of older leans from Marquette as well as the small sample size makes
it difficult to interpret this trend. Additionally, the Michipicoten leans had higher GSIs than
Marquette leans, but Marquette leans appear to be increasing fecundity with age. Perhaps since
Michipicotens span a range of ages that include older lake trout, the Michipicotens are producing
fewer larger eggs, whereas the Marquette leans are producing greater amounts of smaller eggs.
However, PCBs and PBDEs are endocrine disruptors and have the ability to disrupt the ability of

59

lake trout to reproduce (Halden et al., 2010). The exposure to PCBs has been linked to reduction
in fecundity (Rose et al., 2003). A difference in exposure rates between Marquette and
Michipicoten leans could lead to differences in fecundity due to endocrine disruption by PCBs
and PBDEs and exposure to other endocrine-disrupting pollutants.
There were no Marquette leans captured that were older than age 15, suggesting that
Marquette leans face higher mortality rates than the other populations; as a result, the Marquette
lean population is facing different pressures that result in the production of higher volumes of
eggs at younger ages (Figure 9B). However, the use of size-selective gill nets may have
contributed to the lack of older lean lake trout collected from Marquette (Sitar, personal
communication). Michigan DNR surveys have shown that Marquette lean lake trout live past the
age of 15, however the abundance of lean lake trout older than 15 near Marquette is low (Sitar,
personal communication).
The Michipicoten leans had significantly higher unadjusted fecundities and gonad
weights than the Marquette leans and the siscowets, particularly at older ages (Figure 7A and
8A). The higher fecundities and gonad weight of the Michipicotens indicate a greater investment
into reproduction than siscowets over their lifespan, which is consistent with other studies
regarding the energy investments of leans and siscowets (Hansen et al., 1995). However, the
leans and siscowets from Marquette did not have significantly different fecundities adjusted for
body weight, gonad weights or GSIs (Figure 9A, 8A, 10A). The Marquette leans are not
producing more eggs per gram of body weight than the Michipicotens, despite the age
differences. Additionally, comparing the leans and siscowets from Marquette, it appears that the
investment into reproduction is similar based on fecundity and gonad weight despite Marquette
leans being younger.

60

Marquette siscowets had significantly higher concentrations of lipids in the muscle
compared to Michipicoten leans (Figure 11A). The amount of muscle fat in lake trout has been
determined to be a heritable trait (Goetz et al., 2010; Felt, unpublished). When placed in
common environments and exposed to different rations of food, lake trout follow a predictable
pattern of lipid accumulation in the muscle that is likely heritable (Felt, unpublished). Work
done specifically on lean and siscowet lake trout found a genetic basis for the higher
concentrations of lipids in siscowets (Goetz et al., 2010). Siscowets contain higher amounts of a
gene in their liver containing an apolipoprotein AII gene, which is associated with increased
levels of plasma fatty acids and triglycerides (Goetz et al., 2010). While environmental plasticity
may still play a role in the higher lipid concentrations in the muscle tissue of siscowets, there
evidence to suggest that this trait is heritable.

Liver Contaminant Patterns
The patterns of PBDE and PCB accumulation in the liver appeared to be population
specific with Michipicotens accumulating contaminants at higher rates. The differences in
PBDE and PCB concentrations in the livers between the populations could be linked to the
energy demands required to maintain particular life history patterns, such as age structure.
Alternatively, the difference in accumulation rates of PBDEs and PCBs in the liver with age
should be considered within the limited age ranges of the Marquette leans and siscowets.
Although there were no differences in PBDE or PCB liver concentrations between the three
populations across all ages (Figure 12A and 12B), the Michipicotens and siscowets age 20 and
older were different with siscowets having lower liver concentrations of both PBDEs and PCBs
(Figure 13C and 13D). The higher concentrations of both PBDEs and PCBs in the liver of older

61

Michipicotens may be a result of the greater energy requirements for larger fish. Michipicotens
had much higher body weights and larger fork-lengths than the siscowets resulting from
increased somatic growth and increased energy intake. Basal metabolic rates increase with
increased body size (Moyle & Cech, 2004). Additionally, lean lake trout have higher metabolic
requirements of the swimbladder due to their higher specific density (Henderson & Anderson,
2002). The higher energy demands due to the larger body size of the Michipicoten leans require
more foraging and consumption of more prey items, and consequently more PCBs and PBDEs
through ingestion of contaminated prey (Hinton et al., 2008). Once a prey item is consumed, the
liver is one of the first organs to be exposed to lipophilic contaminants within the prey (Hinton et
al., 2008). While a study has found that the percentage of empty stomach space in leans and
siscowets differed only by about ten percent, the size of the stomach would be relative to body
size (Conner et al., 1993). Therefore, it is not surprising that fork length is a good predictor of
PBDE and PCB liver concentrations; the larger the fish, the more likely to have consumed
contaminated prey.
Michipicoten leans and Marquette leans demonstrated potentially different trends in
PBDE and PCB accumulation with age. Michipicoten leans appeared to accumulate more
contaminants in the liver with increasing age while Marquette leans appear to be stable, although
a larger sample size is necessary to confirm these trends (Figure 12C and 12D). The reason for
the increase of PBDEs and PCBs concentrations in the Michipicoten leans with age may be due
to the differences in growth rates and egg production. The Marquette leans in this study were all
under age 14, while Michipicoten leans ranged from 6 to 27 years old. Michipicoten leans could
be accumulating more contaminants in their livers because of the slowing of somatic growth and
the production of fewer eggs (O’Gorman et al., 1998); Michipicoten leans showed a decreasing

62

trend of fecundity adjusted for body weight with age whereas Marquette leans had an increasing
trend, although the small sample size makes it difficult to confirm these trends (Figure 9B). The
Marquette leans, through the higher production of eggs and use of lipids towards growth, could
sequester more lipophilic contaminants within muscle tissue and the ovaries. Although, when
Michipicoten leans and Marquette leans under age 14 years were compared there was no
difference in accumulation patterns with age. This suggests that perhaps the difference in
accumulation rates in the liver is due to the difference in exposure rates. The older Michipicoten
leans (greater than age 14) are likely consuming more fish and therefore in taking more
contaminants. The exposure through food intake would manifest within days and influence the
liver concentration because lipophilic contaminants can be eliminated from the liver within
weeks of exposure (Karjalainen et al., 2006). Although it is difficult to speculate, if Marquette
leans lived as old as Michipicotens, the pattern of PBDE and PCB accumulation in the liver
would be expected to increase with age until Marquette leans reached senescence and reduced
the elimination of contaminants through egg production.
The concentrations of PBDEs and PCBs in the liver were also predicted by muscle lipid
concentrations (Table 6). Muscle lipids could influence the concentration of lipophilic
contaminants in the liver if considered within the context of lipid mobilization. The stress of low
food availability and consequently low lipid concentrations in the body could mobilize lipids to
supplement energy demands (Paterson et al, 2005). When lipids are mobilized, lipophilic
contaminants such as PCBs and PBDEs would be moved with the lipids, which would result in
higher elimination rates of the contaminants. The elimination of the contaminants could cause
higher liver concentrations as the contaminants are mobilized and processed through the liver.
This concept has been supported by work on fasted and fed arctic charr (Salvelinus alpines)

63

exposed to PCBs (Vijayan et al., 2006). This study found an increase in PCB accumulation and
response of CYP1A enzyme in fasted fish, suggesting increased hepatic activity due to lipophilic
contaminants when food availability is low (Vijayan et al., 2006). Populations facing reduced
lipid concentrations in the liver may be more susceptible to effects of PCBs and PBDEs.
The relatively rapid elimination rate of PBDEs and PCBs from the liver suggested that
the liver could be used as a quantitative indicator of exposure that could be used to adjust for
exposure and tease out effects related to life history. To adjust for exposure, I made the
assumption that the muscle concentrations could be divided by the liver concentrations based on
information generated in my study and guided by information from previous research. In my
regression analyses, the liver concentrations were consistent predictors of muscle concentrations
and a previous study found that PCBs do not differ between the muscle and liver tissue of
rainbow trout (Janz et al., 1992). Also, as previously mentioned, PCBs have been found to be
removed from the liver within 14 days (Karjalainen et al., 2006); whereas muscle tissues
accumulate lipophilic contaminants at a slower rate than other organs based on the internal
transfer of the contaminants (Antunes et al., 2008). Additionally, lake trout in Lake Superior are
chronically exposed to PBDEs and PCBs based on historical understanding of accumulation
patterns, retention times in Lake Superior and delay in atmospheric deposition (Batterman et al.,
2007; Gewurtz et al., 2008). The continuous exposure to these contaminants allows the
utilization of the liver as an indicator of the long-term exposure rates. Therefore, muscle PBDE
and PCB concentrations were adjusted for exposure rates using liver concentrations (see Muscle
Contaminant Patterns below).

64

Muscle Contaminant Patterns
Muscle tissue also showed a population specific pattern of accumulation. A previous
study had found that the concentration of PCBs in the muscle tissue of siscowets increased with
the concentration of lipids in the muscle and concentrations were higher in siscowets than leans
(Miller et al., 2000). My study also showed that siscowets have greater percentages of body fat
and lower growth rates than leans, consistent with previous research (Eschenmeyer and Philips,
1965, Hansen et al., 1995). Also, in my study, siscowets had higher percentages of muscle fat
than Michipicoten leans based on fat meter readings from localized areas along the body (Figure
11A). Marquette siscowets had higher concentrations of PCBs than both lean populations and
higher concentrations of PBDEs than Michipicoten leans when compared at age 17 and 22
(Figure 14B and 15B). Interestingly, the Marquette siscowets showed a slight trend of increasing
accumulation of PBDEs in the muscle with age whereas the Michipicoten leans showed PBDEs a
decreasing trend with age (Figure 14A). The limited age range of the siscowets makes it difficult
to properly interpret patterns of accumulation with age; however, this difference may be due to
depuration of the PBDEs each spawning season and because Michipicoten leans have much
higher fecundities, more PBDEs are subsequently eliminated. On the other hand, PCBs did not
show a different rate of accumulation with age between the Marquette siscowets and the
Michipicoten leans (Figure 15A). The lack of difference in PCB accumulation with age may be
caused in a change in exposure to PBDEs in the relative locations of Michipicoten leans and
Marquette siscowets. This similar pattern of decreasing PBDE concentrations for Marquette
siscowets with age, but increasing patterns of PCBs with age was also shown in the ovary tissue
suggesting that lake trout are not exposed to PBDEs and PCBs in similar ratios; suggesting
contaminant exposure is site specific (see Ovary Contaminant Patterns below).

65

The difference in the rate of PBDE muscle accumulation between Michipicoten leans and
Marquette siscowets may also indicate a difference in exposure levels. When I adjusted the
muscle PBDE and PCB concentrations by liver concentrations and compared between
populations, Michipicotens had significantly lower PBDE muscle concentrations than siscowets
(Figure 16A). The PCB muscle concentrations of both the lean populations were also lower than
the siscowets (Figure 16C). The adjusted muscle data provides similar trends as the unadjusted
data; however after adjustment, there were no longer any differences in the rate of PBDE muscle
accumulation with age between the Marquette siscowets and the Michipicoten leans. This
suggests that the adjustment of muscle concentrations by liver concentrations corrected for
potential differences in exposure rates between the populations.
Muscle PBDE and PCB concentrations were predicted by fork-length (Table 6). The
influence of fork-length on accumulation of lipophilic contaminants is most likely caused by
increased exposure to contaminants because of increased consumption of contaminated prey
items to maintain a larger body size.

Ovary Contaminant Patterns
Based on the AIC method, the concentrations of PBDEs and PCBs in the ovary tissue
were predicted by egg diameter and fecundity or gonad weight, which are all life history specific
traits (Table 6). These results suggest that the amount of energy (in the form of lipoproteins)
taken up by the developing eggs (number and size) will directly influence the concentration of
these contaminants in the ovaries. The Michipicoten leans had significantly higher fecundities
and gonad weights than the siscowets, suggesting that reproductive investment lead to higher
accumulation of lipophilic contaminants in the ovaries.

66

The PBDE concentrations in the ovaries increased with age for Michipicoten leans, but
decreased with age for Marquette siscowets (Figure 17C). However, there was no difference in
accumulation rates for PCBs between Marquette siscowets and Michipicoten leans (Figure 17D).
This may be related to changes in exposure rates in the locations of the populations as similar
patterns were shown in the muscle tissue.
Additionally, age was a consistent predictor of PBDE and PCB ovary concentrations
(Table 6). When the ovary concentrations of PBDEs and PCBs in lake trout older than age 20
were compared, the Michipicoten leans had higher concentrations compared to siscowets (18A
and 18C).

Habitat Specific Influences
My research suggests that life history traits can influence the accumulation patterns of
PCBs and PBDEs; however there are also other factors indirectly related to life history traits that
could influence the intake of contaminants. Siscowets and leans have unique bathymetric
distributions that contribute to the life history differences. However, the bathymetric distribution
may also influence PBDE and PCB concentrations through diet and water temperature. Also,
exposure to PBDEs and PCBs could be attributed to location and the proximity to a pollution
source.
The differences in bathymetric distribution of leans and siscowets result in different diets,
which could influence accumulation and exposure of PBDEs and PCBs. Because PBDEs and
PCBs are highly lipid soluble, the common route of exposure involves trophic transfer (Wong et
al., 2004) and leans and siscowets have shown limited dietary overlap (Harvey & Kitchell, 2000;
Harvey et al., 2003). Rainbow smelt (Osmerus mordax) are the primary prey item for inshore

67

lake trout whereas siscowets consume deepwater and slimy sculpins (Myoxocephalus thompsonii
and Cottus congatus) (Conner et al., 1993). There is also a marked difference between inshore
and offshore lake trout diets. Offshore lake trout consume more coregonines (Coregonus sp.),
opossum shrimp (Mysidacida sp.) and sculpins (Conner et al., 1993). These dietary differences
need to be further explored to fully understand the effect on PCB and PBDE concentrations or
accumulation. Future studies could compare contaminant concentrations in prey fish and such
data would fill data gaps identified by my research.
In addition to diets, the bathymetric distribution may also result in different water
temperatures in the habitats occupied by each population, which could influence the
concentrations of PBDEs and PCBs. The half-life of PCB congeners decreased in rainbow trout
held in warmer water (Buckman et al., 2007). Additionally, biotransformation of PCBs also
increased with increasing water temperature (Buckman et al., 2007). Buckman’s study showed
decreases in biotransformation within a 4°C water temperature decrease. The surface
temperature of Lake Superior ranges between 2 and 4°C (NOAA, 2011) and the greater depth
occupied by the siscowet morphotype would lead to differences in the water temperature
occupied by leans and siscowets, which could influence biotransformation rates of contaminants,
including PCBs and PBDEs.
The location of each population and its relationship to industrialized settings could create
differences in exposure to PBDEs and PCBs. The Michipicoten leans inhabit waters near a less
industrialized area than the Marquette leans and siscowets. Michipicoten Island is uninhabited
by humans and the entire island is controlled by the Canadian government as a provincial park.
Whereas Marquette, Michigan is home to several industries, including a coal-fired power plant
located on the shores of Lake Superior (We-Energies, 2010). This could influence the

68

concentration of contaminants based on the localized pollution rates. However, PCBs and
PBDEs travel atmospherically and the more northern island of Michipicoten could result in
higher deposition rates of PCBs and PBDEs. The evaporation of contaminants in warmer areas
and subsequent condensation in colder climates near the poles has been termed ‘global
distillation’ and results in atmospheric deposition of persistent organic pollutants in remote areas
(Schmid et al., 2007; Datta et al, 199). Atmospheric deposition would be consistent with our
findings of higher liver concentrations of PBDEs and PCBs in Michipicotens older than age 20
compared to Marquette siscowets. The exposure rates of each population could have been
included as part of this study by including contaminant analysis of sediment and water samples
from each fish sampling site. Inclusion of exposure data would have strengthened the evidence
for differences in exposure rates between populations.
The concentrations of PBDEs and PCBs within each tissue did not always follow similar
patterns. PCB concentrations were higher than PBDE concentrations in all tissues. This can be
contributed to the longer period of exposure to PCBs, which began in the early 20th century
(Batterman et al., 2007; Gewurtz et al., 2008; USEPA, 2008a). PBDEs and PCBs also did not
accumulate in similar patterns within tissues. Siscowets and Michipicoten leans had different
rates of accumulation of PBDEs in the muscle and ovary tissue, while accumulation rates of
PCBs in these tissues were not significantly different. This seems to suggest that siscowets are
exposed to different concentrations of PBDEs than Michipicoten leans. The difference between
PCB and PBDE concentrations in this study suggests that populations are experiencing unique
exposure rates to both PBDEs and PCBs.

69

Health Effects of Contaminants
The elevated concentrations of PCBs and PBDEs in the ovary are of concern because of
the potential for adverse effects to the offspring. A previous study found that juvenile lake trout,
exposed to concentrations of PBDEs as low as 6 ng/g dry weight, had lower thyroxine levels
(T4). This thyroid disruption persisted for over 100 days after exposure (Tomy et al., 2004). In
my study, PBDE concentrations in the ovary tissue were found to range from 20 ng/g to 42 ng/g
reported in wet weight. This suggests that offspring of these lake trout may have decreased
levels of T4, which may disrupt the development of the immune system. PCBs have also been
found to directly affect the immune system in fish and cause a decrease in survival when exposed
to a bacterial pathogen (Maule et al., 2005). Furthermore, epidemiological studies suggest that
lake trout eggs show a negative correlation between egg hatchability and total PCB
concentrations (Mac et al., 1992). Strengthening the epidemiological study, a laboratory study
also found PCBs to be estrogenic and embryotoxic, causing caused delayed egg hatchability in
zebrafish (Danio rerio) (Olsson et al., 1999). When zebrafish hatched from mothers exposed to
PCBs were raised to adulthood, they had craniofacial malformations, severe curvature of the tail
and no development of the caudal fin (Olsson et al., 1999). PCB exposure can slow swimming
speeds and slow predator avoidance response, which could impact the survival rates of hatched
offspring (Rose et al., 2003).
The higher concentrations of PBDEs and PCBs found in the muscle tissues of lake trout
are of concern for the health of the adult individual and the population as a whole. A study has
found that exposure to a mixture of PCBs suppressed cortisol dynamics in arctic charr
experiencing handling stress (Jorgensen et al., 2002). Usually, the primary stress response in fish
is to increase cortisol secretion in efforts to maintain homeostasis (Mommsen et al. 1999). The

70

lake trout in Jorgensen’s study had muscle wet weight concentrations of 8.07 µg/g of PCBs,
whereas lake trout in my study had an average of 44.48 µg/g of the sum of the PCB congeners
measured. This suggests that lake trout in Lake Superior are exposed to concentrations of PCBs
high enough to potentially disturb cortisol excretion and therefore disrupt appropriate stress
responses. Additionally, PCBs can interact with the immune system. Lower levels of PCBs can
induce thymocyte apoptosis and higher concentrations can results in thymocyte necrosis (Sweet
et al., 1998). If thymocytes are reduced or damaged, the body may become more susceptible to
disease (Birchmeier et al., 2005). PBDEs have also been found to increase necrosis of
thymocytes and decrease viability (Birchmeier et al., 2005). Chinook salmon (Oncorhynchus
tshawytscha) exposed to an infectious pathogen, Listonella anguillarum, had significantly higher
mortality rates if the salmon had been exposed to PBDEs (Arkoosh et al., 2010). The salmon in
Arkoosh’s study had concentrations of PBDEs in their stomachs averaging 190 ng/g dry weight
(Arkoosh et al., 2010), whereas my study found an average muscle PBDE concentration of 135
ng/g wet weight. However, Arkoosh’s study used juvenile subyearling salmon and another study
has shown that juveniles are more sensitive to the effects of PCBs (Duffy et al., 2003). While
my study examined mature adult lake trout, it is important to also consider the effects to juvenile
lake trout exposed to PBDEs and PCBs.
The effects of PBDEs and PCBs have the potential to be detrimental to lake trout
populations in the Great Lakes. The ability for individuals of a population to not only produce
offspring, but to produce successful offspring that survive past the juvenile stage and continue
reproducing is required for the stability of a population. Exposure of adult lake trout to PBDEs
and PCBs can result in disruption of the immune system and difficulty maintaining homeostasis
due to stress. These effects may lead to the decreased production of offspring or reduced quality

71

of offspring. Additionally, the effect of PBDEs and PCBs on the success of the offspring due to
the maternal transfer of contaminants can result in failure of the offspring to reach maturity.
Although not reported, the observed tissue levels suggest that the combination of the effect of
PBDEs and PCBs on lake trout at all life stages has the potential to cause a population decline.

Assumptions, Limitations and Future directions
The population age structure differed greatly between populations making it difficult to
tease out the effects of life history on contaminant bioaccumulation because age plays an
important role in both life history traits and contaminant accumulation. The mature female leans
and siscowets from Marquette showed no overlap in ages (Figure 5). While this apparent age
difference suggests that each morphotype is experiencing different life history pressures, it
created complications for between population comparisons. Mature siscowets were not collected
at the spawning grounds younger than age 15, suggesting the siscowets are a slower growing
population that matures at an older age. Mature leans were not collected in this sample past the
age of 14, suggesting a higher mortality rate than the other populations and thus they reach
maturity at a younger age. Lean lake trout from Michipicoten had an age range that overlapped
both the leans and siscowets from Marquette. The Michipicoten lean lake trout mature as young
as the Marquette lean lake trout and live longer. The Michipicoten lean lake trout are not under
the same fishing pressure or predation from sea lamprey as the Marquette leans (Ebener, 1998)
and their life history is influenced by the lower adult mortality compared to the Marquette leans.
I addressed the issue of age by using an ANCOVA with age as a covariate for the statistical
analyses. In some cases, I truncated the ages at a particular year to compare between populations
of the same age for a stronger comparison. However, future studies should include more varied

72

populations of lake trout and sample lake trout from a variety of locations in order to avoid
comparing populations with no age overlap. Additionally, a lower minimum length used for
sampling may result in wider age ranges for each population as my study targeted mature lake
trout based on a minimum length requirement (Goetz et al., in review). The number of lake trout
analyzed was relatively low and the age range was limited for this study, so further studies
should include larger sample sizes for each population.
In addition to the difficulties encountered by the age structures of the populations, it is
difficult to separate the effect of life history traits and exposure due to habitat variations.
Populations examined in this study were exposed to different concentrations of contaminants due
to their habitats and prey items. I adjusted for exposure by dividing muscle concentrations by
liver concentrations. However, future studies should consider the influence of exposure and
include data collection and necessary controls for exposure that could be due to both prey items
consumed and habitat. Future sampling events should consider stomach contents, soil samples
and water samples, and compare these samples for differences in contaminant concentrations.
Such a sampling regime would allow for a better understanding of the influence of life history
traits on lipophilic contaminant bioaccumulation patterns.

Implications for Monitoring
While further research is needed to better understand the differences in life history and
their effects on lipophilic contaminants, fish monitoring programs that focus on lean lake trout
should consider life history when making decisions about consumption advisories. Lean and
siscowet lake trout can produce half-breeds that have an intermediate lipid concentration
(Eschmeyer and Phillips, 1965; Burnham-Curtis & Smith, 1994). There is currently no literature

73

that supports these half-breeds are truly genetic hybrids of leans and siscowets, however the
potential for inter-breeding makes it difficult to distinguish morphotypes without determining
lipid concentrations, which can complicate monitoring programs’ aims. The consumption of
lean and siscowet half-breeds may have health consequences for the public who are basing their
consumption rates on the consumption advisories for lean lake trout. Therefore, a better
understanding of lipid concentrations and lake trout morphotypes is necessary to properly guide
lake trout consumption warnings so that morphotypes with potentially higher contaminant
concentrations in their muscle are avoided. Fish consumption advisories would have to be based
on contaminant concentrations observed in the fattiest morphotypes so that human consumers
who would not be able to easily distinguish between a lean, a siscowet or a half-breed, are
protected.
Due to the presence of distinct morphotypes and half-breeds, and based on my data
suggesting a connection between PBDE and PCB concentrations and life history, fish advisory
programs need a comprehensive way to incorporate life history into fish advisories. The
inclusion of life history information in fish advisors is necessary to minimize human
consumption of halogenated pollutants. Federal fish contaminant monitoring in the United
States waters of the Great Lakes consists of a two part program, the first part is the targeted to
monitor contaminant trends and assess the effects of toxics on fish (US EPA, 2010a). This
program analyses lake trout from the Great Lakes by targeting a similar size fish (US EPA,
2010a) and PCBs and PBDEs are included in the list of contaminants analyzed. However, these
data are not used to issue fish consumption advisories. The U.S.’s federal program to monitor
sport fish using fillets was disbanded in order to create a program that monitors for emerging
contaminants. Each Great Lake state and tribe in the U.S. are responsible for issuing fish

74

consumption advice. The states of Wisconsin, Minnesota and Michigan issue a separate PCB
advisory for siscowet and lake trout, however no information was provided on the states’ website
distinguishing lean and siscowet lake trout (WDNR, 2010; MDH, 2010 and MDCH, 2010). The
Canadian government issues fish advisories through the Ontario Ministry of the Environment
(OMOE). A fisheries advisory guide is provided and includes advisories for PCBs, however
only one advisory is given for both morphotypes of lake trout (OMOE, 2011). The OMOE had
the most comprehensive guide because it divided Lake Superior into distinct sections and gave
individual advisories for each region and fish identification information; however there was no
mention of life history differences between morphotypes of lake trout (OMOE, 2011). Fisheries
managers releasing fish advisories to the public should present advisories unique to the lean and
siscowet morphotypes and should provide a description of each morphotypes and the way to
distinguish each morphotype.
In addition to directly consuming siscowets or siscowet/lean half-breeds, due to the
abundance and high lipid content of siscowets, they are being considered for use as high-grade
fish oil (Sitar et al., 2008). Siscowet lake trout contain high levels of omega-3 fatty acids (Wang
et al., 1990). This fish oil would be used as nutritional supplements and to produce high-value
fish meal. Fish oils have many human health benefits including prevention of cardiovascular
diseases (Marik & Varon, 2009). However, if siscowet store large amounts of lipophilic
contaminants in their muscle tissue, consumption of siscowet oils may pose an unacceptable risk
of human exposure to these contaminants.
Fisheries managers should also consider the impact of life history on the response of a
population to a stressor such as PBDEs and PCBs. A study modeled the effect of chronic
toxicity through immune suppression, reproductive toxicity and growth suppression on three

75

species of salmonids (Salmonidae sp.) (Spromberg & Meador, 2006). These salmonids shared a
reproductive strategy, but had different life spans, age of maturity and reproductive
contributions, similar to the lake trout populations used in my study. The work by Spromberg
and Meador demonstrated that contaminant exposure has population-level effects and that
populations can be uniquely susceptible to chronic contaminant exposure; therefore demographic
and life history traits should be included in risk assessments of contaminant exposure
(Spromberg & Meador, 2006).

Implications for Management: Multiple Stressors
Fisheries managers could use the knowledge of the influence of lipophilic contaminants
in tandem with other stressors to understand the lack of naturally reproducing lean lake trout
populations in the Great Lakes. The decline of lean lake trout populations resulted in the
establishment of stocking programs throughout the Great Lakes. However, stocking programs
have yet to see the successful restoration envisioned at the onset of the programs because there
has been no evidence of natural reproduction in Lake Michigan or Lake Erie (Madenjian et al.,
2008; Bronte et al., 2007; Cornelius et al., 1995). The cause for the decline of naturally
reproducing populations as well as the mediocre success of stocking programs could be
contributed to a variety of stressors faced at different life stages. During the adult life stage, lean
lake trout have been under increased fishing effort as a result of increased fishing efficiency
post-World War II (Wilberg et al., 2004). Lean lake trout have greater expression of immunerelated genes than siscowets, which may have been caused by greater exposure to pathogens
(Goetz et al., 2010). The parasitic sea lamprey (Petromyzon marinus), another invasive species,
attack and prey on adult lake trout. Studies conducted on lake trout who have suffered from sea

76

lamprey attacks, found mortality rates between 40% and 69% depending on the genetic strain of
the lake trout (Swink, 2003). The siscowet populations experience sea lamprey wounding that is
comparable to lean wounding rates (Sitar et al., 2008); however siscowets occupy greater depths
and lower host mortality from sea lamprey attacks has been associated with lower water
temperatures (Swink, 2003). Sea lamprey attacks also have sublethal effects. Changes in blood
serum levels were associated with stress from sea lamprey attacks (Edsall & Swink, 2001).
More research needs to be conducted to better understand the sublethal effects of sea lamprey
attacks and how it might interact with additional stressors as these stressors may exacerbate the
effect induced by lipophilic contaminants. As shown in my study, older lean lake trout have
higher concentrations of contaminants, which may also affect survivability and contribute to the
problem of multiples stressors on Great Lakes lake trout populations, thereby preventing
successful recovery.
Multiple stressors are also observed at the egg and larval stages. I observed significant
differences in age structures between lean and siscowet populations in my study. My study did
not find mature siscowets younger than age 16 years; however the mature leans were as young as
6 years (Figure 5). Other studies suggest that older fish produce higher quality offspring (Green,
2008; Venturelli et al., 2010). While there are additional physiological factors that can also
influence offspring viability, the age of the lake trout may influence the quality of the offspring.
Lean lake trout that are reproducing at a younger age may not have as high of a concentration of
PBDEs or PCBs in the eggs as older lake trout, but the lower quality of eggs may result in a
decreased ability of the offspring to overcome the effect of the contaminants. Additionally,
hatchery lake trout have a higher incidence of deformed fry and egg mortality (McDermid et al.,
2009). Other effects on larvae include thiamine deficiency resulting in early mortality syndrome

77

in fry (Fitzsimons et al., 1999). Swim-up syndrome is characterized by anorexia, loss of
equilibrium and death in lake trout fry and leads to slow growth, reduced foraging ability and
reduced predator avoidance (Fitzsimons, 1995; Fitzsimons et al., 2009). Physical disturbances,
such as wind fetch, can cause mortality to lake trout eggs (Fitzsimons et al., 2007). Finally, as
my study demonstrates, older lean lake trout may be delivering higher than average levels of
contaminants to their offspring.
Although studies have found that concentrations of certain contaminants were not high
enough to single-handedly cause mortality of lake trout fry, it has been suggested that exposure
to contaminants could interact with other factors to cause egg mortality (Fitzsimons, 1995). A
previous study determined that exposure levels to 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD)
in Lake Ontario and Lake Superior were below the no observable adverse effect level for lake
trout sac fry mortality (Guiney et al., 1996). TCDD can lead to symptoms resembling blue-sac
disease (Guiney et al., 1996) and certain congeners of PCBs can produce dioxin-like effects
(Tillitt et al., 2008). Lake trout sac fry are the most sensitive to the lethal effects of TCDD than
any other freshwater fish species investigated (Walker et al., 1991).
My study only quantified PCBs and PBDEs, however, there are other contaminants that
are ubiquitous that should be considered as stressors to Great Lakes lake trout populations. My
study only documented PCBs and PBDEs, but it seems that other lipophilic contaminants, such
as mercury, hexachlorodibenzene, polychlorinated dibenzo-p-dioxins and
dichlorodiphenyltrichloroethane (DDT) are likely to follow a similar pattern of accumulation in
lake trout and may be found if they had been measured as part of my study. A recent study
found trends towards synergistic effects of mercury and PCBs on brain function (Andersen et al.,
2009). Lake Superior fish, including lake trout, have the highest concentrations of mercury than

78

fish from the other Great Lakes (Bhavsar et al., 2010). Few studies that examine the mixture
effect of multiple contaminants, however, the potentially additive and synergistic effects of
contaminants should be considered in understanding the stressors faced by lake trout
populations.
Contaminants are only one of many stressors faced by lean lake trout of the Great Lakes;
however the additional stress of contaminants could contribute to the lack of natural recruitment
occurring in lean lake trout populations. Despite the low abundances of lean lake trout
populations, siscowet lake trout abundances have increased in the previous four decades (Sitar et
al., 2008). The increasing abundances have been contributed to lower fishing pressures on
siscowet populations, although the increase in abundances could perhaps be understood in part
by the distribution of lipophilic contaminants within the body of the lake trout; siscowet do not
show as high levels of contaminants in their ovaries as similarly aged leans.

Conclusions
My study suggests that there is a potential link between life history traits and lipophilic
contaminant accumulation patterns in lake trout populations. Life history traits are a
population’s response to environmental constraints and can result in different morphotypes.
Lake trout have evolved into the lean and siscowet morphotypes and within each morphotypes,
different populations have adopted different life history traits for different environments. When
the lean and siscowet morphotypes were compared, there were several major life history
differences, the most dominant difference that the siscowets allocate more energy into storage as
muscle fat, while the leans proportion more energy into reproductive efforts. These life history
differences between leans and siscowets have been well documented in other studies and my

79

results provide more evidence to this body of literature (Miller et al., 2000, Sitar et al., 2008,
Eschmeyer and Phillips, 1965; Hansen et al., 1995, Goetz et al., 2010). These life history
differences appear to impact the accumulation of lipophilic contaminants within the body, with
siscowets accumulating more contaminants in the muscle tissue and older leans transferring more
contaminants to their ovaries. However, these patterns of accumulation of lipophilic
contaminants can also be strongly linked to habitat-specific exposure. It is difficult to separate
direct exposure from life history because of how life history dictates potential for exposure
through habitat selection and energetic requirements due to age schedules and reproductive
efforts. Nevertheless, in addition to the apparent differences in life history traits, the Marquette
siscowet, Marquette lean and Michipicoten Island lean populations in my study show marked
differences in accumulation patterns of PBDEs and PCBs in the muscle and ovary tissues, which
may have implications for the health and survival of lake trout populations in the Great Lakes
and for human consumption.

80

LITERATURE CITED

81

Literature Cited

Andersen, I., O. Voie, F. Fonnum and E. Mariussen. 2009. Effects of Methyl Mercury in
Combination with Polychlorinated Biphenyls and Brominated Flame Retardants on the Uptake
of Glutamate in Rat Brain Synaptosomes: A Mathematical Approach for the Study of Mixtures.
Toxicological Sciences. 112(1): 175-184.
Antunes, P., A.J. Hendriks, M.A.J. Huijbregts, O. Gil and M.A. Reis-Henriques. 2004. Organspecific accumulation and elimination patterns of PCBs in adult seabass (Dicentrarchus labrax).
Science of the Total Environment. 407: 204-210.
Arkoosh, M., D. Boylen, J. Dietrich, B. Anuluacion, G. Ylitalo, C. Bravo, L. Johnson. F. Loge
and T. Collier. 2010. Disease susceptibility of salmon exposed to polybrominated diphenyl
ethers (PBDEs). Aquatic Toxicology. 98: 51-59.
Arnold, D. and M. Feeley. 2003. Polychlorinated Biphenyls. Food Safety: contaminants and
toxins. Ed. J.F. D.Mello. CABI Publishing, Cambridge, MA. 125-152.
Batterman, S., S. Chernyak, E. Gwynn, D. Cantonwine, C. Jia, L. Begnoche and J. Hickey. 2007.
Trends of brominated diphenyl ethers in fresh and archived Great Lakes fish (1979-2005).
Chemosphere. 69:444-457.
Bhavsar, S., S. Gewurtz, D. McGoldrick, M. Keir and S. Backus. 2010. Changes in Mercury
Levels in Great Lakes Fish between 1970s and 2007. Environmental Science and Technology.
44: 3273-3279.
Birchmeier, K., K. Smith, D. Passino-Reader, L. Sweet, S. Chernyak, J. Adams and G. Omann.
2005. Effects of selected polybrominated diphenyl ether flame retardants on lake trout
(Salvelinus namaycush) thymocyte viability, apoptosis, and necrosis. Environmental Toxicology
and Chemistry. 24(6): 1518-1522.
Bligh, E.G. and W.J. Dyer. 1959. A rapid method of total lipid extraction and purification.
Canadian Journal of Biochemistry and Physiology. 37(8): 911-917.
Bronte, C. R., M.E. Holey, C.P. Madenjian, J.L. Jonas, R.M. Claramunt, P.C. McKee, M.L.
Toneys, M.P. Ebener, B. Breidert, G.W. Fleischer, R. Hess, A.W. Martell Jr, and E.J. Olsen.
2007. Relative Abundance, Site Fidelity, and Survival of Adult Lake Trout in Lake Michigan
from 1999 to 2001: Implications for Future Restoration Strategies. North American Journal of
Fisheries Management. 27: 137-155.
Bronte, C.R. and S. A. Moore. 2007. Morphological Variation of Siscowet Lake Trout in Lake
Superior. Transactions of the American Fisheries Society. 136: 509-517.
Buckman, A., S. Brown, J. Small, D. Muir, J. Parrott, K. Solomon and A. Fisk. 2007. Role of
Temperature and Enzyme Induction in the Biotransformation of Polychlorinated Biphenyls and
82

Bioformation of Hydroxylated Polychlorinated Biphenyls by Rainbow Trout (Oncorhynchus
mykiss). Environmental Science and Technology. 41: 3856-3863.
Burnham-Curtis, M. and G. Smith. 1994. Osteological Evidence of Genetic Divergence of Lake
Trout (Salvelinus namaycush) in Lake Superior. Copeia. 4: 843-850.
Conner, D., C. Bronte, J. Selgeby and H. Collins. 1993. Food of Salmonine predators in Lake
Superior 1981-1987. Great Lakes Fisheries Commission. Technical Report 59.
Cornelius, F.C., K.M. Muth and R. Kenyon. 1995. Lake Trout Rehabilitation in Lake Erie: A
Case History. Journal of Great Lakes Research. 21(Supplement 1): 65-82.
Datta, S., K. Ohyama, D.Y. Dunlap and F. Matsumura. 1999. Evidence for Organochlorine
Contamination in Tissues of Salmonids in Lake Tahoe. Ecotoxicology and Environmental Safety.
42:94-101.
DeVault, D., R. Hesselberg, P. Rodgers and T. Feist. 1996. Contaminant Trends in Lake Trout
and Walleye from the Laurentian Great Lakes. Journal of Great Lakes Research. 22(4): 884-895.
DiGiulio, R. and D. Hinton, 2008. Introduction. Toxicology of Fishes. CRC Press, Boca Raton,
FL. 3-8.
Duffy, J., E. Carlson, Y. Prophete and J. Zelikoff. 2003. Age-related Differences in the
Sensitivity of the Fish Immune Response to a Coplanar PCB. Ecotoxicology. 12: 251-259.
Dulfer, W., H. Govers and J. Groten. 1998. Kinetics and conductivity parameters of uptake and
transport of polychlorinated biphenyls in the Caco-2 intestinal cell line model. Environmental
Toxicology and Chemistry. 17(3): 493-501.
Ebener, M. [Ed.]. 1998. A lake trout rehabilitation guide for Lake Huron. Great Lakes Fisheries
Commission. 48.
Edsall, C. and W. Swink. 2001. Effects of Nonlethal Sea Lamprey Attack on the Blood
Chemistry of Lake Trout. Journal of Aquatic Animal Health. 13: 51-55.
Eschmeyer, P. and A. Phillips. 1965. Fat Content of the Flesh of Siscowets and Lake Trout from
Lake Superior. Transactions of the American Fisheries Society. 94(1): 62-74.
Eshenroder, R. 2008. Differentiation of deep-water lake charr Salvelinus namaycush in North
American lakes. Environmental Biology of Fishes. 83(1): 77-90.
Fitzsimons, J. 1995. The Effect of B-Vitamins on a Swim-up Syndrome in Lake Ontario Lake
Trout. Journal of Great Lakes Research. 21(Supplement 1): 286-289.

83

Fitzsimons, J., S. Brown, D. Honeyfield and J. Hnath. 199. A Review of Early Mortality
Syndrome (EMS) in Great Lakes Salmonids: Relationship with Thiamine Deficiency. Ambio.
28(1): 9-15.
Fitzsimons, J., S. Brown, B. Williston, G. Williston, L. Brown, K. Moore, D. Honeyfield, D.
Tillitt. 2009. Influence of Thiamine Deficiency on Lake Trout Larval Growth, Foraging, and
Predator Avoidance. Journal of Aquatic Animal Health. 21: 302-314.
Fitzsimons J., J. Jonas, R. Claramunt, B. Williston, G. Williston, J. Marsden, B. Ellrott and D.
Honeyfield. 2007. Influence of egg predation and physical disturbance on lake trout Salvelinus
namaycush egg mortality and implications for life-history theory. Journal of Fish Biology. 71: 116.
Gewurtz, S., L. Shen, P. Helm, J. Waltho, E. Reiner, S. Painter, I. Brindle and C. Marvin. 2008.
Spatial Distributions of Legacy Contaminants in Sediments of Lakes Huron and Superior.
Journal of Great Lakes Research. 34: 153-168.
Goetz, F., D. Rosauer, S. Sitar, G. Goetz, C. Simchick, S. Roberts, R. Johnson, C. Murphy, C.
Bronte and S. Mackenzie. 2010. A genetic basis for the phenotypic differentiation between
siscowet and lean lake trout (Salvelinus namaycush). Molecular Ecology. 19(Suppl. 1): 176-196.
Goetz, F., S. Sitar, D. Rosauer, P. Swanson, C. Bonte, J. Dickey and C. Simchick. 2010. The
reproductive biology of siscowet and lean lake trout (Salvelinus namaycush) in southern Lake
Superior. Transactions of the American Fisheries Society. (in review).
Green, B.S. 2008. Maternal effects in fish populations. Advances in Marine Biology.51: 1-105.
Guiney, P., P. Cook, J. Casselman, J. Fitzsimons, H. Simonin, E. Zabel and R. Peterson. 1996.
Assessment of 2,3,7,8-tetraclorodibenzo-p-dioxin induced sac fry mortality in lake trout
(Salvelinus namaycush) from different regions of the Great Lakes. Canadian Journal of
Fisheries and Aquatic Sciences. 53: 2080-2092.
Halden, A., J. Nyholm, P. Andersson, H. Holbech and L. Norrgren. 2010. Oral exposure of adult
zebrafish (Danio rerio) to 2,4,6-tribromophenol affects reproduction. Aquatic Toxicology. 100:
30-37.
Hammar, J., P. Larsson and M. Klavins. 1993. Accumulation of persistent pollutants in normal
and dwarfed arctic char (Salvelinus-alpinus sp. complex). Canadian Journal of Fisheries and
Aquatic Sciences. 50(12): 2574-2580.
Hansen, M., J. Peck, R. Schorfhaar, J. Selgeby, D. Schreiner, S. Schram, B. Swanson, W.
MacCallum, M. Burnham-Curtis, G. Curtis, J. Heinrich and R. Young. 1995. Lake Trout
(Salvelinus namaycush) Populations in Lake Superior and Their Restoration in 1959-1993.
Journal of Great Lakes Research. 21(Supplement 1): 152-175.

84

Harvey, C. and J. Kitchell. 2000. A stable isotope evaluation of the structure and spatial
heterogeneity of a Lake Superior food web. Canadian Journal of Fisheries and Aquatic
Sciences. 57(7): 1395-1403.
Harvey, C. S. Schram and J. Kitchell. 2003. Trophic Relationship among Lean and Siscowet
Lake Trout in Lake Superior. Transactions of the American Fisheries Society. 132: 219-228.
Hawker, D. and D. Connell. 1991. An evaluation of the relationship between bioconcentration
factor and aqueous solubility. Chemosphere. 23(2): 231-241.
Henderson, B. and D. Anderson. 2002. Phenotypic differences in buoyancy and energetic of lean
and siscowet lake charr in Lake Superior. Environmental Biology of Fishes. 64: 203-209.
Hinton, D., H. Segner, D. Au, S. Kullman and R. Hardman. 2008. Liver Toxicity. Toxicology of
Fishes. Ed. R. DiGiulio and D. Hinton. CRC Press, Boca Raton, FL. 327-400.
James, M. Polychlorinated Biphenyls: Metabolism and Metabolites. PCBs: recent advances in
environmental toxicology and health effects. ED. L. Robertson and L. Hansen. The University
Press of Kentucky, Louisville, KY. 35-46.
Janz, D., T. Metcalfe and C. Metcalfe. 1992. Relative concentrations of cytochrome P450-active
organochlorine compounds in liver and muscle of rainbow trout from Lake Ontario. Journal of
Great Lakes Research. 18(4): 759-765.
Jorgensen, E., M. Vijayan, N. Aluru and A. Maule. 2002. Fasting modifies Aroclor 1254 impact
on plasma cortisol, glucose and lactate responses to a handling disturbance in Arctic charr.
Comparative Biochemistry and Physiology Part C. 132: 235-245.
Karjalainen, A., J. Pääkkönen and J. Karjalainen. 2006. Tissue-specific and whole-fish
accumulation of polychlorinated biphenyls by juvenile Baltic salmon (Salmo salar L.) after oral
gavage exposure. Boreal Environmental Research. 11: 421-430.
Kleinow, K., J. Nichols, W. Hayton, J. McKim and M. Barron. 2008. Toxicokinetics in Fishes.
Toxicology of Fishes. Ed. R. DiGiulio and D. Hinton. CRC Press, Boca Raton, FL. 55-152.
Lavigne, M., M. Lucotte and S. Paquet. 2010. Relationship between Mercury Concentration and
Growth Rates for Walleyes, Northern Pike, and Lake Trout from Quebec Lakes. North American
Journal of Fisheries Management. 30: 1221-1237.
Luthe, G., J. Jacobus and L. Robertson. 2008. Receptor interactions by polybrominated diphenyl
ethers versus polychlorinated biphenyls: A theoretical structure-activity assessment.
Environmental Toxicology and Pharmacology. 25(2): 202-210.
Mac, M.J. and T.R. Schwartz. 1992. Investigations into the effects of PCB congeners on
reproduction in lake trout from the Great Lakes. Chemosphere. 25:189-192.

85

Madenjian, C.P., M.P. Ebener and T.J. Desorcie. 2008. Lake Trout Population Dynamics at
Drummond Island Refuge in Lake Huron: Implications for Future Rehabilitation. North
American Journal of Fisheries Management. 28: 979-992.
Marik, P. and J. Varon. 2009. Omega-3 Dietary Supplements and the Risk of Cardiovascular
Events: A Systematic Review. Clinical Cardiology. 32(7): 365-372.
Maule, A., E. Jorgensen, M. Vijayan,and J.E. Killie. 2005. Aroclor 1254 Exposure reduced
disease resistance and innate immune responses in fasted artic charr. Environmental Toxicology
and Chemistry. 24(1): 117-124.
McDermid, J., W. Sloan, C. Wilson and B. Shuter. 2009. Early Life History Variation among
Hatchery- and Wild-Origin Lake Trout Reared in a Hatchery Environment. Transactions of the
American Fisheries Society. 139: 21-28.
Meredith, B. 2008. Michipicoten Island Wild Egg Collection. Ontario Ministry of Natural
Resources report.
Michigan Department of Community Health. 2010. 2010 Michigan Fish Advisory: A Family
Guide to Eating Michigan Fish. <
http://www.michigan.gov/documents/FishAdvisory03_67354_7.pdf>.
Miller, M. 1993. Maternal transfer of organochlorine compounds in Salmonines to their eggs.
Canadian Journal of Fisheries and Aquatic Sciences. 50(7): 1405-1413.
Miller, M. and S. Schram. 2000. Growth and Contaminant Dynamics of Lake Superior Lake
Trout. Journal of Great Lakes Research. 26(1): 102-111.
Minnesota Department of Health. 2010. Consumption Advice for Lake Superior. <
http://www.health.state.mn.us/divs/eh/fish/eating/lakesuperior.html>.
Mommsen, T., M. Vijayan and T. Moon. 1999. Cortisol in teleosts: dynamics, mechanisms of
action, and metabolic regulation. Reviews in Fish Biology and Fisheries. 9(3): 211-268.
Monteverdi, G.H., R. T. DiGiulio. 2000. Oocytic accumulation and tissue distribution of 2,3,7,8tetrachlorodibenzo-p-dixon and benzo[a]pyrene in gravid Fundulus heteroclitus. Environmental
Toxicology and Chemistry. 19(1): 2512-2518.
Moyle, P. and J. Cech Jr.. 2004. Fishes: an introduction to ichthyology. Prentice-Hall, Upper
Saddle River, NJ.
National Oceanic and Atmospheric Administration. 2011. Great Lakes Coastal Forecasting
System: Lake Superior. <
http://www.glerl.noaa.gov/res/glcfs/glcfs.php?lake=s&ext=swt&type=N&hr=48>.

86

Ng, C. and K. Gray. 2009. Tracking bioaccumulation in aquatic organisms: A dynamic model
integrating life history characteristics and environmental change. Ecological Modelling. 220:
1266-1273.
O’Gorman, R. and J. Elrod. 1998. Reproductive Potential and Fecundity of Lake Trout Strains in
Southern and Eastern Waters of Lake Ontario, 1977-1994. Journal of Great Lakes Research.
24(1): 131-144.
Olsson, P., L. Westerlund, S.J. The, K. Billsson, A. Hakan Berg, M. Tysklind, J. Nilsson, L.
Eriksson and D.E. Hinton. 1999. Effects of Maternal Exposure to Estrogen and PCB on Different
Life Stages of Zebrafish (Danio rerio). Ambio. 28(1): 100-106.
Ontario Ministry of the Environment. 2011. Guide to Eating Ontario Sport Fish.
Ontario Ministry of the Environment. 2004. The Determination of Polybrominated Dipheyl
Ethers (PBDEs) In Environmental Matrices by Gas Chromatography-High Resolution Mass
Spectrometry (GC-HRMS), method code BDE-E3430.
Page, K., K. Scribner and M. Burnham-Curtis. 2004. Genetic Diversity of Wild and Hatchery
Lake Trout Populations: Relevance for Management and Restoration in the Great Lakes.
Transactions of the American Fisheries Society. 133: 674-691.
Paterson, G., S. Huestis, D. Whittle, K. Drouillard and G. Haffner. 2005. In situ measurement of
tissue turnover and energy conversion efficiencies in lake trout (Salvelinus namaycush) using a
novel toxicokinetic approach. Canadian Journal of Fisheries and Aquatic Sciences. 62: 464-471.
Qunice, C., B. Shuter, P. Abrams and N. Lester. 2008. Biphasic growth in fish II: Empirical
assessment. Journal of Theoretical Biology. 254: 207-214.
Roff, D. 2002. Life History Evolution. Sinauer Associates, Sunderland, MA.
Rose, K., C. Murphy, S. Diamond, L. Fuiman and P. Thomas. 2003. Using Nested Models and
Laboratory Data for Predicting Population Effects of Contaminants on Fish: A Step Toward a
Bottom-Up Approach for Establishing Causality in Field Studies. Human and Ecological Risk
Assessment. 9(1): 231-257.
Rossignol, O., J. Dodson and H. Guderley. 2011. Relationship between metabolism, sex and
reproductive tactics in young Atlantic salmon (Salmo salar L.). Comparative Biochemistry and
Physiology. Part A 159: 82-91.
Schlenk, D., M. Celander, E. Gallagher, S. George, M. James, S. Kullman, P. van den Hurk and
K. Willett. 2008. 2008. Biotransformation in Fishes. Toxicology of Fishes. Ed. R. DiGiulio and D.
Hinton. CRC Press, Boca Raton, FL. 154-234.

Schmid, P., M. Kohler, E. Gujer, M. Zennegg and M. Lanfranchi. 2007. Persistent organic pollutants,
brominated flame retardants and synthetic musks in fish from remote alpine lakes in Switzerland.
Chemosphere. 67: 16-21.

87

Simoneau, M., M. Lucotte, S. Garceau and D. Laliberté. 2005. Fish growth rates modulate
mercury concentration in walleye (Sander vitreus) from eastern Canadian lakes. Environmental
Research. 98: 73-82.
Sitar, S., H. Morales, M. Mata, B. Bastar, D. Dupras, G. Kleaver and K. Rathbun. 2008. Survey
of Siscowet Lake Trout at Their Maximum Depth in Lake Superior. Journal of Great Lakes
Research. 34: 276-286.
Streets, S., S. Henderson, A. Stoner, D. Carlson, M. Simcik and D. Swackhamer. 2006.
Partitioning and Bioaccumulation of PBDEs and PCBs in Lake Michigan. Environmental
Science and Technology. 40: 7263-7269.
Streit, B., E. Siré, G. Kohlmaier, F. Badeck and S. Winter. 1991. Modelling ventilation
efficiency of teleost fish gills for pollutants with high affinity to plasma proteins. Ecological
Modelling. 57: 237-262.
Stromberg, J. and J. Meador. 2006. Relating chronic toxicity responses to population-level
effects: A comparison of population-level parameters for three salmon species as a function of
low-level toxicity. Ecological Modelling. 199: 240-252.
Sweet, L., D. Passino-Reader, P. Meier and G. Omann. 1998. Fish thymocyte viability, apoptosis
and necrosis: in-vitro effects of organochlorine contaminants. Fish and Shellfish Immunology. 8:
77-90.
Swink, W. 2003. Host Selection and Lethality of Attacks by Sea Lampreys (Petromyzon
marinus) in Laboratory Studies. Journal of Great Lakes Research. 29(Supplement 1): 307-319.
Tietge, J., R. Johnson, K. Jensen, P. Cook, G. Elonsen, J. Fernandez, G. Holcombe, D.
Lothenbach and J. Nichols. 1998. Reproductive toxicity and distribution of 2,3,7,8tetrachlorodibenzo-p-dioxin in adult brook trout (Salvelinus fontinalis) following a dietary
exposure. Environmental Toxicology and Chemistry. 17(12): 2395-2407.
Tillitt, D., P. Cook, J. Giesy, W. Heideman and R. Peterson. 2008. Reproductive Impairment of
Great Lakes Lake Trout by Dixon-Like Chemicals. Toxicology of Fishes. Ed. R. DiGiulio and D.
Hinton. CRC Press, Boca Raton, FL. 819-876.
Tomy, G., V. Palace, T. Halldorson, E. Braekevelt, R. Danell, K. Wautier, B. Evans, L.
Brinkworth and A. Fisk. 2004. Bioaccumulation, Biotransformation and Biochemical Effects of
Brominated Diphenyl Ethers in Juvenile Lake Trout (Salvelinus namaycush). Environmental
Science and Technology. 38: 1496-1504.
United Nations Environment Programme Chemicals. 2002. PCB Transformers and Capacitors:
From Management to Reclassification and Disposal. United Nations Environment Programme.
Issue 1.

88

United States Environmental Protection Agency. 2008a. Great Lakes Binational Toxics Strategy.
< http://www.epa.gov/glnpo/bnsdocs/pcbsrce/pcbsrce.html>.
United States Environmental Protection Agency. 2008b. Toxicological Review of 2,2’,4,4’,5Pentabromodiphenyl ether (BDE-99).
United States Environmental Protection Agency. 2010a. Great Lakes Monitoring. <
http://www.epa.gov/glnpo/monitoring/fish/index.html>.
United States Environmental Protection Agency. 2010b. Polybrominated Diphenyl Ethers
(PBDE) Project Plan.
United States Environmental Protection Agency. 2011. Pollution Prevention and Toxics:
Polybrominated diphenylethers. <http://www.epa.gov/oppt/pdbe>.
United States Environmental Protection Agency and Environment Canada. 2009. States of the
Great Lakes Technical Report.
United States Geological Society. 2010. Octanol-Water Partition Coefficient (Kow). <
http://toxics.usgs.gov/definitions/kow.html>.
Ungerer, J. and P. Thomas. 1996. Transport and Accumulation of Organochlorines in the Ovaries
of Atlantic Croaker (Micropogonias undulates). Marine Environmental Research. 42 (1-4): 167171.
Valters, K., H. Li, M. Alaee, I. D’Sa, G. Marsh, A. Bergman and R. Letcher. 2005.
Polybrominated diphenyl ethers and hydroxylated and methoxylated brominated and chlorinated
analogues in the plasma of fish from the Detroit River. Environmental Science and Technology.
39: 5612-5619.
Venturelli, P., C. Murphy, B. Shuter, T. Johnston, P. de Groot, P. Boag, J. Casselman, R.
Montgomerie, M. Wiegand and W. Leggett. 2010. Maternal influences on population dynamics:
evidence from an exploited freshwater fish. Ecology. 91(7): 2003-2012.
Vijayan, M., N. Aluru, A. Maule and E. Jorgensen. 2006. Fasting Augments PCB Impact on
Liver Metabolism in Anadromous Arctic Char. Toxicological Sciences. 91(2): 431-439.
Walker, M., P. Cook, A. Batterman, C. Berini, J. Libal, L. Hufnagle and R. Peterson. 1994.
Translocation of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) from adult female lake trout
(Salvelinus namaycush) to oocytes: Effects on early life stage development and sac fry survival.
Canadian Journal of Fisheries and Aquatic Sciences. 51:1410-1419.
Walker, M., J. Spitsbergen, J. Olson and R. Peterson. 1991. 2,3,7,8-tetrachlorodibenzo-p-dioxin
(TCDD) Toxicity during Early Life Stage Development of Lake Trout (Salvelinus namaycush).
Canadian Journal of Fisheries and Aquatic Sciences. (48):875-884.

89

Wang, Y., L. Miller, M. Perren and P. Addis. 1990. Omega-3 Fatty Acids in Lake Superior Fish.
Journal of Food Science. 55(1): 71-76.
We Energies. 2010. We Energies generating system: Presque Isle Power Plant. < http://www.weenergies.com/home/PresqueIsle.pdf>.
Wilberg, M., C. Bronte and M. Hansen. 2004. Fleet Dynamics of the Commerical Lake Trout
Fishery in Michigan Waters of Lake Superior during 1929-1961. Journal of Great Lakes
Research. 30(2): 252-266.
Wisconsin Department of Natural Resources. 2010. 2010 PCB advisory. <
http://dnr.wi.gov/fish/consumption/FishAdvPCBs2010lo.pdf>.
Wong, C., S. Mabury, D.M. Whittle, S. Backus, C. Teixeira, D. Devault, C. Bronte and D. Muir.
2004. Organochlorine Compounds in Lake Superior: Chiral Polychlorinated Biphenyls and
Biotransformation in the Aquatic Food Web. Environmental Science and Technology. 38: 84-92.

90