. V . .. .. I. r. , ‘. ‘m- ...
’ . . ' .

I] I
til I

 

-._u.-r-..,.x....u. .

     

 

 

4 . ‘ .
. . ,
' .

 

LIBRARY
Michigan State
University

 

 

 

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

DATE DUE DATE DUE DATE DUE

 

 

ENERGY DYNAMICS OF LAKE MICHIGAN CHINOOK SALMON
By

Amber Keasey Peters

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

DOCTOR OF PHILOSOPHY
Department of Fisheries and Wildlife

2006

ABSTRACT

This project investigated the life history and energy dynamics of Chinook salmon,
Oncorhynchus tshawytscha. Age at maturity and two surrogates for growth rate, size at
maturity and size at age, were compared for 16 populations of Chinook salmon
worldwide and within six populations from annual data. Among populations, the trend of
decreasing age at maturity with increasing size at age was clear. Trends among
populations were more difficult to detect.

The energy dynamics of Chinook salmon in Lake Michigan were analyzed using
proximate composition analysis. Caloric content, protein, lipid and water were measured
for 3 size classes in the spring and the fall from locations on both the eastern and western
sides of Lake Michigan. ANOVA of changes in lipid content showed that small fish had
significantly lower lipid levels than medium or large fish. The seasonal patterns of lipid
dynamics were different between small and large individuals. Specifically, small fish
show a decline in lipid levels from fall to spring while large fish show a decline in lipid
levels from spring to fall. The energy allocation strategy shown by each size class was
different and small Chinook salmon seem to be using a strategy which could lead to a
high risk of severe nutritional stress, especially in the novel environment of Lake
Michigan.

Results from the energetic study were used to determining an efficient method for
monitoring the nutritional status of the Lake Michigan Chinook salmon population.
Condition factor performed poorly as an indicator of whole-fish lipids (r2=0.07). Water

content in a dorsal muscle plug was

found to be correlated with whole-fish lipids (r2=0.50) for all samples. For the subset of
samples that included small fish collected in the spring, the strength of the relationship
between water content in a dorsal muscle plug and whole-fish lipids increased (r2=0.70).
The metric of water content in a dorsal muscle plug was determined to provide an
adequate surrogate of whole-fish lipid content and, therefore, overall nutritional status.
We propose a monitoring program that involves collecting small individuals in the spring
and reporting the proportion of samples with over 78% water content in muscle tissue.
Small individuals collected in the spring had the lowest whole-fish lipid levels of any
segment of the population and would be the most prone to nutritional stress; therefore we
recommend focusing on them for monitoring.

A model was built which coupled a bioenergetics approach with dynamic
programming to estimate the optimal strategy of energy allocation on a monthly basis for
Chinook salmon in environmental conditions model for Lake Michigan and for the
Pacific. Final egg mass was used a surrogate measure for fitness and length and lipid
levels were tracked throughout the monthly time steps. The optimal energy allocation
strategy differed for the two modeled environments, with Chinook salmon modeled in
Lake Michigan growing more rapidly and maintaining lower lipid reserves than those

modeled in the Pacific environment.

ACKNOWLEDGMENTS

I am indebted to many people for helping me during my doctoral program. I
especially thank my major advisor, Dr. Michael Jones, who was constantly guided and
supported me, answered endless questions, and prodded me along. Mike has become my
gauge of what a superb mentor should be and it has been insightful and delightful to work
with him. I thank the rest of my graduate committee members, Drs. Jim Bence, Mary
Bremigan, and Mohamed Faisal, for providing assistance and a well-balanced knowledge
base. Special thanks to Drs. Faisal and Ehab Elsayed and Dale Honeyfield for teaching
me laboratory techniques. Dr. Mike Wilberg mastered dynamic programming and wrote
a great model for me. Nathan Nye was my tireless lab technician never complained while
grinding and analyzing salmon parts. Other assistance was provided by Dr. Elizabeth
Marschall, Dr. Marc Trude], Jonathan Petit, and Steve Nimcheski. Data and advice used
in chapter 1 were collected and provided by Dr. Kim Scribner, Paul Peeters, Dan Bishop,
Don Schreiner, Llyod Mohr, Mark Ebner, Molly Negus, Dr. James Johnson, Dr. Martin
Unwin, Dr. David Welch, Roger Bergstedt, Randy Claramunt, and Dr. Thomas Quinn.
Collection, transporting and analysis of samples was done by Fritz Peterson, Ken Jeske,
Brad Eggold, Mike Toneys, Jerry Rackozy, Greg Wright, Dave Clapp, and the Steelhead
crew. Laboratory space and training was provided by Dr. John Riebow, Bob Burnett, Dr.
Steven Rust, Dr. Mohamed Faisal, Dr. Ehab Elsayed, and Dr. Alicia Orta-Ramirez.

The J ones/Bence lab provided encouragement, assistance, and much needed
distraction. Thank you to my parents for their years of encouragement. Finally, and most

importantly, I thank my husband, James Peters for his love and his unconditional support

iv

and my daughter, Avery Peters for the laughter and smiles she constantly provides in my
life.

This research was supported in part by the Great Lakes Fishery Trust, the United
States Geological Survey, the Great Lakes Fishery Commission, the US. Geological
Survey, the US. Fish and Wildlife Service Federal Aid for Sportfishing Restoration
program through the Michigan Department of Natural Resources, the Michigan
Department of Environmental Quality, and the Michigan State University College of

Agriculture and Natural Resources.

TABLE OF CONTENTS

LIST OF TABLES ........................................................................................................... viii
LIST OF FIGURES ............................................................................................................ x
INTRODUCTION .............................................................................................................. 1
Objective One .............................................................................................................. 2
Objective Two ............................................................................................................. 2
Objective Three ........................................................................................................... 3
Objective Four ............................................................................................................. 4
References .................................................................................................................... 5
CHAPTER 1

WITHIN AND AMONG POULATION RELATIONSHIPS BETWEEN GROWTH
RATE AND AGE AT MATURITY OF CHINOOK SALMON (ONCHORHYNCHUS

TSHA WYTSCHA) IN NATIVE AND INTRODUCED POPULATIONS .......................... 6
Introduction .................................................................................................................. 6
Methods ..................................................................................................................... 12
Results ........................................................................................................................ 13
Discussion .................................................................................................................. 14
References .................................................................................................................. 20

CHAPTER 2 ..................................................................................................................... 35

FACTORS AFFECTING ENERGY DYNAMICS IN LAKE MICHIGAN, CHINOOK

SALMON, ONC ORH YNCH US T SHA WYTSCHA ........................................................... 35
Abstract ...................................................................................................................... 35
Introduction ................................................................................................................ 35
Methods ..................................................................................................................... 39
Results ........................................................................................................................ 44
Discussion .................................................................................................................. 46
References .................................................................................................................. 51

CHAPTER 3 ..................................................................................................................... 62

A PROPOSED MONITORING PROGRAM FOR LAKE MICHIGAN CHINOOK

SALMON (ONC ORH YNCH US TSHA WYTSHCA) .......................................................... 62

vi

Abstract ...................................................................................................................... 62

Introduction ................................................................................................................ 62

Methods ..................................................................................................................... 66

Field Study ..................................................................................................... 66

Winter Simulation Study ............................................................................... 68

Data Analysis ................................................................................................. 69

Results ........................................................................................................................ 71

Discussion .................................................................................................................. 73

References .................................................................................................................. 80

CHAPTER 4 ..................................................................................................................... 92
OPTIMAL ENERGY ALLOCATION FOR CHINOOK SALMON, ONCORHYNCHUS

T SA WYTSCHA, IN THEIR NATIVE AND A NOVEL ENVIRONMENT ..................... 92

Introduction ................................................................................................................ 92

Methods ..................................................................................................................... 95

Bioenergetics Model ...................................................................................... 95

Dynamic Programming Model ...................................................................... 98

Results ...................................................................................................................... 102

Optimal energy allocation and model sensitivity ........................................ 102

Optimal age at maturation ........................................................................... 106

Virtual transplant of Pacific strategy to Lake Michigan .............................. 106

Discussion ................................................................................................................ 107

References ................................................................................................................ 1 12

vii

LIST OF TABLES

Table 1.1. Location, years sampled, race and aging method for 16 populations of
Chinook salmon. ....................................................................................................... 25

Table 1.2. Relationship between mean age at maturity and mean length at age 3 by brood
year for six populations of Chinook salmon. ............................................................ 26

Table 2.1. Summary of samples collected by location, sex, maturity status, and size class.
................................................................................................................................... 54

Table 2.2. AIC differences from the best model for all possible combinations of fixed
effects and higher interactions where SN=season, Size=size class, and
Loc=Location. ........................................................................................................... 55

Table 2.3. ANOVA for Lipid Content in the Whole Fish. .............................................. 56

Table 2.4. Mean calories (per gram dry), % lipids (per gram dry), % protein (per gram
dry), and % water of small, medium and large fish by season with standard error of
mean and sample size (n). ......................................................................................... 57

Table 3.1. Summary of samples collected by location, sex, maturity status, and size class.
................................................................................................................................... 83

Table 3.2. Sample size required for desired absolute error for two measures of proximate
composition for small, spring samples ...................................................................... 84

Table 4.1 Optimal Lipid Allocation Strategy for Chinook salmon modeled in Lake
Michigan and in the Pacific beginning at age 1 in April and maturing at age 3 in
September. .............................................................................................................. 1 15

Table 4.2. Final values at time of maturity for weight (g), length (mm), % gonad (per
gram whole fish) and final egg mass (a surrogate measure of fitness) for age 3 fish
modeled under Lake Michigan condition, Pacific conditions, and fish employing the
optimal strategy for the Pacific model, but transplanted to Lake Michigan
conditions. ............................................................................................................... 1 16

Table 4.3. The effect of increasing energy allocation from the optimal strategy to a 10%
in somatic growth and to a 10% increase in lipid reserves on length, weight, the
percent of the whole body composed of gonads and final egg mass for both fish
modeled under the Pacific and the Lake Michigan environmental conditions. ...... 117

Table 4.4. Expected final egg mass (g) in April of Year 2 for fish modeled in the Lake

Michigan environment with age at maturation set to 2, 3, and 4. Results are shown
under 4 different mortality regimes: 0.1, 0.2, 0.3, and 0.4. .................................... 118

viii

Table 4.5. Expected final egg mass (g) in April of Year for fish modeled in the Pacific
................................................................................................................................. 119

Table 4.6. Lipid levels (percent lipid per gram whole fish) during June and August of
years 1-3 for fish modeled under Pacific conditions, Lake Michigan conditions, and
actual fish collected and analyzed from Lake Michigan ......................................... 120

ix

LIST OF FIGURES

Figure 1.1. Hypothesized nonlinear relationship between juvenile mortality and growth
rate, where mortality falls most rapidly with increasing growth rate when growth
rate is low. ................................................................................................................. 27

Figure 1.2. A hypothesized dome-shaped relationship between age at maturity and
growth rate, with age at maturity reaching a maximum at an intermediate growth
rate ............................................................................................................................. 28

Figure 1.3. Geographic location of 16 Chinook salmon populations. l-Andreafsky,
Alaska. 2-Tuluksak, Alaska. 3-Kwethluk, Alaska. 4-Washington. S-Bonneville
Dam Ocean Type, Oregon. 6-Bonnville Dam Stream Type, Oregon. 7-Lake Oahe,
South Dakota. 8-Lake Superior. 9-Lake Michigan. lO-Lake Huron. ll-Lake
Ontario. 12-Glenariffe Stream, New Zealand. l3-Hydra Waters, New Zealand. 14-
Rakaia River, New Zealand. lS-Rangitata River, New Zealand. 16-Waitaki River,
New Zealand. ............................................................................................................ 29

Figure 1.4. The relationship between mean age of maturity and mean length at maturity,
a surrogate for growth rate, for 16 populations of Chinook Salmon (r = -0.85,
p<0.0001). A mean value for populations with multiple years of data is shown. 30

Figure 1.5. The relationship between mean age of maturity and mean length at maturity,
a surrogate for growth rate, for 16 populations of Chinook Salmon with values for
all populations for all years (r =0.57, p<0.0001). AKA-Andreafsky, Alaska. AKT-
Tuluksak, Alaska. AKK-Kwethluk, Alaska. WAS-Washington. ORO-Bonneville
Dam, Ocean Type, Oregon. ORS-Bonnville Dam, Stream Type, Oregon. SDK-Lake
Oahe, South Dakota. NZG. NZH—Hydra Waters, New Zealand. NZR-Rakaia
River, New Zealand. NZN-Rangitata River, New Zealand. NZW-Waitaki River,
New Zealand. ............................................................................................................ 31

Figure 1.6a. The relationship between mean age at maturity and mean length at age 3, a
surrogate for growth rate, for populations of Chinook Salmon by brood year for (a)
Lake Huron ............................................................................................................... 32

Figure 1.6b. The relationship between mean age at maturity and mean length at age 3, a
surrogate for growth rate, for populations of Chinook Salmon by brood year for (b)
Glenariffe Stream, New Zealand. Note: scales differ between populations. .......... 33

Figure 1.7. The relationship between mean age at maturity and survival rate, measured as
percent return to weir, for Strawberry Creek Weir, Lake Michigan. ........................ 34

Figure 2.1. Pearson correlation for caloric content per gram wet mass for (a) % lipid and
(b) % protein. Percentage data was transformed using the arcsine, square root
transformation. .......................................................................................................... 58

Figure 2.2. The joint effects of size class and location of collection on % lipid per gram
dry mass. Least squares means from ANOVA (back-transformed) are shown with
bars at 95% confidence intervals, p<0.01. ................................................................ 59

Figure 2.3. The joint effects of year and season on % lipid per gram dry mass. Least
squares means from ANOVA (back-transformed) are shown with bars at 95%
confidence intervals, p<0.0001. ................................................................................ 60

Figure 2.4. The joint effects of size class and season on % lipid per gram dry mass.
Least squares means from ANOVA (back-transformed) are shown with bars at 95%
confidence intervals, p<0.00001. .............................................................................. 61

Figure 3.1. Scatterplots of whole-fish lipid content versus (a) caloric content r2=0.79,
p<0.001 (b) condition factor r2=0.07, p<0.001 (c) water content r2=0.68, p<0.001
(cont.) ........................................................................................................................ 85

Figure 3.1 continued. Scatterplots of whole-fish lipid content versus ((1) hepatosomatic
index r2=0.008, p<0.001and (e) protein content r2=0.09, p<0.001. .......................... 86

Figure 3.2. Scatterplots of lipid content in whole-fish versus (a) lipid content in muscle
r2=0.61, p<0.001 (b) lipid content in liver r2=0.1 l, p<0.001 (cont.) ....................... 87

Figure 3.2 continued. Scatterplots of lipid content in whole-fish versus (c) water content
in muscle r2=0.50, p<0.001 and (d) water content in liver r2=0.01, p=0.19. ........... 88

Figure 3.3. Results from the winter simulation study showing means with standard errors
for (a) percent whole body lipid content for each treatment and (b) percent water in
the muscle tissue for each treatment. ........................................................................ 89

Figure 3.4. Required sample size (n) to estimate the proportion of the population in the
critical range for nutritional stress when that true proportion in the population ranges
from 0.05 to 0.50. Sample size requirements are shown for three levels of desired
absolute error (d) as 50%, 40%, and 60% of p. Lines illustrate the following
example: the true proportion of fish that exceeded the critical threshold, p, was 0.20
and the estimate of the proportion of the population which exceeded the critical
threshold was set to fall between 0.1 and 0.3 (ie. d is set to 50% of p) 95% of the
time, then a sample size of 60 would be required. .................................................... 90

Figure 3.5. Comparison of the proportion of the population with water content in the
muscle above the threshold of 78% with the proportion of the population with
whole-fish lipid content below the threshold of 2.45% for all combinations of size
class and collection periods (r2=0.94). A 1:1 Line is provided for reference. ......... 91

xi

Figure 4.1. Temperature values (°C) for the Lake Michigan and Pacific environments
used in the model (unpublished data from Roger Bergstedt and Dr. David Welch).
................................................................................................................................. 121

Figure 4.2. Monthly surplus energy (calories) available for allocation (above horizontal
line) or monthly energy deficit (below horizontal line) for Chinook salmon modeled
in Pacific and in Lake Michigan and set to mature at age 3. .................................. 122

Figure 4.3. Weight (g) of Chinook salmon modeled in the Pacific and in Lake Michigan,
the Pacific, and a virtual transplant from the Pacific to Lake Michigan ................. 123

Figure 4.4. The percent of lipid (per gram body weight) of Chinook salmon modeled in
the Pacific, Lake Michigan, and a virtual transplant from the Pacific to Lake
Michigan. ................................................................................................................ 124

xii

INTRODUCTION

In the late 19503, the food web of Lake Michigan was severely disrupted due to a
combination of over-fishing, pollution, and the arrival of non-indigenous species. The
abundance of native lake trout, which previously filled the role of top predator declined
to low levels. Commercial fisheries for lake trout were closed, but this species still
became extirpated from the lake. With few top predators in the system, prey species,
especially the non-native alewife (Alosa pseudoharengus), increased substantially.
During winter, when food was scarce for the prey species, huge numbers of fish starved.
When the lake thawed in the spring, millions of kilograms of dead alewife washed up on
the shores of Lake Michigan, clogged water intake pipes, and deterred tourism. In order
to control alewife populations and to create a sport fishery, Chinook salmon
(Oncorhynchus tshawytscha) were introduced in Lake Michigan in the 19608 through
large-scale stocking efforts (Kocik and Jones, 1999). Results were impressive; the
alewife population was controlled, and the new sport fishery lured anglers and tourists to
Lake Michigan. During the 19705 and early 19803, the numbers of Chinook salmon
stocked in Lake Michigan approached 8 million. However, after two decades of steady
population growth, Chinook salmon in Lake Michigan suffered a massive mortality event
in the late 19805 and harvest decreased by 90% (Benjamin and Bence 2003). This
mortality event was associated with low growth rates and a disease outbreak, both of
which are thought to have been exacerbated by severe nutritional stress in the population
(Holey et al. 1998; Wesley 1996; Peeters and Royseck 1997). The outbreak increased
concern about the nutritional condition of Chinook salmon in Lake Michigan and their

1

ability to maintain adequate energy reserves to survive the conditions in their novel
environment. In Lake Michigan, Chinook salmon are exposed to water temperatures that
are both colder in the winter and warmer in the summer than in their native Pacific range
(Welch pers. comm, Bergstedt pers. comm.) It is also thought that Lake Michigan
Chinook salmon have very little energy intake in the winter compared to their Pacific
counterparts. The history of Lake Michigan Chinook salmon and the devastation that
followed the collapse in the 19803 led to the following objectives being addressed
regarding energy dynamics of Lake Michigan Chinook salmon in this dissertation:
Objective One

We used life history theory to make predictions about how Chinook salmon might
alter their life history strategies to maximize fitness in different environments. This
involved analyzing populations of Chinook salmon in native and exotic environments to
compare how the relationship between growth rate and age at maturation vary both
between populations and within a population from year to year. Populations examined
for this review included those from Alaska, Washington, Oregon, New Zealand, Lake
Oahe in South Dakota, and Lakes Superior, Huron, Michigan, and Ontario. The between
population analysis showed that when size at maturity increased (a surrogate for growth
rates), age at maturation decreased. For populations with multiple years of data to
evaluate, a clear trend was difficult to determine.
Objective Two

We evaluated the effects of maturity, size, season, location and year on energy

content, and specifically lipid levels, in Lake Michigan Chinook salmon. The prediction

that the long, cold winter would result in much lower energy content in the spring
compared to the fall and that this low energy content could lead to nutritional stress and
the possibility of a second outbreak of bacterial kidney disease or a different
opportunistic disease was tested. Results from this study supported the hypothesis that
energy content was lower in the spring compared to the fall, likely due to over-winter
stress, but only for the small size class. The large size class showed the opposite pattern,
with energy content being much higher in the spring than in the fall. These findings led
to the suggestion that the maturation process, which occurs in the large fish during the
summer, depletes the fish of their energy reserves.
Objective Three

We developed an efficient and inexpensive approach to monitoring the nutritional
health of the Lake Michigan Chinook salmon population in order to provide reliable
evidence of when whole-fish lipids are reaching low enough levels to be of concern. For
this objective, we investigated what variable to measure, what segment of the population
should be measured, when and where samples should be collected, and what values
would be interpreted as signs of severe nutritional stress. Our proposed monitoring
program advocates using water content in a muscle plug as a predictor of total lipids in
the whole fish. This method will allow fishery managers to perform a quick and
inexpensive assessment of nutritional condition in a population and avoid future periods

of severe nutritional stress.

Objective Four

We determined the energy allocation strategy and associated life-history schedule
that would maximize fitness for Chinook salmon living in the Pacific Ocean and for
Chinook salmon living in Lake Michigan and we compared these two strategies. My
approach to this objective was to construct a bioenergetics and energy allocation
computer model that predicted the number of eggs produced by a Chinook sahnon given
the environment it was exposed to and the allocation schedule. Parameters that were
different between the two environments included monthly temperature, monthly
proportion of maximum consumption, prey density, and natural mortality rates. The
metabolic rate was appropriately modified to reflect the salinity conditions of each
environment. Once the optimal energy allocation was determined for each environment,
the optimal energy allocation strategy from the Pacific was input into the Lake Michigan
model to see how fitness was affected when this suboptimal strategy was used. This was
of special interest because actual strategies used by Chinook salmon in lake Michigan
could reflect their evolutionary history in the Pacific Ocean. This modeling approach
also allowed us to explore which parameters had the largest effect on fitness for Chinook

salmon.

References

Benjamin, D.M. and Bence, J .R. 2003. Spatial and temporal changes in the
Lake Michigan Chinook Salmon fishery, 1985-1996. Michigan Department
of Natural Resources, Fisheries Division, Fisheries Research Report 2065.

Holey, M. E., Elliot, R.F., Marcquenski, S.V., Hnath, J .G., and Smith, K. D. 1998.
Chinook salmon epizootics in Lake Michigan: possible contributing factors and
management implications. J. Aquatic Animal Health 10: 201-210

Kocik, J .F ., Jones, ML. 1999. Pacific salmonines in the Great Lakes basin. In
Great Lakes fisheries policy and management: a binational perspective, eds.
W.W. Taylor and CR Ferreri, pp. 455-488. East Lansing, MI: Michigan State
University Press.

Peeters, P. and Royseck, K. 2003. Harvest, Age, and Size at Age of Chinook and Coho
Salmon at Strawberry Creek Weir and Besadny Anadrarnous Fisheries Facility Fall 2003.
Wisconsin Department of Natural Resources Technical Report.

Wesley, J .K. 1996. Age and growth of Chinook salmon in Lake Michigan. Michigan
Department of Natural Resources, Fisheries Division, Fisheries Research Report 2029,
Lansing, Michigan.

CHAPTER 1
WITHIN AND AMONG POULATION RELATIONSHIPS BETWEEN GROWTH
RATE AND AGE AT MATURITY OF CHINOOK SALMON (ONCHORHYNCHUS
T SHA WYTSCHA) IN NATIVE AND INTRODUCED POPULATIONS
Introduction
A major life history question for any organism is its timing of maturation. The
optimal size and age at maturation is predicted to be a balance between the costs and
benefits of delaying sexual maturity (Roff 1984; Stearns and Crandall 1984; Stearns and
Koella, 1986). In organisms with a positive relationship between size and fecundity,
maturation at a small size, or young age, often means low fecundity and therefore low
fitness. Maturation at a large size, or older age, allows for more and/or larger propagules,
but could also mean an increased risk of mortality before the large size is attained and
reproduction occurs. When environmental conditions exist that allow for rapid growth,
such as abundant prey, life history theory predicts that age at maturity should decrease
since an organism would have ample energy for allocation to both growth and
reproduction (Roff, 1984; Stearns 1992). When environmental conditions do not provide
ample energy to allow for both rapid growth and reproduction, life history theory predicts
that organisms should delay reproduction to either (1) allow for attainment of a larger
body size or (2) wait for a period of better grth conditions and compensate for the
period of poor growth conditions (Roff 1991; Nicieza and Metcale 1997; Grover, 2005).
However, if environmental conditions result in extremely low growth rates, long delays

in maturation are predicted to be unlikely because the associated risks such as nutritional

stress leading to mortality, could exceed the benefits (Stems and Koella 1986; Stearns,

1992). Under such conditions it would be predicted that maturation should occur at an

early age. An early maturity would result in a small size at maturity and low fecundity,
but that strategy is more optimal than delaying maturity and risking mortality.

Due to their indeterminate growth and their often highly plastic life histories, fish
have been the focus of many studies examining the effects of growth rate on age at
maturity. Age at maturity was found to be significantly lower in populations with high
growth rates for pumpkinseed sunfish, Lepomis gibbosus (Villeneuve et al. 2005), brook
trout, Salvelinusfontinalis (Hutchings 1993; Hutchings 1996), Atlantic salmon (Salmo
salar) (Hutchings and Jones 1998); and churn salmon (Oncorhynchus keta) (Morita and
Morita 2002). In central Norway, growth rates were found to affect age at maturity on
an individual basis for grayling, Thymallus thymallus (Haugen 2000).

Similarly, there has been evidence for a shift to an older age at maturity when
growth rate decreases. Grover (2005) found that during a period of declining grth
conditions in Bucks Lake, California, the average age of maturity in a population of
kokanee salmon (Oncorhynchus nerka) shifted from age 2 to age 3. In such instances, by
delaying age at maturation, size at maturity does not decrease despite the slower growth
rate and there is no accompanying reduction in fecundity (McKinnell 1995; Morita et al.
2005).

A study which looked at Eurasian perch, Percafluviatz'lis L., in five Swedish lakes
found that the populations with the earliest maturing females were those in the lake with

the fastest growth and the lake with the slowest growth (Heibo and Magnhagen 2005).

The authors concluded that for females in the lake with the slowest growth, the early
maturation could be a strategy to minimize generation time, thereby maximizing fitness
(Heibo and Magnhagen 2005; Taborsky et al. 2003). F uthermore, delaying maturation
with such slow growth rates would not increase either size or fecundity of the individuals
(ibid).

How would the three different growth situations described in the studies described
above (rapid growth, average growth, and slow growth) be explained by life-history
theory? Stearns and Koella (1986) modeled the relationship between size and optimal
age at maturity for different growth rates (see Stearns and Koella 1986; Stearns 1992).
Their model also suggested that changes in juvenile mortality should affect how growth
rates affect the optimal age at maturity. Size-dependent mortality is often critically
important for fish (Winemiller and Rose 1992), with higher and more variable mortality
rates being associated with poor growth. In an environment with generally low juvenile
growth rates, because of an expectation of elevated mortality rates when growth is
especially poor, a strong negative relationship is expected between growth rate and
juvenile mortality. However, at higher growth rates variations in growth are less likely to
affect juvenile mortality (Fig. 1.1). Using the model developed by Stearns and Koella
(1986) this leads to the prediction of a dome-shaped relationship between growth rate and
age at maturity (Fig. 1.2). During periods of high growth and thus a weak relationship
between growth and juvenile mortality, age at maturity is predicted to decrease from a
maximum value. Under these conditions, fish are growing quickly with low mortality

and would be able to mature at a large size, but a young age. Conversely when growth

rates are low and juvenile mortality becomes much higher, age at maturity is again
predicted to decrease from the maximum value because the risk of delaying maturity is
too great.

The goal of this paper is test the predictions of the dome-shaped relationship by
evaluating how age at maturity covaries with two surrogates for growth rate, size at
maturity and length at size at age from a wide range of Chinook salmon (Onchorhynchus
tsawytscha) populations. The observed mean age at maturity for a population is a result
of both the timing of maturation for individual members of that population and survival
rates prior to maturity. Because survival can affect mean age at maturity, we will also
examine the relationship between survival rate and age at maturity for one population of
Chinook salmon.

Chinook salmon are an ideal species for such an investigation because of the large
amount of life history variation they show, their wide-ranging native distribution, and
their introduction into a number of novel environments. Also, it has been established that
Chinook salmon fecundity increases with size, so the trade-off between maturing at a
smaller size versus risking another year of mortality to increase in size does exist for this
species (Healey and Heard 1984). Furthermore, due to their economic importance, they
are a well-studied species (Groot and Margolis 1998).

Chinook salmon are anadromous and semelparous. Most Chinook salmon spawn
in coarse gravel substrate of freshwater streams during the fall. The fry emerge during
the spring. Chinook salmon have been separated into two races: stream-type and ocean-

type. In stream-type Chinook salmon, the fry spend up to two winters in freshwater

before migrating to the ocean (Bigler et al. 1995). After the ocean phase, they return to
their natal river in the spring or summer, several months prior to spawning. This type of
life history is mostly found in Asian populations and populations north of 56° in North
America (Healey 1991). In ocean-type Chinook salmon, the young-of-year migrate to the
ocean by their first winter. Once in the ocean, these Chinook salmon spend between 2
and 6 years in the ocean before returning to their natal streams in the late summer or fall,
just a few days or a week before spawning (Taylor 1991). The ocean-type Chinook
salmon are most commonly found south of 56° in North America (Healey 1991). While
it is clear that some of this variation in life history is controlled genetically (Healey and
Heard 1984; Hankin et al. 1993; Hutchings 2003), environmental influences are also
important in shaping life history traits of Chinook salmon populations (Quinn 2005). In
their native distribution, there is a pattern of increasing age at maturity among
populations with increasing latitude (Myers et al. 1998). This pattern has been seen for
other species of fishes as well, including dace, Leuciscus leuciscus (Labon-Cervia et a1.
1996), largemouth bass, Micropterus salmoides, (Garvey et a1. 2003), and Eurasian
perch, Percafluviatilis (Heibo et al. 2005).

Life history variation among Chinook salmon can be thought of as a consequence
of their extensive native geographical distribution which includes a broad range of
environmental conditions (Groot and Margolis 1998). In North America, their range
extends from the San Francisco Bay in California northwards along the coastal areas of
Oregon, Washington, British Columbia and Alaska and in Asia they are found from

northern Hokkaido in Japan to the Anadyr River in Russia (Healey 1991). Well over a

10

thousand spawning populations of Chinook salmon exist within this range and can
include fewer than a thousand to over a million spawners each year (Healey 1991).
Among these populations, spawning can occur near tidewaters to over 3200 kilometers
upstream (ibid). Together, the diverse environmental conditions which are found in the
natural range of Chinook salmon and the wide variation in life history parameters,
including age of maturity, provide an ideal opportunity to look for trade-offs between
growth rates and age at maturity. Furthermore, since late nineteenth century, Chinook
salmon have been introduced into ecosystems around the world with varying degrees of
success (Lever 1996). Life history changes that have occurred following introduction
into novel environments provide further data to test our hypotheses about the
relationships between growth rates and age/size at maturity.

In this paper, we will use data obtained from both Chinook salmon populations in
their native range and those in introduced environments to explore two predictions:
(1) There will be a decrease in age at maturity when growth rates are high; and (2)
There will also be a decrease in age at maturity when growth rates are low.
The first prediction would be likely to result from an environment in which a fish could
obtain a large size and high fecundity at an early age and thus would be predicted to
minimize mortality risks by maturing at a young age. The second situation would be
predicted when food availability is so low that the mortality risk of feeding for another

year to attain large size and high fecundity is too great to warrant delaying maturity.

ll

Methods

Data were compiled from 16 populations of Chinook salmon, six from their native
range and ten from introduced populations (Table 1). All data were collected from
mature fish that were returning to spawn. We only used data from fall run fish with
brood year sample sizes of over 40 individuals. Age determination methods varied
among studies, but included aging of scales and otoliths, and recovery of coded wire tags
(Table 1). In all cases, age was assumed to be correct. Male salmon returning to spawn
at age 1, commonly known as “jacks”, were excluded from the analysis as they represent
a different life history strategy (Hooff et al. 1999a). Only four of the studies reported
male and female data separately, therefore data were pooled for male and female samples
30 that comparisons could be made with studies that did not report gender.

The limitations of the data resulted in growth rate being approximated in two
different ways for this study, mean length at maturity and mean length at age 3. Using
mean length at maturity as a surrogate for growth assumes that when growth rates are
high, length at maturity will increase. Mean length at maturity was considered a suitable
indicator of growth rate only when the relationship between mean age at maturity and
mean length at maturity was negative. In these cases, the large size and young age of
maturity clearly shows that age at maturity is decreasing as growth increases.
Furthermore, for some of the populations analyzed, maturation was rare among age 3
individuals and sample sizes for mean length at age 3 would have been quite low. When
the relationship between mean age at maturity and mean length at maturity was not

significantly negative, mean length at age 3 was used as a surrogate for growth rate.

12

For the spatial analysis, mean length at maturity and age at maturity were
determined for each of the 16 populations as a weighted average of the number of fish
collected. Mean length at maturity was used as an indicator of growth rate. For the
populations with multiple years of data, we calculated a weighted average over the years
for which there were estimates, with each estimate weighted by its associated sample
size.

The temporal analysis involved looking at changes in age at maturity for a single
population for different years. Six of the 16 populations had sufficient data to construct a
time series of at least six brood years. For these populations, mean age at maturity was
again calculated as a weighted average of the number of fish collected of each age. In
these cases, we used mean length of the fish collected at age 3 as an indicator of growth
rate.

For the population from Lake Michigan Strawberry Creek Weir, yearly data were
available on the percentage of stocked fish that retum‘to the counting weir. These data
were used as a surrogate for mortality to look at the relationship between mortality and

age at maturity.

Results

Mean length at maturation among the 16 populations ranged from 666 mm to 906
mm and the mean age of maturity ranged from 2.39 years to 4.07 years (Fig. 1.4). A
negative correlation was found between mean age at mattu'ity and length at maturity

across the 16 populations (Fig. 1.4) (r = -0.85, p<0.001). The populations from South

13

Dakota and the three sites in Alaska matured at a substantially smaller mean length and at
a higher mean age compared to the other 12 populations (Fig. 1.4). Lakes Michigan and
Ontario had the greatest mean length of maturity and the lowest mean age at reproduction
of the populations (Fig. 1.4). When all the years of data are shown for populations in
which multiple brood years were available, the negative relationship is still evident (Fig.
1.5), but with considerably greater scatter (r = -0.5 7, n=83, p<0.001).

For the temporal analysis, only one of the populations analyzed showed a

significant positive correlation between mean age at maturity and mean length at age 3:
Lake Huron (r = 0.65, n=l6, p<0.006) (Fig. 1.6a). One other population, Glenariife
Stream, New Zealand showed a weak positive relationship (I = 0.34, n=17, p = 0.17).
For the other four populations, Lake Michigan, Lake Superior, Bonneville Dam Ocean
Type and Bonneville Dam Stream Type, the relationship between mean age at maturity
and mean length at age 3 was negative and non-significant (Table 2). The range in length
at age 3 was highest for Lake Michigan and the New Zealand population and lowest for
the Bonneville Dam populations (Table 2).

Finally, age at maturity was not significantly related to smolt-to-adult survival

rates for the Lake Michigan population (r=0.012, n=18, p=0.64) (Fig. 1.7).

Discussion
Our first prediction, that a decrease in age at maturity would occur when growth
rates were high, was supported by the data from the spatial analysis. In the populations

with lowest size at maturity, assumed to be an indicator of slow growth, like Alaska and

14

South Dakota, age at maturity was highest. In Lakes Ontario and Michigan, individuals
reached a larger size in a shorter time and age at maturity decreased. One explanation for
the strong negative relationship between mean age at maturity and mean length at
maturity among the 16 populations is a reflection of the adaptive responses to differences
among the growth conditions for these populations. Stearns (1992) argues that two main
reasons to delay maturity are (a) if delaying maturity permits an increase in fecundity
which outweighs the fitness lost through longer generation time or lower survival (b) if
delaying maturity increases the quality of the offspring produced enough that the
reduction in the instantaneous juvenile mortality rate of the offspring offsets the fitness
lost through longer generation time or lower survival. In populations with slow growth
rates, the postponement of maturation may be necessary to allow individuals to attain a
size that is large enough to make a sufficient quantity and quality of gametes, thereby
maximizing fitness. Indeed, the data used in this analysis showed a positive relationship
between age and length for all populations. Populations from locations with slow growth
rates could possibly be postponing their maturation for these reasons, creating the pattern
seen in the spatial analysis.

The data from the Lake Huron population were the only intra-population data that
supported the second prediction that age at maturity would decrease when growth rates
decreased. The population from New Zealand showed a positive, but non-significant
relationship between age at maturity and length at age 3. The Stearns and Koella (1986)
model would predict this to happen when mortality rates are very sensitive to changes in

growth. If food availability decreases and growth rates slow down, one would predict

15

that individuals may maximize fitness by maturing early because delaying maturation
would result in another year of vulnerability to mortality without a compensatory increase
in size. In some instances, like the 1999 and 2000 brood years in Lake Huron, lengths at
age 3 were very low and age at maturation was earlier than in other years, lending some
support to this prediction. A decrease in age at maturity during slow growth was also
observed for perch, Percafluviatilis L. (Heibo and Magnhagen 2005), brook trout
(Hutchings 1993) pumpkinseed sunfish (Fox 1994) and white-spotted charr (Morita and
Morita 2002).

The other intra-population results did not strongly support either of the
predictions; the direction of the relationship between age at maturity and length at age 3
was variable. For three of these populations, Bonneville Ocean Type, Bonneville Stream
Type and Lake Superior, perhaps the lack of any significant pattern was a result of the
small number of brood years or the limited contrast in mean length at age 3 for these
populations. The two populations with the largest range in values for length at age 3
(Lake Huron, New Zealand) were also the ones which showed a positive relationship
between age at maturity and length at age 3. Perhaps the positive relationship between
age at maturity and growth rates was only seen in populations which have undergone
large changes in growth rates.

Mortality rates have been found to affect age and size at maturity in some studies
(Hutchings 1993; Haugen 2000), but not in others (Cardinale and Modin 1999; Heibo et
al. 2005). For the Lake Michigan population, there was no relationship evident between

age at maturity and mortality rate, measured by percentage of stocked individuals

16

returning to the weir. While the data for this relationship were only available for one
population, it supports the hypothesis that growth rate is the major factor affecting age at
maturity.

By comparing Chinook salmon across these 16 populations, it is clear that latitude
plays an important role in shaping both the age at maturity and the length at maturation.
As expected from previous work on Chinook salmon (Myers 1998), the data in this
analysis showed an increase in mean age at maturity as latitude increased. This trend is
usually explained by a combination of factors, such as temperature, food availability, and
predation risk (Garvey et al. 2003; Heibo et al. 2005; Quinn 2005).

In our assessment of 16 Chinook populations, we did not see the relationship
hypothesized in figure 1.2, either within or among populations. Possibly, there were not
sufficient data for any one population that had experienced periods of both rapid and slow
growth or periods during which the strength of the relationship between juvenile
mortality rate and growth rates varied. The Lake Huron population clearly shows a
decrease in age at maturity during a recent period of slow growth, and during the early
19803, when growth was extremely high, age at maturity was also low (pers. comm. J .R.
Bence). Perhaps if more data are collected as changes continue in the populations, we
will see a complete dome-shaped curve for a single population.

There are many ways to measure growth rates. Use of von Bertalanffy’s growth
function is one of the best models for looking at growth patterns over the life of a fish
(He and Stewart 2001). For this study, there were not enough data to utilize such a model

because most of the populations only have size data for spawning fish in the fall of each

17

year for two or three age classes. He and Stewart (2001) found that ages at first
reproduction among a wide range of fish species were highly correlated to a suite of
variables, such as growth coefficient, size at first reproduction, and maximum size.
Therefore, a combination of variables would be more informative for determining age at
maturity compared to only using growth rates. Future data collection for Chinook salmon
could focus on obtaining information which would provide better surrogates for grth
rate than were available for this study. Additional data could also aid in addressing any
significant gender differences for the relationship between growth and age at maturation.
For most populations of Chinook salmon, the females mature at a larger size and older
age than males (Healey 1991). If enough data were available to look at the trends in
growth rate and age at maturity for males and females separately, other interesting
patterns may emerge.

Better understanding of the relationship between growth rate and age at maturity
will also help managers make better decisions about stocking rates and fishing target
rates. The Lake Huron population had the greatest range for mean age at maturity, from
2.5 years to 3.6 years. Such a wide ranging mean age at maturity presents challenges for
agencies trying to manage a population, to forecast population fluctuations, or to maintain
equilibrium. Another variable of interest for management would be the slope of the
relationship between age at maturity and growth rates. The slope of this relationship
differed between the populations in our study. In populations with a flatter slope, small
changes in growth conditions would not substantially affect age at maturation, but in

populations with steeper slopes, a small change in growth rates could drastically affect

18

age at maturation and would have implications on the management of the population.
Finally, populations which showed a negative relationship for size at maturity and age at
maturity could be those for which juvenile mortality rate increases strongly as growth
rates decrease. Such populations might be suffering from density dependent mortality
from lack of forage or by some other mechanism. Increasing our understanding of the
mechanisms responsible for age at maturation together with knowledge of how stocking

densities affect growth rates can help managers better define optimal stocking strategies.

19

References

Barnes, M.E., Hanten, R.P., Lott, J .P., and M. Gabel. (2001). Environmental Influences
on Landlocked Fall Chinook Salmon Reproductive Characteristics. North American
Journal of Aquaculture 63, 58-65.

Barnes, M.E., Hanten, R.P., Codes, R.J., Sayler, W.A., and J. Carreriro. (2000).
Reproductive Performance of Inland Fall Chinook Salmon North American Journal of
Aquaculture 62, 203-211.

Bigler, B.S., Welch, D.W., and J .H. Helle. (1995). A review of size trends among North
Pacific salmon (Oncorynchus spp.). Canadian Journal of Fisheries and Aquatic Sciences
53, 455-465.

Cardinale, M. and J. Modin. (1999). Changes in size-at-maturity of Baltic cod (Gadus
morhua) during a period of large variations in stock size and environmental conditions.
Fisheries Research 41, 285-295.

Fox, M.G. (1994). Growth, density, and interspecific influences on pumpkinseed sunfish
life histories. Ecology 75, 1157-1171.

Garvey, J .E., D.R. Devries, R.A. Wright and J .G. Miner. (2003). Energetic adaptations
along a broad latitudinal gradient: implications for widely distributed assemblages.
BioScience 53, 141-150.

Gates, KS. and KC. Harper. (2002). Run timing and abundance of adult Pacific salmon
in the Tuluksak River, Yukon Delta National Wildlife Refuge, Alaska, 2001. US. Fish
and Wildlife Service, Alaska Fisheries Data Series Number 2002-6, Kenai, Alaska.

Grover, MC. (2005). Changes in size and age at maturity in a population of kokanee,
Oncorhynchus nerka, during a period of declining growth conditions. Journal of Fish
Biology 66, 122-134.

Hankin, D. G. N. J. W. a. D. T. W. (1993). Evidence for Inheritance of Age of Maturity
in Chinook Salmon (Oncorhynchus tshawytscha). Canadian Journal of Fisheries and
Aquatic Sciences 50, 347-358.

Harper, KC, and Watry, CB. (2001). Abundance and Run Timing of Adult Salmon in
the Kwethluk River, Yukon Delta National Wildlife Refuge, Alaska, 2000. Alaska
Fisheries Data Series Number 2001-4. US. Fish and Wildlife Service, Kenai, Alaska.

Haugen, TO. (2000). Growth and survival effects on maturation patterns in populations

20

of grayling with recent common ancestors. Oikos 90, 107-118.

He, J .X., and Stewart DJ. (2001). Age and Size at First Reproduction of Fishes:
Predictive Models based Only on Growth Trajectories. Ecology 82, 784-791.

Healey, M. C. (1991). Life History of Chinook Salmon (Oncorhynchus tsawytscha). In
Pacific Salmon Life Histories, ed. C. M. L. Groot, Vancouver: UBC Press.

Healey, MC. and W.R. Heard. (1984). Inter- and intra- population variation in the
fecundity of chinook salmon and its relevance to life history theory. Canadian Journal of
Fisheries and Aquatic Sciences 41, 476-483.

Heibo, E and C. Magnhagen. (2005). Variation in age and size at maturity in perch,

Percafluviatilis L., compared across lakes with different predation risk. Ecology of
Freshwater Fishes 14, 344-351.

Heibo, E., Magnhagen, C., and LA. Vollestad. (2005). Latitudinal Variation in Life-
History Traits in Eurasian Perch. Ecology 86, 3377-3386.

Hooff, R.C., Fryer, J. and Netto, J. (1999a). Age and Length Composition of Columbia
Basin Chinook, Sockeye, and Coho Salmon at Bonneville Dam in 1998. Columbia River
Inter-Tribal Fish Commission Technical Report 99-3. Portland, Oregon.

Hooff, R.C., Ritchie, A., Fryer, J. and Netto, J. (1999b). Age and Length Composition of

Columbia Basin Chinook, Sockeye, and Coho Salmon at Bonneville Dam in 1999.
Columbia River Inter-Tribal Fish Commission Technical Report 99-4. Portland, Oregon.

Hutchings, J. A. (2003). Norms of Reaction and Phenotypic Plasticity in Salmonid Life
Histories. In Evolution Illuminated: Salmon and Their Relatives, eds. A. P. Hendry and
S. C. Stems, pp. 154-174. Oxford: Oxford University Press.

Hutchings, J .A. and M.E.B. Jones. (1998). Life history variation and growth rate

thresholds for maturity in Atlantic salmon, Salmo salar. Canadian Journal of Fisheries
and Aquatic Sciences SSS, 22-47.

Hutchings, J .A. (1996). Adaptive phenotypic plasticity in brook trout, Salvelinus
fontinalis, life histories. Ecoscience 3, 25-32.

Hutchings, J .A. (1993). Adaptive life history effected by age-specific survival and
growth. Ecology 74, 673-684.

Kelsey, DA. and J. Fryer. (2003). Age and Length Composition of Columbia Basin
Chinook, Sockeye, and Coho Salmon at Bonneville Dam in 2001. Columbia River Inter-
Tribal Fish Commission Technical Report 03-1. Portland, Oregon.

21

Kelsey, DA. and J. Fryer. (2001). Age and Length Composition of Columbia Basin
Chinook, Sockeye, and Coho Salmon at Bonneville Dam in 2000. Columbia River Inter—
Tribal Fish Commission Technical Report 01-1. Portland, Oregon.

Lever, C. (1996). Naturalized Fishes of the World. San Diego, Academic Press.

Lobén-Cervia, Y. Dgebuadze, C. G. Utrilla, P. A. Rincon, C. Granado-Lorencio. (1996).
The reproductive tactics of dace in central Siberia: evidence for temperature regulation of
the spatio-temporal variability of its life history. Journal of Fish Biology 48, 1074 -1087.

McKinnell, S. (1995). Age-specific effects of sockeye abundance on adult body size of
selected British Columbia sockeye stocks. Canadian Journal of Fisheries and Aquatic
Sciences 52, 1050-1063.

Miranda, D.P., Whiteaker, J. and J. Fryer. (2004). Age and Length Composition of
Columbia Basin Chinook, Sockeye, and Coho Salmon at Bonneville Dam in 2003.
Columbia River Inter-Tribal Fish Commission Technical Report 04-2. Portland, Oregon.

Morita, K. and SH. Morita. (2002). Rule of age and size at maturity: individual variation
in the maturation history of resident white-spotted char. Journal of Fish Biology 61,
1230-123 8.

Morita, K., Morita, S.H., Fukuwaka, M., and Matsuda, H. (2005). Rule of age and size at
maturity of chmn salmon (Oncorhynchus keta): implications of recent trends among
Oncorhynchus spp. Canadian Journal of Fisheries and Aquatic Sciences 62, 2752-2759.

Myers, J .M and 10 coauthors. (1998). Status review of Chinook salmon from
Washington, Idaho, Oregon, and California. NOAA Technical Memorandum NMFS-
NWFSC-35.

Nicieza, A. G., and N. B. Metcalfe. (1997). Growth compensation in juvenile Atlantic
salmon: responses to depressed temperature and food availability. Ecology 78, 23 85—
2400

Ontario Ministry of Natural Resources. (2005). Lake Ontario Fish Communities and
Fisheries: 2004 Annual Report of the Lake Ontario Management Unit. Ontario Ministry
of Natural Resources, Picton, Ontario, Canada.

Peeters, P. and Royseck, K. (2003). Harvest, Age, and Size at Age of Chinook and Coho
Salmon at Strawberry Creek Weir and Besadny Anadramous Fisheries Facility Fall 2003.
Wisconsin Department of Natural Resources Technical Report.

Quinn, TR (2005). The Behfavior and Ecology of Pacific Salmon and Trout. University
of Washington Press, Seattle.

22

Quinn, TR and M. Unwin. (1993). Variation in life history patterns among New
Zealand chinook salmon, Oncorhynchus tshawytscha, populations. Canadian Journal of
Fisheries and Aquatic Sciences 50, 1414-1421.

Quinn, T.P., Vollestad, L.A., Peterson, J. and V. Gallucci. (2004). Influences of
Freshwater and Marine Growth on the Egg Size—Egg Number Tradeoff in Coho and
Chinook Salmon. Transactions of the American Fisheries Society 133, 55-65.

Roff, D.A. (1991). The evolution of life history variation in fishes, with particular
references to flatfishes. Netherlands Journal of Sea Research 27, 316-323.

Roff , D.A. (1984). The evolution of life history parameters in teleosts. Canadian
Journal of Fisheries and Aquatic Sciences 41, 989-1000.

Roettiger, T.G., Harris, F ., and KC. Harper. (2005). Abundance and Run Timing of

Adult Salmon in the Kwethluk River, Yukon Delta National Wildlife Refuge, Alaska,
2004. Alaska Fisheries Data Series Number 2005-7. US. Fish and Wildlife Service,
Kenai, Alaska.

Roettiger, T.G., Harris, F., and KC. Harper. (2004). Abundance and Run Timing of

Adult Salmon in the Kwethluk River, Yukon Delta National Wildlife Refuge, Alaska,
2003. Alaska Fisheries Data Series Number 2004-8. US. Fish and Wildlife Service,
Kenai, Alaska.

Roettiger, T.G., Harper, KC, and Nolfi, D. (2003). Abundance and Run Timing of
Adult Salmon in the Kwethluk River, Yukon Delta National Wildlife Refuge, Alaska,
2002. Alaska Fisheries Data Series Number 2003-6. US. Fish and Wildlife Service,
Kenai, Alaska.

Roettiger, T.G., Harper, KC, and Chikowski, A. (2002). Abundance and Run Timing of
Adult Salmon in the Kwethluk River, Yukon Delta National Wildlife Refuge, Alaska,
2002. Alaska Fisheries Data Series Number 2002-8. US. Fish and Wildlife Service,
Kenai, Alaska.

Stearns, SC. (1992). The Evolution of Life Histories. Oxford : Oxford University Press.

Stearns, SC. and RE. Crandall. (1984). Plasticity for age and size at sexual maturity: a
life-history response to unavoidable stress. In Fish Reproduction: Strategies and Tactics,
G.W. Potts and R.J. Wooton, eds. Pp. 13-33. London: Academic Press.

Stearns, SC. and J .C. Koella. (1986). The evolution of phenotypic plasticity in life-
history traits: predictions of reaction norms for age and size at maturity. Evolution 40,
893-913.

Taborsky, B., Dieckmann, U., and M. Heino. (2003). Unexpected discontinuities in life-

23

history evolution under size-dependent mortality. Proceedings of the Royal Society of
London Series B—Biological Sciences. 270, 713-721.

Taylor, E. B. (1991). A review of local adaptation in Salmonidae, with particular
reference to Pacific and Atlantic salmon. Aquaculture 98, 185-207.

Tobin, J. H. 111 and Harper, KC. (1999). Abundance and Run Timing of Adult Salmon
in the East Fork Andreafl<sy River, Yukon Delta National Wildlife Refuge, Alaska, 1998.
Alaska Fisheries Data Series Number 99-3. US. Fish and Wildlife Service. Kenai,
Alaska

Tobin, J. H. 111 and Harper, KC. (1998). Abundance and Run Timing of Adult Salmon
in the East Fork Andreafl<sy River, Yukon Delta National Wildlife Refuge, Alaska, 1997.
Alaska Fisheries Data Series Number 98-2. US. Fish and Wildlife Service. Kenai,
Alaska

Villeneuve, F., G.H. Copp, M.G. Fox, and S. Stakenas. (2005). Interpopulation variation
in growth and life-history traits of introduced sunfish, pumpkinseed Lepomis gibbosus, in
Southern England. Journal of Applied Ichthyology 21, 275-281.

Winemiller, K.O. and K.A. Rose. 1992. Patterns of Life-History Diversification in
North-American Fishes—Implications for Population Regulation. Canadian Journal of
Fisheries and Aquatic Sciences 49, 2196-2218.

Zabkar, L.M., F. Harris, and K. Harper. (2005). Abundance and Run timing of Adult
Pacific salmon in the Tuluksak River, Yukon Delta Wildlife Refuge, Alaska, 2004.
Alain Fisheries Data Series Report Number 2005-6. US. Fish and Wildlife Service
National Wildlife. Kenai, Alaska.

Zabkar, L.M. and KC. Harper. (2004). Abundance and run timing of adult Pacific
salmon in the Tuluksak River, Yukon Delta National Wildlife Refuge, Alaska, 2003. US.
Fish and Wildlife Service, Alaska Fisheries Data Series Report Number 2004-6, Kenai,
Alaska.

Zabkar, L.M. and KC. Harper. (2003). Abundance and Run Timing of Adult Pacific
Salmon in the East Fork Andreafsky River, Yukon Delta National Wildlife Refuge,
Alaska, 2001 and 2002. Alaska Fisheries Data Series Number 2003-5. US. Fish and
Wildlife Service. Kenai, Alaska.

24

Location
Alaska

Washington
Oregon

Great Lakes

8. Dakota
New Zealand

Andreafsky River (AKA)1

Kwethluk River (AKK)2

Tuluksak River (AKT)3

University of Washington Hatchery (WAS)4
Bonneville Dam (ORS)5

Bonneville Dam (0R0)5

Strawbeny Creek, Lake Michigan (MCH)6
Salmon River, Lake Ontario (ONT)7

Lake Huron (HUR)8

French River, Lake Superior (SUP)9

Lake Oahe (sok)10

GIenan'ife(NZG)11

Hydra (NZH)12

Rakaia (NZR)12

Rangitata (NZN)12

Waitaki (NZW)12

Yearsa

Sampled R_acs'°

1997-01
5 years
2002-04
2 years
1998—03
1998-03
1985-04
1999-05
1994-04
1991-00
1 year
1976-85
6 years
7 years
12 years
5 years

1 Zabkar and Harper, 2003, Tobin and Harper, 1998, Tobin and Harper, 1999
2 Roettiger et al. 2005, Roettiger et al. 2004, Roettiger et al. 2003, Roettiger et al. 2002, Harper and
Watry 2001, Tobin and Harper 1998
3 Zabkar et al., 2005, Zabkar and Harper 2004, Gates and Harper 2002

4 Quinn et al. 2004

S

S
S
O
S
O
O
O
O
O
O
M
M
M
M
M

Aging
Method

Scales
Scales
Scales
CWT
Scales
Scales
CWT
Otoliths
Otoliths
Scales
CWT
Scales
Scales
Scales
Scales
Scales

5 Miranda et al., 2004, Kelsey and Fryer, 2003, Kelsey and Fryer, 2001, Hooff et al., 1999a andl999b
6 Peeters and Royseck, 2003
7 Ontario Ministry of Natural Resources 2005, Dan Bishop pers comm.
8 J. Johnson pers. comm.

9 D. Schreiner and M. Negus pers. comm.
10 Barnes et al. 2000, Barnes et al., 2001
11 Unwin pers. comm.

12 Quinn and Unwin, 1993

a A range of years are listed when data were available by year, when data were pooled over many years,
the number of years is listed.
b Ocean type (0), Stream type (S), Mix of Ocean and Stream type (M)

Table 1.1. Location, years sampled, race and aging method for 16 populations of

Chinook salmon.

25

Population r value p value Sample Size Range of length at

age 3 (mm)

Lake Huron 0.65 0.006 16 783-916
New Zealand Glenriffe 0.34 0.176 17 720-807
Lake Superior -0.10 0.834 7 883-975
Lake Michigan 021 0.427 17 697-811
Bonneville Dam -0.39 0.447 6 808-845
(Ocean Type)

Bonneville Dam -0.45 0.366 6 747-776
(Stream Type)

Table 1.2. Relationship between mean age at maturity and mean length at age 3 by

brood year for six populations of Chinook salmon.

26

strong

 

 

93-:
E
t’
o
E
2
'E
o
3 weak
_,
Growth Rate

Figure 1.1. Hypothesized nonlinear relationship between juvenile mortality and growth

rate, where mortality falls most rapidly with increasing growth rate when growth rate is

low.

27

Age at Maturi

 

 

Growth Rate

Figure 1.2. A hypothesized dome-shaped relationship between age at maturity and

growth rate, with age at maturity reaching a maximum at an intermediate growth rate.

28

 

 

 

New Zealand

 

 

 

Figure 1.3. Geographic location of 16 Chinook salmon populations. l-Andreafsky,
Alaska. 2-Tuluksak, Alaska. 3-Kwethluk, Alaska. 4-Washington. 5-Bonneville Dam
Ocean Type, Oregon. 6-Bonnville Dam Stream Type, Oregon. 7-Lake Oahe, South
Dakota. 8-Lake Superior. 9-Lake Michigan. 10-Lake Huron. 11-Lake Ontario. 12-
Glenariffe Stream, New Zealand. 13-Hydra Waters, New Zealand. 14-Rakaia River,

New Zealand. 15-Rangitata River, New Zealand. 16-Waitaki River, New Zealand.

29

:‘i‘
N

 

fl Tr

..AKT

 

 

70‘

a AKA SD.K AKK

(D o

33 3.8

3.. .WAS

5 3.4 - ‘SUP NZW

4...! O

m .NZH

(U 30 i NZG. o'Rs .

8) .ORO

<

E, 2.6 . JV'CH

é’ 9'"-
2.2

 

640 680 720 760 800 840 880 920
Mean Length at Maturity(mm)

Figure 1.4. The relationship between mean age of maturity and mean length at maturity,
a surrogate for growth rate, for 16 populations of Chinook Salmon (r = -0.85, p<0.0001).

A mean value for populations with multiple years of data is shown.

30

 

 

 

 

0 Lake Superior
,0? 4 0 _ OAKT A Glenariife. NZ,
L- ' OK I Lake Huron
g 0 ’ 'AKK 0 Lake Michigan,
> AKA 0 Lake Ontario
v 3 5 . WAS _
a ' ' -
-: I <>ORS °
3 NZH O .NZW' i
m 3.2 i an NZR.’ . 980' '-. 1
ORS
42.- ABESQ' AA ' I': . O
(O 8 AA . a. o o
(D 2.8 ' O A A ORS 0R0 .ORO ° .
2) .- ORO - ,
c ’ 9 ‘ O
(u 2.4 r ’. °
3’ ° 0
O
20 . I m ma 1 - I m m
650 700 750 800 850 900 950 1000

Mean Length at Maturity (mm)

Figure 1.5. The relationship between mean age of maturity and mean length at maturity,
a surrogate for growth rate, for 16 populations of Chinook Salmon with values for all
populations for all years (r =0.57, p<0.0001). AKA-Andreafsky, Alaska. AKT-
Tuluksak, Alaska. AKK—Kwethluk, Alaska. WAS-Washington. ORO-Bonneville Darn,
Ocean Type, Oregon. ORS-Bonnville Darn, Stream Type, Oregon. SDK-Lake Oahe,
South Dakota. NZG. NZH-Hydra Waters, New Zealand. NZR-Rakaia River, New

Zealand. NZN-Rangitata River, New Zealand. NZW-Waitaki River, New Zealand.

31

:5
o
i

i

 

o.)
oo
93
v
'33

.0"
0)

9°
.5

95 85 l

.9"
N
O

97
96 g4 .
92 o ' 98

9’
o
0

Mean Age at Maturity (Years)

9?, 00

.N'
a)

 

 

760 780 800 820 840 860 880 900 920
Mean Length at Age 3 (mm)

 

Figure 1.63. The relationship between mean age at maturity and mean length at age 3, a
surrogate for growth rate, for populations of Chinook Salmon by brood year for (a) Lake

Huron

32

3.25

 

 

 

 

3.20 - b) .66
g 81
g 3.10- .68 o .
9;" 3.05- 57
3 72 '77
m
2 3.00- ' .76 i
‘5
73 75
g 2.95- 69 - . .74
< ' .71 79
g 2.90 ’ . .80 '1
78
E 2.85» '
2.80~ .70
2.75 ~ . . e - e , . -
680 700 720 740 760 780 800 820
Mean Length at Age 3 (mm)

Figure 1.6b. The relationship between mean age at maturity and mean length at age 3, a

surrogate for growth rate, for populations of Chinook Salmon by brood year for (b)

Glenariffe Stream, New Zealand. Note: scales differ between populations.

33

 

.w s»
N «h
O

9’
(3

Mean Age at Maturity (years)

 

 

 

2.8 - ° , i
2.6 ~ '. . . .

2.41 - ', -

2.2- '

2'00 1 2 s A s 6

Return to Weir (percent)

Figure 1.7. The relationship between mean age at maturity and survival rate, measured as

percent return to weir, for Strawberry Creek Weir, Lake Michigan.

34

CHAPTER 2
FACTORS AFFECTING ENERGY DYNAMICS IN LAKE MICHIGAN, CHINOOK
SALMON, ONCORHYNCHUS TSHAWYTSCHA
Abstract
To better understand the energy dynamics of Chinook salmon (Oncorhynchus

tshawytscha) in Lake Michigan, nutritional status of 345 individuals was analyzed using
proximate composition analysis. Caloric content, protein, lipid and water were measured
for 3 size classes in the spring and the fall from locations on both the eastern and western
sides of Lake Michigan. ANOVA of changes in lipid content showed that small fish had
significantly lower lipid levels than medium or large fish. The seasonal patterns of lipid
dynamics were different between small and large individuals. Specifically, small fish
show a decline in lipid levels from fall to spring while large fish show a decline in lipid
levels from spring to fall. The energy allocation strategy shown by each size class is
different and small Chinook salmon seem to be using a strategy which could lead to a

high risk of severe nutritional stress, especially in the novel environment of Lake

Michigan.

Introduction

Managing individual energy reserves is critical to survival. Fluctuations in energy
reserves are commonly observed in environments where seasonal changes in temperature
and prey availability occur. A wide range of fishes show a pattern of increasing energy
reserves in summer, when temperatures are warm and prey availability is high, and

mobilizing those reserves during winter, when temperatures decrease and prey are scarce

35

(Dawson and Grimm, 1980; Jonas et al. 1996; Craig et al. 2000; Pedersen and Hislop,
2000; Brown and Murphy, 2004). Other stressors, such as migration (Phleger et al. 1989;
J onsson, Jonsson and Hansen 1997; Tocher 2003), disease, or maturation (Jobling et al.
1998; Silverstein et al. 1998; Amdt, 2000), can also initiate mobilization of energy
reserves. The events which trigger energy storage and mobilization lead to a dynamic
energy profile for many fishes and investigation of these patterns can aid in interpretation
of observed nutritional status and provide insight into the ability of a species to withstand
novel stressors.

Energy, or caloric content, can be divided into protein, lipid, and carbohydrate. In
fish, carbohydrate levels are usually very low, less than 1% and are often considered
negligible (Jonsson and Jonsson 1997; Edsall et al. 1999). Generally, small fish grow by
allocating energy into protein which increases muscle mass (Rikardsen and Elliott 2000,
Sutton et al. 2000, Morgan et al. 2002). This early growth may be important to decrease
vulnerability to predation and to allow intake of larger prey items (J onsson and Jonsson
1998). Once a certain size is reached, a greater amount of energy is allocated to lipid
stores (Morgan et al. 2000). Lipids are more calorically dense than protein and are the
most efficient method for energy storage in fish. It is these stored lipids that are
mobilized first during periods of stress; protein is only mobilized after a majority of the
lipid stores have been depleted and nutritional stress is occurring (Hendry and Berg,
1999; Morgan et al. 2002; Tocher 2003). Therefore, focusing on the dynamics of lipids is

appropriate for measurement of early fluctuations in energy stores. In addition to being

36

crucial for energy storage, lipids are important in triggering immune responses and
maintaining the integrity of cell membranes (Adams 1999; Lall 2000).

Migratory salmonids are ideal for studies on energy dynamics because they
inhabit temperate climates, undergo migrations, and are semelparous, which means a
large amount of energy is spent on their single spawning bout. In this study, we
investigated the energy dynamics of Chinook salmon, Oncorhynchus tshawytscha, in
Lake Michigan. Chinook salmon were introduced in Lake Michigan in the 19603 through
large-scale stocking efforts. After two decades of steady population growth, Chinook
salmon in Lake Michigan suffered a massive mortality event in the late 19803. This
event was associated with an outbreak of bacterial kidney disease (BKD), and it is
thought that this outbreak was linked to severe nutritional stress in the population which
made them more vulnerable to disease (Holey et al. 1998). Just prior to the outbreak of
this disease, observed growth rates of Chinook salmon were at the lowest levels since
their introduction (Wesley 1996, Peeters and Royseck 1997). The population crash
increased concern over the ability of Chinook salmon in Lake Michigan to maintain
adequate energy reserves in order to survive winter periods when energy intake is
presumed to be negligible.

Stocked Chinook salmon grow rapidly in Lake Michigan and mature between
ages 2 to 4, although some males are mature at age 1. This contrasts to Chinook salmon
in the Pacific, where average age of maturation is 3 to 5 years for males and 4 to 6 years
for females (Healey 1991). The rate of growth and timing of maturation are affected by

the availability, storage, and mobilization of energy. By assessing the nutritional state of

37

Lake Michigan Chinook salmon, we can gain insight into how energy content is related
to size and maturation. Specifically, we can look for evidence of changes in energy
allocation from smaller fish allocating energy to protein in order to increase in size to
larger fish allocating energy to produce lipid reserves in order to better survive times of
low food intake and to prepare for maturation. Such a pattern has been seen for many
fish species, including Atlantic salmon, Salmo salar (Dempson et al. 2004), red drum,
Sciaenops ocellatus (Craig et al. 2000), and brown trout, Salmo trutta (J onsson and
Jonsson 1998).

In Lake Michigan, Chinook salmon are exposed to a novel environment compared
to their native environment of the Pacific Ocean; specifically, water temperatures are
colder in the winter and the available forage is different. Previous research on another
introduced Pacific salmon in Lake Michigan, coho salmon (Oncorhynchus kistuch),
found that seasonal variation in lipid content was pronounced for coho salmon
(Oncorhynchus kistuch) while it was only modest for the native lake trout (Salvelinus
namaycush), and no seasonal pattern was found for native bloater (Coregonus hoyz)
(Madenjian et al. 2000). These findings, along with the stresses of the Lake Michigan
environment, lead to an expectation of a pronounced effect of season on nutritional state
for Chinook salmon. We predict that, due to the long, cold winter, Chinook salmon
should have much lower energy content in the spring compared to the fall.

There is also evidence that location in Lake Michigan can also have an effect on
lipid levels. Madenjian et al. (2000) found that location significantly affected lipid levels

in bloater, but not lake trout or coho salmon. Finally, variations in environmental

38

conditions from year to year can also affect nutritional state. A particularly strong year
class of prey can have a positive effect on nutritional state of its predators while an
extremely harsh winter or a decline in prey might negatively affect energy content.
Accordingly, in this study we have evaluated the effects of maturity, size, season,
location and year on energy content, and specifically lipid levels, in Lake Michigan
Chinook salmon. A better understanding of Chinook salmon energy dynamics in Lake
Michigan will provide insight about how an introduced species adjusts, or fails to adjust,
its life history to cope with a novel environment.
Methods

We collected Lake Michigan Chinook salmon beginning in the fall of 2000 and
continuing until the spring of 2003. Spring samples were collected from April 15th to June
15th and fall samples were collected fi'om August 15th to September 15th. Fish were
collected from a variety of sources including volunteer contributions from anglers and
charter operators and from an ongoing Michigan Department of Natural Resources
(MDNR) survey. All fish were caught in the open waters of Lake Michigan. The fish
provided by the MDNR were caught using gill nets and for each fish weight was recorded,
length was measured, stomachs were removed and gender and maturity status were
recorded by MDNR staff aboard the USS Steelhead. Fish were then placed on ice, and
transported to Michigan State University (MSU). Fish caught by anglers and charter
operators were put on ice and transported to MSU. Once at MSU, these samples were
weighed, measured, had stomachs removed and were examined for gender and maturity

status. Date and location of collection for all samples was recorded. Fish were designated

39

into length categories using the following criteria: small (<490 mm spring, <550 mm fall),
intermediate or large (>700 mm spring, >775 mm fall). Previous data on size at age were
used to determine these length categories so that they approximated age 1, 2, and 3 fish
(Robert Elliott, US. Fish and Wildlife Service, Green Bay, WI, unpublished data).

Stomach content analysis was done by weighing the total contents of the stomach
and then dissecting the contents to determine the number of prey items. Prey items were
identified to species. Each stomach was also coded as either empty or containing at least
one food item.

While still frozen, each whole fish was sawed into long, thin strips with a band saw
and then homogenized using a Hobart Meat grinder. Samples were passed through the
grinder two times to ensure a well-mixed sample. About 500 grams of each homogenized
fish was then stored in a freezer and saved for proximate composition analysis (PCA). In
our study, PCA included analyses of water, caloric density, protein, and lipid.

For each individual, water content analysis was done by placing a known amount of
wet material (approximately 2 grams) in a pre-weighed aluminum pan, and then heating it
in a 100°C oven for 24 hours. The pans and samples were then cooled in a dessicator for at
least one hour. After cooling, the pans and dry material were weighed again to determine
water content by difference.

The remaining compositional analyses were performed on freeze-dried material.
Freeze-drying was performed to remove water from the samples without heating using a

FTS Kinetics unit (FTS Kinetics, Stone Ridge, New York) .

40

The caloric content for the homogenized, freeze-dried whole fish was assessed
using a 1241 Adiabatic calorimeter (Parr Inc, Moline, Illinois). Pellets of approximately
0.75 grams were made for each sample and were burned inside the bomb chamber. The
water temperature surrounding the bomb chamber was measured before and after the
burning of the pellets and the increase in water temperature caused by the burning of the
pellets was converted into calories.

The protein content was determined for the homogenized, freeze-dried whole fish
using a LECO 2000 Nitrogen Analyzer (Leco Inc, St. Joseph, Michigan). This apparatus
burns freeze-dried material in a controlled helium and oxygen atmosphere and then
determines percent protein based upon nitrogen content during combustion (AOAC
990.03).

The Soxtec method was used to determine lipid content (Soxtec System HT6;
Tecator, Sweden). This method involved placing freeze-dried whole fish in a cellulose
thimble through which hot ether is percolated. The ether extracts the lipid which is
collected in a metal cup and the ether is then evaporated off. The extracted lipid is then

weighed to determine lipid content (AOAC, 1995).

Data Analysis:

Mean values of two replicate samples for each individual fish were used for all
analyses. Values for lipids and proteins are reported as a percentage of dry mass of the
homogenized whole fish and caloric content is reported as calories per gram of dry mass

of the homogenized whole fish. No qualitative difference was found in the results or

41

conclusions when analyses were repeated on a wet weight basis. All percentage data
were transformed (arcsine-square root) for statistical analyses in order to normalize the
data. Relationships between the caloric content, lipid, and protein were evaluated by
calculating pair-wise Pearson correlations.

A mixed-effects ANOVA was used to estimate the effect of year, season (fall or
spring), size class (small, medium and large) and location (eastern or western side of Lake
Michigan) on lipid content. Year and its interaction with season were modeled as random
effects in the analysis. A year by season interaction was not significant, and a year by
location interaction was not estimable. The other variables were modeled as fixed effects.
Maturity and sex were included as fixed-effects in some initial data analysis, but were not
included in the final ANOVA for two reasons: (1) for some combinations of year and size
class there was a sample size of zero when a certain sex or maturity level was included and
(2) when other factors were included neither maturity nor sex were significant factors.

Thus, the full model we considered was:

Yj=P+Bn+Ys+51+Wns+lnl+93l+wsy +3j
5s1~ N(0.625) and 8j ~ N(0,62 a)

where sz Sine'1 (% Lipid)”2 for the jth fish, II is the overall mean, 13 is the effect of size
class n, y is the effect of season s, 5 is the effect of location I ,w is the interactive effect of
season and size, A is the interactive effect of size and location, 0 is the interactive effect of

season and location, y is year, a) is the random interactive effect of year and season, and 8
is the residual error term.

42

We evaluated this full model and reduced models where some effects were omitted. Model
selection was performed in two steps using Akaike Information Criterion (AIC) computed
as

AIC = -2L + 2q

where L is the maximized log likelihood and q is the number of parameters (Bumham and
Anderson, 2004). In the first step, we defined the error structure by determining which
random effects, if any, to include. We did this by comparing our full model and the same
fiilly saturated fixed effect model where only year or only the year by season interaction
were included as random effects. The error structure of the saturated model with the lowest
AIC score was then used in subsequent analyses to determine which fixed effects should be
retained. Fixed interactions were only included in models that contained the corresponding
main effects. Again, the models with the lowest AIC scores were chosen. Finally, to

assess the support for the competing models Akaike weights were computed as

Wi = exp(-Ai/2j

Zr exp (”Ar/2)

where wi is the weight of evidence in favor of model i, Ai is the difference in AIC between

the best model and the model i, and -Ar is the difference between model r and the best

model (Bumham and Anderson, 2004). Least squares means were computed from the best

model to assess how different variables influenced lipid content.

43

Results

A total of 345 chinook salmon were collected during 6 sampling periods,
beginning in the fall of 2000 and ending in the spring of 2003 (Table 2.1). The samples
were composed of 80% gill net caught fish and 20% angler caught fish. Stomach content
analysis showed that 99% of prey items were alewife (Alosa pseudoharengus) and less
than 1% were rainbow smelt (Osmerus mordax). A higher percentage of empty stomachs
were found in the spring (69%) than in the fall (33%). In the spring, fish caught in gill
nets had a slightly higher percentage of empty stomachs (68%) than those caught by
anglers (64%). In the fall, the fish caught in gill nets had a lower percentage of empty
stomachs (28%) than those caught by anglers (47%). Also, a higher percentage of empty
stomachs were found in the samples collected from the western side of Lake Michigan
(64%) compared to the eastern side (41%).

Linear regression analysis showed that both lipids and protein explained a
significant amount of the variation in caloric content (Fig. 2.1). As the percentage of
lipid in the dry mass increased, caloric content in the fish increased (r2=0.68). The
opposite is true for protein, as the percentage of protein in the dry mass increased, caloric
content in the fish decreased (r2=0.66).

The AIC value for the saturated fixed effects model which included only a year by
season interaction as a random effect was 10 AIC units less than the next best model,
which included only year as a random effect. This large difference indicates that the
alternative error structures were not viable contenders, and we proceeded with subsequent

analyses using only a year by season interaction for the error structure. There were only

44

four fixed effects models that were within 20 AIC units of the best model (Table 2.2), In
further work we only considered Model 1 (Akaike weight of 0.60) and Model 2 (Akaike
weight of 0.29) because weights for other models were less than .05. As noted above all
the models considered at this stage included only the interaction of season and year as a
random effect. Both models 1 and 2 included main effects of size class, season, and
location and the interactions of size class by season, and size class by location. Model 2
also included the interaction between season and location that was absent from model 1.
Other interactions between fixed effects were not included in the best two models (Table
2.2).

Lipid content varied significantly with size class (Table 2.3), with smaller fish
having lower lipid levels (LSM 13.5%, 95% CI {1 l.6%,15.5%}) than the intermediate
(LSM 26.9%, 95% CI {25.2%,28.7%}) or the large (LSM 26.7%, 95% CI {24.5%,
28.9%}) size classes. Small fish had lower lipid levels on the western side of Lake
Michigan than on the eastern side, whereas lipid levels were similar on both sides of the
lake for other size classes (Fig. 2.2). This pattern produced significant main effects of
location and the interaction between size class and location (Table 2.3). The influence of
small fish on the statistical results is reinforced by noting that both the main effect of
location and its interaction with size class are no longer significant when small fish were
excluded from the ANOVA model.

Lipid levels changed differently from spring to fall for different sizes of fish (Fig
2.4), and averaged over sizes of fish showed inconsistent patterns among years (Fig. 2.3).

Lipid levels dropped from fall to spring for small fish, changed little for medium fish, and

45

dropped from spring to fall from large fish. These patterns were reflected in a significant
interaction of size class and season. In Year 1, lipids averaged over size classes
decreased from fall to spring, whereas in Year 3 they increased, with Year 2 showing a
moderate decrease. These patterns led to a significant effect of the random season by
year interaction (Table 2.3). The varying seasonal patterns among size classes and
seasons led to a non-significant main effect of season.

This difference in seasonal patterns of lipids based on size class was also seen in
seasonal trends in overall body composition for each size class (Table 2.4). In small fish,
lipids and caloric content decreased from fall to spring, but there was an increase from
fall to spring in mean water content (72.7% to 77.6%). The opposite seasonal pattern was
true for large fish. The lipid content and caloric content were higher in the spring than in
the fall while mean water content (70.5% to 72.5%) and mean protein content (62.9% to
74.6%) were lower in the spring than in the fall.

Discussion

As found in studies of other fish, percent lipid content increased in Lake Michigan
Chinook salmon as they grew larger (Jonsson and J onsson 1998; Madenjian 2000;
Dempson et al. 2004). This is consistent with the idea that as fish grow they begin to
allocate surplus energy to lipid storage. Overall, lake location explained a significant
amount of the variation in lipid content. However, it is important to note that the effect
of size class on lipid content was more than three times as large as the effect of location.
It is clear that small fish are driving the significant difference in lipid levels between

locations, as evident from the interaction between size and location. Perhaps food

46

sources for small individuals vary enough between the eastern and western sides of Lake
Michigan to cause this observed difference in lipid levels. This is supported by the
stomach content data which show a higher percentage of empty stomachs in the samples
collected from the western side of the lake than from those collected from the eastern
side. However, the majority of the samples collected on the eastern side of the lake were
collected by the MDNR using gill nets while the majority of samples collected on the
western side of the lake were from anglers or charter captains. This difference in
collection methods could be an alternative explanation to the difference in percentage of
empty stomachs between the two groups.

Previous studies have explained fluctuations in lipid levels in fishes by seasonal
changes in temperature and prey availability (Jonas et al. 1996, Metcalfe et al. 2002,
Morgan et al., 2002) or by energy requirements for reproduction (Kadri et al. 1996, Craig
et al. 2000). In this study, we see evidence for both of these explanations depending on
the size class of the samples. The small Chinook salmon showed a significant decline in
lipid levels from fall to spring. Because these fish are not reproducing, this drop is likely
caused by lipid mobilization to meet metabolic requirements during winter when fish
feed very little, if at all, and temperatures are cold. For the intermediate and large size
classes, however, we did not see the same pattern of seasonal fluctuation in whole-fish
lipid content. The large size class had higher lipid levels in the spring which dropped
significantly by fall, suggesting that energy requirements for reproduction were driving
this pattern. Kadri et al. (1996) found a drop in lipid levels in maturing Atlantic salmon

in mid-summer while immature individuals continued to increase lipid content. A similar

47

decrease was noted in muscle and visceral fat during summer months prior to maturity in
captive sockeye salmon (Hendry et al. 2000). We suggest that as large Chinook salmon
begin the maturation process in the spring, they mobilize lipid reserves to produce
gametes and secondary sexual characteristics resulting in a drop in overall lipid content.
Female salmonids produce large, lipid-rich eggs which require a large energetic
investment and the males develop significant secondary sexual characteristics which also
require energy (Hendry et al. 2000). Since all our samples were caught in the open
waters of Lake Michigan, this significant decline in lipid content for large fish occurred
before the large fish began to stage at the mouths of rivers to begin spawning migrations,
a period during which active feeding stops. Our stomach content analysis shows that
50% of the large individuals had at least one prey item in their stomach when they were
caught in the fall, showing that active feeding was still occurring.

For the intermediate fish, we found no significant change in lipid levels between
the spring and the fall. Perhaps since the intermediate size group is already large enough
to avoid predation and consume large prey and the majority of the intermediate-sized fish
are not preparing energetically for reproduction in the fall, they are able to maintain lipid
levels at a constant percent of their mass. Maintaining a more constant lipid level is a
strategy which is found among the native piscivores of Lake Michigan (Madenjian et al.
2000) and seems to be a strategy which would reduce the likelihood of nutritional stress.
The marked seasonal fluctuation of lipid levels in small Chinook salmon in Lake
Michigan results from a strategy of early, rapid growth with little energy put toward lipid

storage. Indeed, age 1 salmon from our study had a mean length of 490 mm, which falls

43

within the observed mean range for age 2 Pacific Chinook salmon of 450-550 mm
(Healey, 1991). This energy allocation strategy, which is more risky than maintaining a
constant lipid level, could lead to severe nutritional stress among the small fish in the
spring.

Each year of the study showed a different pattern of overall seasonal change in
lipid content, as seen by the significance of the random effect of the interaction of year by
season. With a sample size of only three, it is difficult to draw any strong conclusions
from this result. However, it does provide an important insight in that some other
variable is affecting lipid levels differently each year. In sockeye salmon, 0. nerka,
somatic energy levels at the time of entry into rivers for spawning were positively
correlated with ocean productivity (Crossin et al., 2004). For Lake Michigan Chinook
salmon, factors such as winter severity or prey abundance could be important in
explaining year effects.

If Chinook salmon are to continue being an economically important recreational
fishery for the Great Lakes, we must better understand the stresses which the novel
environment of Lake Michigan places on their energy budget. In years with particularly
cold or long winters, or in times of decreased food availability, it might be extremely
difficult for a small Chinook salmon to maintain the lipid levels required to survive
winter. Our data show that the energy allocation strategy seems to be different for each
size class and the strategy of rapid growth and low lipid reserves seen in small Chinook
salmon exacerbates the challenge of over-wintering in a novel environment which is

thermally harsher than the native environment. Since their introduction into Lake

49

Michigan, Chinook salmon have grown faster and matured at a younger age than Pacific
Chinook salmon (Healey, 1991). Such changes are not ideal for a recreationally
important fish. Researching the energy allocation strategy and energy dynamics of Lake
Michigan Chinook salmon will allow us to better understand how the novel environment
has led to these changes and how the current nutritional status of this population will
affect its ability to sustain fishing pressure and to withstand other stressors such as

disease or parasitism in the future.

50

References

Adams, SM. 1999. Ecological role of lipids in the health and success of fish
populations. In Lipids in freshwater ecosystems. Edited by M. T. Arts and B. C.
Wainman, Springer-Verlag, New York, NY. pp. 132-160.

AOAC. 1995. Official Methods of Analysis. 16th ed. Assoc. Offic. Anal. Chem,
Arlington, VA.

Amdt, S.K.A. 2000. Influence of sexual maturity on feeding, grth and energy stores
of wild Atlantic salmon parr. J. Fish. Biol. 57: 589-596.

Brown, ML. and Murphy, BR. 2004. Seasonal dynamics of direct and indirect
condition indices in relation to energy allocation in largemouth bass Micropterus
salmoides. Ecol. Freshw. Fish 13: 23-36.

Burnham, K.P. and Anderson, DR. 2004. Model Inference Understanding AIC and BIC
in Model Selection. Sociol. Methods and Res. 33: 261-304.

Craig, S.R., MacKenzie, D.S., Jones, G., and Gaitlin, D.M. 2000. Seasonal changes in
the reproductive condition and body composition of free-ranging red drum, Sciaenops
ocellatus. Aquaculture 120: 89-102.

Crossin, G.T., Hinch, S.G., Farrell, A.P., Higgs, D.A., and Healey, MC. 2004. Somatic
energy of sockeye salmon Oncorhynchus nerka at the onset of upriver migration. Fish.
Oceanogr. 13: 345-349.

Dawson, AS. and Grimm, AS. 1980. Quantitative seasonal changes in the protein, lipid
and energy content of the carcass, ovaries and liver of adult female plaice, Pleuronectes
platessa L. J. Fish. Biol. 16: 493-504.

Dempson, J .B., Schwarz C.J., Shears, M. and Furey, G. 2004. Comparative proximate
body composition of Atlantic salmon with emphases on parr from fluvial and lacustrine
habitats. J. Fish Bio. 64: 1257-1271.

Edsall, T.A., Frank, A.M., Rottiers, D.V., and Adams, J .V. 1999. The effects of
temperature and ration size on the growth, body composition, and energy content of

juvenile coho salmon. J. Great Lakes Res. 25: 355-362.

Healey, MC. 1991. Pacific salmon life histories. UBC Press, Vancouver, British
Columbia.

51

Hendry, AP. and Berg, OK. 1999. Secondary sexual characters, energy use,
senescence, and the cost of reproduction in sockeye salmon. Can. J. Zool. 77: 1663-
1675.

Hendry, A.P., Dittman, A.H., Hardy, R.W. 2000. Proximate composition, reproductive
development, and a test for trade-offs in captive sockeye salmon. Trans. Am. Fish. Soc.
129: 1082-1095.

Holey, M. E., Elliot, R.F., Marcquenski, S.V., Hnath, J .G., and Smith, K. D. 1998.
Chinook salmon epizootics in Lake Michigan: possible contributing factors and
management implications. J. Aquatic Animal Health 10: 201-210.

Jobling, M., Johansen, S.J.S., Foshaug, H., Burkow, I.C., Jorgensen, EH. 1998. Lipid
dynamics in anadromous Arctic charr, Salvelinus alpinus (L.): seasonal variations in lipid
storage depots and lipid class composition. Fish Physiol. and Biochem. 18: 225-240.

Jonas, J ., Kraft, C., and Margenau, T. 1996. Assessment of seasonal changes in energy
density and condition in age-0 and age-1 muskellunge. Trans. Am. Fish. Soc. 125: 203-
210.

Jonsson, N., and Jonsson, B. 1998. Body composition and energy allocation in life
history stages of brown trout. J. Fish Biol. 53: 1306-1316.

Jonsson, N., J onsson, B. and Hansen, LP. 1997. Changes in proximate composition and
estimates of energetic costs during upstream migration and spawning in Atlantic salmon,
Salmo salar. J. Animal Ecology 66: 425-436.

Kadri, Sunil, Mitchell D.F., Metcalfe, N.B., Huntingford, EA, and Thorpe, JE. 1996.
Differential patterns of feeding and resource accumulation in maturing and immature
Atlantic salmon, Salmo salar. Aquaculture 142: 245-257.

Lall, S. P. 2000. Nutrition and Health of Fish. In Avances en Nutricion Acuicola V.
Edited by L. E. Cruz-Suarez, D. Ricque-Marie, M. Tapia-Salazar, M. A. Olvera-Novoa,
and R. Civera-Cerecedo. Merida, Yucatan, Mexico. pp. 13-23.

Madenjian, C.P., Elliot, R.F., DeSorcie, T.J., Stedman, R.M., O'Connor, D.V., and
Rottiers, D.V. 2000. Lipid concentrations in Lake Michigan fishes: seasonal, spatial,
ontogenetic, and long-terrn trends. J. Great Lakes Res. 26: 427-444.

Metcalfe, N.B., Bull, CD, and Mange], M. 2002. Seasonal variation in catch-up growth
reveals state-dependent somatic allocations in salmon. Evol. Eco. Res. 4: 871-881.

Morgan, I.J., McCarthy, I.D., and Metcalfe, NE. 2000. Life-history and protein
metabolism in over-wintering juvenile Atlantic salmon. J. Fish. Bio. 56: 63 7-647.

52

Morgan, I.J., McCarthy, I.D., and Metcalfe, NE. 2002. The Influence of Life-History
Strategy on Lipid Metabolism in Over-wintering Juvenile Atlantic salmon. J. Fish. Bio.
60 674-686.

Pedersen, J. and Hislop, J .R.G. 2001. Seasonal variations in the energy density of fishes
in the North Sea. J. Fish Biol. 58: 380-389.

Peeters, PL and Royseck, KR 1997. Harvest, age and size at age of Chinook and coho
salmon at Strawberry Creek weir and Besadny Anadromous Fisheries Facility Fall 1996.
Wisconsin Department of Natural Resources, Report, Sturgeon Bay, Wisconsin.

Phleger, C.F. Laub, R.J. and Benson, AA. 1989. Skeletal lipid depletion in spawning
salmon. Lipids 24: 286-289.

Rikardsen, AH, and Elliot, J .M. 2000. Variations in juvenile growth, energy allocation,
and life-history strategies of two population of Artie charr in North Norway. J. Fish Biol.
56: 328-346.

Silverstein, J .T., Shearer, K.D., Dickhoff, W.W., and Plisetskaya, EM. 1998. Effects of
growth and fatness on sexual development of Chinook salmon. Can. J. Fish. Aquat. Sci.
55: 2376-2382.

Sutton, S.G., Bult, T.P., and Haedrich, KL. 2000. Relationship among fat weight, body
weight, water weight, and condition factors in wild Atlanic salmon parr. Trans. Am. Fish.
Soc. 129: 527-538.

Tocher, DR. 2003. Metabolism and Functions of Lipids and Fatty Acids in Teleost
Fish. Rev. Fish. Sci., 11: 107-184.

Wesley, J .K. 1996. Age and growth of Chinook salmon in Lake Michigan. Michigan

Department of Natural Resources, Fisheries Division, Fisheries Research Report 2029,
Lansing, Michigan.

53

Period Location Maturig Status Size Class* Total

West East 1mm. Mat Sm. Med. Lg
Fall 2000 0 38 15 23 12 18 8 38
Spr 2001 7 24 25 6 4 20 7 31
Fall 2001 19 68 58 29 33 44 10 87
Spr 2002 7 43 26 24 8 24 18 50
Fall 2002 24 45 24 45 19 22 28 69
Spr 2003 8 62 28 42 5 39 26 70
Total 65 280 176 169 81 167 97 345

* Fish were categorized as small if they measured <495mm in the spring or < 550mm in fall.
Fish were categorized as large if they measured >700mm in spring or >775mm in the fall.

Table 2.1. Summary of samples collected by location, sex, maturity status, and size

class.

54

Model of Fixed Effects AIC difference from best model

SN, Size, Loc, SN by Size, Size by Loc 0
SN, Size, Loc, SN by Size, Size by Leo, SN by Loc 1.4
SN, Size, Loc, SN by Size, Size by Loc, SN by Loc, SN by Size by Loc 4.6
SN, Size, Loo, SN by Size, SN by Loc 18
SN, Size, Loc, SN by Size 18.7
SN, Size, Loc, Size by Loc 81.9
SN, Size, Loc 91.7
SN, Size, Loc, SN by Loc 93.1
SN, Size 97.1
Size, Loc 102.5
Size 112.3
SN, Loc 116.4
SN 122.3
Loc 123.4

Table 2.2. AIC differences from the best model for all possible combinations of fixed

effects and higher interactions where SN=season, Size=size class, and Loc=Location.

55

DF
Intercept (fixed) 1
Season (fixed)1 1
Size class (fixed)2 2
Location (fixed)3 1
Season x Size class 2
(fixed)

Season x Year (random)4 4

Location x Size class 2
(fixed)

Error 326

Total 339

1 Season was Spring or Fall

SS

34.1456

0.02001

0.73292

0.06076

0.64384

0.17678

0.10943

2.31461
4.0583

MS

34.1456

0.02001

0.33646

0.06076

0.032192

0.04419

0.05471

0.00710

F

1304.74

0.574

51.614

8.557

45.341

6.225

7.706

<0.001

0.4867

<0.001

0.0037

<0.001

<0.001

0.0005

2 Samples were divided into small (<495 mm spring, <550 mm fall), medium, large (>700 mm

spring, >775 mm fall) size classes.

Location of samples was noted as East or West

4 Three years were included in the study. Year 1 includes Fall 2000 and Spring 2001, Year 2
include Fall 2001 and Spring 2002, Year 3 includes Fa112002 and Spring 2003

Table 2.3. ANOVA for Lipid Content in the Whole Fish.

56

Size Season
Small Fall (62)

Spr.(16)

Med. Fall (88)

Spr. (83)

Large Fall (59)

Spr. (50)

Calories
5597 :i: 55

5287 :l: 99

5680 i 34

5876 i 37

5503 i 53
6008 :I: 36

Lipid
22.0 i 0.81

10.8 :I: 2.3

23.9 i 0.76
26.1: 0.86

16.4 i 0.98
31.3 i 0.90

Protein
68.0 :1: 1.6
83.6 i 2.9

66.3: 1.1
69.5:t1.1

74.7 t 1.6

62.9 i 0.95

Water
72.7 t. 2.7

77.6 i 1.3

71.3 i 2.3

73.0 i 2.1

72.5 :I: 2.8
70.5 i 2.3

Table 2.4. Mean calories (per gram dry), % lipids (per gram dry), % protein (per gram

dry), and % water of small, medium and large fish by season with standard error of

mean and sample size (n).

57

2400 -.-..--.--..--.--

 

 

 

 

 

 

 

 

 

 

 

 

 

2200i a) [ r=0.83; p<0.0001] °
I O

E 2000 l q
(6
E i
as 1800 -
3
9 1600 - '
g .
‘5‘, 1400 - . .
E l o 00 .
O . o
3 1200~ .. .o .-
5 . . : O O
8 1000 r '00 ~. . O . o

800 - .

600 - l . k - . - . - 1 . A - 4 - -

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
% Lipid (g dry mass, transformed)
2400 . 1 . - - . . - . - . - - . - - . .....
. [r= 0.79; p < 0.0001]
2200 . - . b)
:. ..o.~

’9‘ 2000~ . '.
g 0 . .0 ’ . .3. 0 9.
8 1800 ’ o . .0: .'O o~$0 o.
3 ;-{. . o .0 0 o
O o“ o o o .
L. 1600 . g . o. gafi'fi ‘:o'.o. . .2 O
a o 0 Or. . (?‘.o? .0: ‘0.
5 .0 ‘ . 8. . o... 0.
81200, . ....° ‘. . ,
.9 O . o O
:3 1000 l . . . o. . 4

800 -

600 - Rh 1 A I - A 1 . - - n - n 1 - -

0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5

% Protein (9 dry mass. transformed)

Figure 2.1. Pearson correlation for caloric content per gram wet mass for (a) % lipid and
(b) % protein. Percentage data was transformed using the arcsine, square root

transformation.

58

35 . fl

 

 

 

 

 

 

 

 

 

 

     

 

 

 

 

30 ~
................... M,,,____,,,,______..-—-—---—-—-1
,P___ .. ............................................
a ...... ,,
to
E
a —_
‘U
221’ 20 _ __
F i
It
-; 15 . ----
°\ —— Small
— — — Medium —
1O . .............. Large ’
5
East West

Figure 2.2. The joint effects of size class and location of collection on % lipid per gram
dry mass. Least squares means from ANOVA (back-transformed) are shown with bars at

95% confidence intervals, p<0.01.

59

35

 

30-

25-

 

 

 

 

20-

 

 

 

 

15» 1

% Lipid (per gram dry mass)

 

Yearl
10 , _-—Year2
.............. Year3

 

 

 

Fall Spring

Figure 2.3. The joint effects of year and season on % lipid per gram dry mass. Least
squares means from ANOVA (back-transformed) are shown with bars at 95% confidence

intervals, p<0.0001.

60

40

 

 

 

 

 

 

 

 

 

 

 

 

 

 

35 i
..... ii
7; 30 - __ ...........
m ......
g 0—————-———..,.". ------- 1 I
25 I __ ....... __
E —— ..........
E (D ......
‘3 20 . ‘
3 __\—
5:
‘o 15 >
E
.J
g. 10 — Small
-- — - Medium ,
5 .............. Large
0
Fall Spring

Figure 2.4. The joint effects of size class and season on % lipid per gram dry mass.
Least squares means from ANOVA (back-transformed) are shown with bars at 95%

confidence intervals, p<0.00001.

61

CHAPTER 3
A PROPOSED MONITORING PROGRAM FOR LAKE MICHIGAN CHINOOK
SALMON (ONC ORHYNCH US T SHA WYTSHCA)
Abstract
Monitoring health indicators of fish populations can be an expensive and time

consuming process. This study analyzed energy dynamics of Lake Michigan Chinook
salmon using proximate composition analysis with the goal of determining an efficient
method for monitoring the nutritional status of the population. Condition factor
performed poorly as an indicator of whole-fish lipids (r2=0.07). Water content in a dorsal
muscle plug was found to be correlated with whole-fish lipids (r2=0.50) for all samples.
For the subset of samples that included small fish collected in the spring, the strength of
the relationship between water content in a dorsal muscle plug and whole-fish lipids
increased (r2=0.70). The metric of water content in a dorsal muscle plug was determined
to provide an adequate surrogate of whole-fish lipid content and, therefore, overall
nutritional status. We propose a monitoring program that involves collecting small
individuals in the spring and reporting the proportion of samples with over 78% water
content in muscle tissue. Small individuals collected in the spring had the lowest whole-
fish lipid levels of any segment of the population and would be the most prone to

nutritional stress; therefore we recommend focusing on them for monitoring.

Introduction
Chinook salmon (Oncorhynchus tshawytscha) provide a popular and
economically important recreational fishery in Lake Michigan and other Great Lakes

62

(Bence and Smith 1999). This species was introduced into the Great Lakes, and the
fishery developed from a stocking program that began in the late 19603 and continues to
the present day (Holey et al. 1998, Kocik and Jones 1999). After two decades of
uninterrupted increase, the population, and its associated fishery, collapsed in the late
19803 before rebuilding during the mid to late 19903 (Benjamin and Bence 2003). There
is now considerable evidence to suggest that the collapse was associated with an outbreak
of bacterial kidney disease (Johnson and Hnath, 1991). In addition, infections caused by
Pseudomonasfluorescens, Aeromonas hydrophila, and acathocephalus (E. salmonis)
were also observed (Johnson and Hnath, 1991; Holey et al. 1998). It has been postulated
that nutritional stress resulting from an excess of Chinook salmon relative to their primary
prey, the alewife (Alosa pseudoharengus), was at the root of the problem (Holey et al.
1998). Therefore, an understanding of the nutritional status of the population would be
useful information to guide management action in order to avoid another catastrophic
decline in Chinook salmon abundance. Management options could include reducing
stocking rates or increasing Chinook harvests to restore the balance between predators
and prey when nutritional stress is present in the Chinook salmon population.

Chinook salmon are an anadromous species and thus adapted to life as juveniles
in oceanic, rather than freshwater, environments. Great Lakes Chinook salmon are
forced to occupy environments that are thermally different from oceanic habitats.
Limited food availability during this time of thermal stress could accelerate the depletion
of lipid reserves, and increase vulnerability to disease because immune responses in

ectotherrns have been shown to be temperature dependent (Lall 2000) and lipids are the

63

source of fatty acids for immune function regulation (Sargent et al. 1989; La112000).
Thus, cold winter conditions in the Great Lakes may substantially increase disease risk
for Chinook salmon, particularly if a cold winter is preceded by a poor growing season
and a limited opportunity for accumulation of lipids. The Great Lakes contain abundant
coldwater habitats, so even under plausible scenarios of future climate change it is very
unlikely that Chinook salmon in the Great Lakes would encounter a limited supply of
cold water (Magnuson et al., 1990).

We have found that the lipid content in Lake Michigan Chinook salmon varies
among both size classes of fish, and seasons, and that these two sources of variation
interact. We found yearling salmon (small salmon, <495 mm), contain significantly
lower lipid levels than larger fish (Peters et al. , in review). We also showed that seasonal
lipid dynamics differed between small and large Chinook salmon; spring-caught small
fish had significantly lower lipid content than they had the previous fall, while larger
salmon showed evidence of declining lipids from spring to fall. Small fish showed the
greatest decline in lipid levels during winter and the lowest absolute levels of lipid
content in the spring. These findings suggested that small Chinook are at the greatest risk
of stress due to low energy reserves if challenged by extreme environmental conditions
(Peters et al., in review). As well, an analysis of Chinook salmon mortality dynamics
during the late 19803 epizootic indicated that the greatest increase in natural mortality
rates was experienced by small fish (Benjamin and Bence, 2003).

Several indices have been developed to assess the health or nutritional status of

fishes. Fulton’s condition factor looks at the relationship between length and weight of

64

the fish. Hepatosomatic index (HSI), relates the weight of the liver to the weight of the
fish. Goede and Barton (1990) developed a suite of characteristics including the condition
of the gill rakers, liver, spleen, eyes, and kidney. The latter method appears to have more
merit than the other two, but is labor intensive and therefore used less frequently.
Fulton’s condition index and HSI may not accurately determine fish nutritional status
(Jonas et al. 1996; Sutton et al. 1999; Copeland et al. 2004, Trude] et al. 2005).
Recently, more emphasis has been placed on the role of lipids in fish health
(Madenjian et al. 2000; Morgan et al. 2002). Energy rich lipids accumulate during
periods of food abundance and are mobilized during periods of food scarcity (Adams
1999; Morgan et al, 2002). Over-winter survival for many fish species has been shown
to depend on adequate lipid reserves (Oliver et al. 1979; Adams et a1. 1985; Pedersen et
al. 2001). Madenjian et al. (2000) reported large declines in total lipid levels in Lake
Michigan coho salmon (0. kisutch) between fall and spring during their second winter.
The primary prey, alewife, (Alosa pseudoharengus) for both coho and Chinook salmon,
have also been shown to undergo pronounced seasonal changes in lipid content in Lake
Michigan (Flath and Diana 1985; Madenjian et al. 2000). Lipids are clearly important
indicators of nutritional health and vulnerability to disease, but measurement of lipid
levels in whole fish is expensive and time-consuming. The objective of this study was to
examine alternative methods of assessing nutritional health and to recommend an
efficient and inexpensive approach to assessing Great Lakes Chinook salmon
populations. We sought a method which would provide reliable evidence of when whole-

fish lipids in a population, particularly for spring-caught small fish, are reaching low

65

enough levels to be of concern. We investigated what variable to measure, whether a
certain segment of the population was more prone to low lipid levels than other segments,
during what month(s) samples should be collected, what area of the lake should be
targeted for collection, and what values would be interpreted as signs of severe nutritional
stress.
Methods
Field Study

We collected Lake Michigan Chinook salmon beginning in the fall of 2000 and
continuing until the spring of 2003. Spring samples were collected from April 15th to
June 15th and late summer/fall samples were collected from August 15th to September
15‘“. Fish were collected from a variety of sources including volunteer contribution from
anglers and charter operators and from an ongoing Michigan Department of Natural
Resources (MDNR) survey. All fish were caught in the open waters of Lake Michigan.

All tissue samples and the whole carcasses were kept on ice and then stored in
freezers. The fish provided by the MDNR were caught using gill nets and were
processed on the boat. Fish caught by anglers and charter operators were put on ice and
transported to Michigan State University for processing. Fish were weighed and
measured for length. The liver, and a dorsal muscle plug of approximately 50 grams
from directly behind the dorsal fin were collected and bagged for later analysis. Fish
gender, maturity status, along with the date and location of collection were recorded.
Fish were designated into three size categories using the following length criteria: small

(<490 mm spring, <550 mm fall), medium, or large (>700 mm spring, >775 mm fall).

66

Previous data on size at age were used to determine these length categories. Size
categories of small, medium and large corresponded with fish age of approximately 1, 2,
and 3 years, respectively (Robert Elliott, US. Fish and Wildlife Service, Green Bay, WI,
unpublished data).

While still frozen, each whole fish was sawed into long, thin strips with a band
saw and then ground and thoroughly homogenized using a Hobart Meat grinder. Samples
were passed through the grinder two times to ensure a well-mixed sample.
Approximately 500 grams of each homogenized fish was stored frozen for proximate
composition analysis. In our study, proximate composition analysis included analyses of
water content, caloric content, protein content, and lipid content.

Water content or dry matter analysis was done on the whole fish, the muscle
sample, and the liver sample of each fish (AOAC 1995). The remaining compositional
analyses were performed on freeze-dried material. F reeze-drying was performed to
remove water from the samples without heating (FTS Kinetics, Stone Ridge, New York).
Each sample was weighed before and after freeze-drying so that values from the protein,
lipid, and caloric analyses could be expressed on both dry and wet weight bases.

The caloric content for the homogenized, freeze-dried whole fish was assessed
using a 1241 Adiabatic calorimeter (Parr Inc, Moline, Illinois). Crude protein content
was determined for the homogenized, freeze-dried whole carcass and a freeze-dried
muscle plug using a LECO 2000 Nitrogen Analyzer (Leco Inc, St. Joseph, Michigan)
(AOAC 990.03). Sample measures were repeated until the coefficient of variation of the

mean estimate was acceptably low (<3%) for each samples. The Soxtec method was

67

used to determine lipid content (Soxtec System HT6; Tecator, Sweden) (AOAC, 1995).
This analysis was conducted on freeze-dried, homogenized whole fish, muscle tissue and
liver tissue. Duplicate samples were done on each individual. For this study, we report
all analyses on a wet weight basis because we are most interested in looking at
predictions of whole fish composition.

We analyzed all fish to look for associations among indicators, but because of
earlier work suggesting that small fish in the spring were of the greatest interest, we also
compared patterns between indicators for this subset of the overall sample.

Winter Simulation Study

To provide baseline data for compositional values of fish in a known nutritional
state, a winter simulation study was performed in a controlled laboratory setting. Thirty
Chinook salmon approximately one year of age were obtained from the Salmon River
Hatchery in Altmar, New York. Fish were from Lake Ontario stocks spawned at the
hatchery. The fish were held in large, rectangular tanks (58 x 305 cm) with continuous
flowing (4-5 gal/m) 9°C well water. The size range of the 30 Chinook salmon was 323
mm-368 mm and 341 g -612 g at the beginning of the experiment. Fish were fed
commercial feed until the start of the study. Fish were randomly divided into three
groups and the first group was immediately processed to obtain muscle plugs, and carcass
samples at the beginning of the study. Then, water temperature was lowered
incrementally over a two week period from 90°C to between 4.5-4.8°C and no feed was
given to the fish during the remainder of the study. A subset of the fish was randomly

selected and processed after three months time and again after six months of exposure to

68

the cold water temperatures. We conducted proximate composition analysis on all the
fish in this winter simulation study using the same procedure as on the Lake Michigan
wild-caught fish.
Data Analysis

We examined the relationship between whole-fish lipid levels and a variety of the
other metrics of fish nutritional status. These metrics included Fulton’s Condition Factor
Index (K), hepatosomatic index (HSI), and the lipid and water content of whole-fish and

tissues. K uses fish length and weight to compare samples:

K=M
L3

where W is weight (g), L is length (mm), and c is a constant (105). HSI was determined

by dividing the weight of the liver by the weight of the entire carcass. We also examined
the relationship between whole-fish lipid levels and proximate composition analysis
measures from individual tissues using Pearson correlations. Percentage data were
evaluated after an arc sine square root transformation.

Once good predictors of whole-fish lipid level were determined, we tested
whether the small fish samples collected in the spring showed a different pattern than the
rest of the samples collected in the spring using ANCOVA. We also examined the
relationship between whole-fish lipid levels and proximate composition analysis

measures from individual tissues for the small fish samples collected in the spring.

69

Data from the winter simulation study were evaluated using a one-way AN OVA
to determine if the different treatments (initial, after 3 months, after 6 months)
significantly affected measures of lipid and water.

Adequate sample size required for firture monitoring of nutritional status was
determined based on the variability in measures of nutritional status. We determined the
sample sizes required such that the expected size of the confidence interval would be 2d,
using the following equation:

d2
where n is the required sample size, ta is the appropriate value from the Student’s t

distribution, s is the observed standard deviation from a sample of observations, and d is
the specified error (Krebs, 1989). We calculated sample sizes for two absolute
differences (d), which were set equal to 1% and 0.75% of the least squares mean.
However, we believe that mean values for these nutritional indicators are not necessarily
the best indicators of population status. We consider the proportion of fish in a
population whose nutritional state exceeds a critical value deemed indicative of high
stress to be a more informative statistic. We determined the sample size required so that
an estimated proportion will fall within d of the true population proportion 95% of the
time using:

n=[p-(I—pii96].
d2

70

where p is the proportion of fish that are in the critical range for nutritional stress. Our
use of this equation assumes that sample sizes will be large enough so that the estimated
proportion will approximately follow a normal distribution with a standard deviation
based on the fact that the number of fish in the critical range will follow a binomial
distribution. We determined sample size for three values of desired absolute error (d) , d
set to 40% of p, 50% of p, and 60% of p.

Results

A total of 345 samples were collected during the six sampling periods. More
samples were collected from the west side of Lake Michigan than the east side. In
addition, more fall samples were collected than spring samples and the medium size class
had more samples than either the small or large size classes (Table 3.1).

Linear regression on all 345 samples showed that whole-fish lipid was
significantly correlated with whole-fish caloric content (positively) (r2 = 0.78), and with
whole-fish water (negatively) (r2 = 0.68) (Fig. 3.1). The relationship between whole-fish
lipid and Fulton’s condition factor (r2 = 0.03) Was very poor (Fig. 1). Whole-fish lipid
was not related to HSI either (r2 = 0.008) (Fig. 31). There was also no significant
relationship observed between whole-fish lipid and whole-fish protein (r2 = 0.09) (Fig.
18). Lipid content in the muscle tissue was positively correlated with lipid content in the
whole-fish (r2 = 0.61) and negatively correlated with water content in the muscle (r2 =
0.50)(Fig. 3.2). Lipid content in the liver was not significantly correlated with whole-fish

lipid content (r2 = 0.14) or water content in the liver (r2 = 0.01) (Fig. 3.2).

71

Results of the heterogeneity of slope test to determine whether small fish
collected in the spring were significantly different from the rest of the spring samples
were non-significant for both lipid content in the muscle (F = 0.95, df = l, p = 0.331) and
water content in the muscle (F = 0.04, df = 1, p = 0.84). Sample sizes for small fish
collected in the spring were small (n=17), but the relationships are not different from the
medium and large size classes. Samples from the medium and large size classes were
included in the ANCOVA if their values for the predictor variable (either lipid content in
the muscle or water content in the muscle) fell within the range observed for the small
size class.

Further, correlation analysis results using only samples from the small size class
collected in the spring showed that the whole-fish lipid level remained positively
correlated with lipid level in the muscle (r2 = 0.75) and negatively correlated with muscle
water content (r2 = 0.70). For the small, spring-caught samples there was no significant
correlation between whole-fish lipid content and either lipid content in the liver (r2 =
0.02) or water content in the liver (r2 = 0.05)

Based on the observed among-fish variability in the small fish collected in the
spring, a sample size of 40 individual fish should provide an estimated mean that fall
within 0.75% of the true mean 95% of the time for measurements of whole-fish lipid or
water content in the muscle (Table 3.2).

The winter simulation laboratory study showed that mean lipid levels decreased
significantly and progressively from levels seen at the start of the experiment to below

43% of initial levels after 3 months of treatment and below 20% of initial levels after 6

72

months of treatment (Fig. 3.3a). Similarly, the water content in the muscle increased
significantly and progressively with time. The initial mean water content in the muscle
tissue was 75.6%, and this increased to 78.9% after 3 months and to 80.9% after 6
months (Fig. 3.3b).

In addition to calculating the sample size necessary to estimate the mean of the
population, we also calculated the sample size that would be necessary to determine the
proportion of the fish whose value exceeded a critical threshold. This calculation
depends on the proportion of fish in the population being sampled that are actually
exceeding the threshold and on the desired absolute error. As one example, our analysis
shows that if the true proportion of fish that exceeded the critical threshold was 0.20, then
a sample size of 60 would provide estimates of the proportion precisely enough so that
they would fall between 0.1 and 0.3 (ie. d set to 50% of p) 95% of time (Fig. 3.4). When
the allowable absolute error is increased, sample sizes are reduced and if the allowable
absolute error is decreased, necessary sample sizes are larger (Fig. 3.4).

Discussion

The approach we adopted in this study reflects our intent to develop an efficient
and inexpensive method of monitoring the nutritional status of Chinook salmon
populations in Lake Michigan. We feel that whole-fish body lipid content provides a
good indication of an individual’s nutritional state, but analyzing samples for whole-fish
body lipid content is expensive and time consuming. Entire carcasses must be obtained,
transported, and kept frozen until band-sawing and homogenization. Obtaining carcasses

can be difficult since this species is not well sampled by surveys from all areas of the

73

lake and anglers are generally unwilling to part with the fish they catch. The freeze-
drying and lipid analysis are also time consuming and require expensive lab equipment
and supplies.

Finding an alternative metric which would inform us about the nutritional state of
a population without incurring the high cost and time requirements of whole-body lipid
analysis would be beneficial. Condition factor and HSI are sometimes used as a
representation of nutritional state because only the weight of the liver and weight and
length of the sample are needed, which can easily be measured during biological
sampling of either sport fishery harvest or survey catches. Although condition factor has
been found to be significantly correlated with body composition in Atlantic cod (Gadus
morhua) (Lambert and Dutil, 1997; Foster et al., 1993) and lake herring (Coregonus
artedi) (Pangle and Sutton, 2005), we found that condition factor is not a good indicator
of body composition for Chinook salmon and we found direct measure of lipid or water
to be much more valuable (Fig. 3.1). Trudel et al. (2005) reported a similarly poor
relationship between condition factor and energy content for Pacific Chinook and Coho
salmon.

The other whole-fish measures we considered were protein content, caloric
content and water content. Both caloric content and water content showed significant
correlation with whole-fish lipid content and would perform well as surrogates to
measuring whole-fish lipid content (Fig. 3.1), but these measures still require collection,

transportation, and homogenization of the whole carcass.

74

To avoid these drawbacks, we also looked at the correlation between muscle
tissue samples or liver tissue samples and whole-fish lipid content. In addition to being
easier to transport and process, such samples are easier to obtain while collecting
biological samples from the sport fishery at cleaning stations. Liver tissues did not show
a clear relationship to whole-fish lipid contents and we do not reconunend them as
surrogates. On the other hand, there was a strong relationship between whole-fish lipid
content and both the lipid content and the water content of the muscle tissue (Fig. 3.2).

A recent study in the Pacific by Trudel et al. (2005) which focused on predicting energy
density and lipid levels in juvenile coho and Chinook salmon using other metrics, found
that water content of the muscle tissue explained 65% of the variance in lipid levels. The
fish sampled in this study were approximately an order of magnitude smaller than in the
Lake Michigan study, ranging from 0-1200 g. Their lipid contents were also smaller,
ranging from 0-8% per gram wet mass. Other studies have reported a similar significant
negative relationship between lipids and water in other species (Plante et al., 2005; Cox
and Hartman, 2005; Hendry et al. 2000). Analysis of water content in muscle tissue is
clearly the most inexpensive and time efficient analysis to perform, among the proximate
composition analysis surrogates we considered. Therefore, we recommend focusing on
water content in the muscle plug for future monitoring efforts. We make this
recommendation because (1) it will make it easier to obtain large numbers of samples;
(2) processing and analytical costs will be lower; (3) the relationship between muscle
water content and whole-fish lipids is statistically significant, and is particularly

diagnostic when water content is relatively high or low; (4) the relationship holds for

75

small fish captured in the spring, and (5) the objective of monitoring is to detect trends at
the population level, not the individual level.

This proposed monitoring approach seeks to look at the overall nutritional health
of the Chinook salmon population in Lake Michigan. To do this, we recommend
focusing on the segment of the population which is most likely to experience nutritional
stress. In another study of this population we showed that small Chinook salmon
collected in the spring had significantly lower lipid levels than any other size class or
season combination (Peters et al., in review), and we recommend focusing on this
segment of the population for monitoring. The relationship between water content in the
muscle and lipid content in the whole fish is highly correlated (r2=0.70) within this
segment of the population. If practical considerations make it impossible to sample
small fish in the spring, we would recommend collecting immature, medium sized
samples in the spring as the relationship between water content in the muscle and lipid
content in the whole fish is similar. In our study, the sample size for small fish in the
spring was relatively small, but we were not specifically targeting this group and the
collection protocol could be changed by the management agencies to increase numbers
of small fish collected. We also found that for small salmon, lipid content varied by
geographic location in Lake Michigan (Peters et al., in review). However, this location
difference was only evident in fall samples. We recommend collecting Chinook salmon
from locations throughout the lake, but if practical restrictions prevent sampling from
both the eastern and western side of the lake, we feel the data collected would still be

usefirl and informative.

76

Another crucial part of this proposed monitoring program is the decision of what
values will be reported and used as an early indicator of nutritional stress in the Lake
Michigan Chinook salmon population. We recommend that the mean value of water
content in the muscle tissue for small fish collected in the spring be reported. This
measure would provide a useful indicator of changing nutritional status that should be
sensitive to changes over a wide range of mean lipid levels. We also recommend
reporting the estimated proportion of small fish in the spring whose nutritional status
exceeds a threshold value. To determine this threshold value, data from the winter
simulation study were used. These data show that the mean value for whole-fish lipids
after 3 months of starvation fell from 6.9% to 2.45%. Based on these laboratory results,
we set the threshold value at 2.45% whole-fish lipid content. A critical threshold for
water content in the muscle of 78% was determined by selecting a value for which the
proportion of samples exceeding it was similar to the proportion of samples showing
lipid levels below 2.45%. The fish from the winter simulation study had mean lipid
content at the beginning of the study (6.9%) only slightly higher than the mean value for
the small fish in the field study collected in the fall (6.1%). We then examined how well
the estimated proportion of the population that exceeded the 78% water content in the
muscle corresponded to the proportion of the population that had less that 2.45% whole-
fish lipid levels by plotting these proportions for the 18 combinations of size class and
period of collection (Fig. 3.5). These proportions were highly correlated (r2=0.94,

p<0.0001) and closely followed a 1:1 line.

77

In conclusion, our study has yielded several important insights of practical
significance to fishery management agencies with responsibilities on Lake Michigan.
We recommend that management agencies adopt a Chinook salmon monitoring program
that:

1. measures water content of the dorsal muscle plug as an indicator of
nutritional status

2. obtains samples in spring
3. focuses on immature individuals in the small length category (<<490mm)
4. samples a minimtun of 40 individuals per sampling occasion

5. reports on the mean water content in muscle tissue and the proportion the
sampled fish with over 78% water content in the muscle tissue.

We hope that this study will prompt continued research on energy dynamics of
Chinook salmon in the Great Lakes. In the future, newer technologies such as
bioimpedence and microwave meters might be useful for bioenergetics studies (Crossin
and Finch, 2005; Cox and Hartman, 2005). Additional research is recommended to
establish a longer time series of nutritional condition for this species. An understanding
of the environmental factors, both biotic and abiotic, that determine Chinook salmon
nutritional status would be gained with such a data set and allow for management
decisions that could prevent future collapses of this population. Finally, future work
should also focus on elucidating the relationship between nutritional stress and increased
mortality or risk of disease in Chinook salmon.

This type of nutritional stress monitoring program would be applicable to

Chinook salmon populations in the other Great Lakes although lake-specific calibration

78

may be necessary because our preliminary data for Lakes Huron and Ontario suggest that
the relationship between lipid and water is not the same in all the lakes. We also feel that
similar monitoring programs could be effective for a range of fish species which show
fluctuations in energy and are susceptible to nutritional stress. Plante et al. (2005)
suggested water content in the muscle be used to predict energy reserves in winter
flounder (Pseudopleuronectes americanus). When condition factor or other indices are
not good indicators of energy content (e. g.Trudel et al. 2005; Sutton et al., 2000;
Copeland et al. , 2004), following a protocol similar to the one we have developed could

be an efficient method for monitoring changes in energy content.

79

References

Adams, SM. 1999. Ecological role of lipids in the health and success of fish
populations. In Lipids in freshwater ecosystems, eds. M.T. Arts and BC.
Wainman, pp. 132-160. New York, NY: Springer-Verlag.

Adams, S.M., Breck, J .E., and McLean, RB. 1985. Cumulative stress-
induced mortality of gizzard shad in a southeastern US. reservoir.
Environmental Bio. Fishes. 13:103-112.

AOAC. 1995. Oflicial Methods of Analysis. 16th ed. Assoc. Offic. Anal. Chem,
Arlington, VA.

Bence, J .R., Smith, K.D. 1999. An overview of recreational fisheries of the
Great Lakes. In Great Lakes fisheries policy and management: a binational
perspectiv., eds. W.W. Taylor and GP. Ferreri, pp. 259-306. East Lansing,

MI: Michigan State University Press.

Benjamin, D.M. and Bence, J .R. 2003. Spatial and temporal changes in the
Lake Michigan Chinook Salmon fishery, 1985-1996. Michigan Department
of Natural Resources, Fisheries Division, Fisheries Research Report 2065.

Copeland, T. and Carline, RF. 2004. Relationship of lipid content to size and
condition in walleye fingerlings from natural and aquacultural environments.
N. Am. J. Aquaculture. 66:237-242.

Cox, MK. and Hartman, K.J. 2005. Nonlethal estimation of proximate
composition in fish. Can. J. Fish. Aquat. Sci. 62:269-275.

Crossin, GT. and Hinch, S.G. 2005. A Nonlethal, Rapid Method for
Assessing the Somatic Energy Content of Migrating Adult Pacific Salmon.
Trans. Am. Fish. Soc. 134:184-191.

Flath, LE. and Diana, J .S. 1985. Seasonal energy dynamics of the alewife in
southeastern Lake Michigan. Trans. Am. Fish. Soc. 114:328-337.

Foster, A.R., Houlihan, DE, and Hall, 8.]. 1993. Effects of nutritional
regime on correlates of growth rate in juvenile Atlantic cod (Gadus morhua):

comparison of morphological and biochemical measurements. Can. J. Fish.
Aquat. Sci., 50:502-512.

Goede, R.W., Barton, BA. 1990. Organismic indices and an autopsy-based
assessment as indicators of health and condition of fish. In Biological
indicators of stress in fish, ed. S.M. Adams, pp. 93-108. Bethesda, MD:
American Fisheries Society.

80

Hendry, A.P., Dittman, AH, and Hardy, R.W. 2000. Proximate
Composition, Reproductive Development, and a Test for Trade-Offs in
Captive Sockeye Salmon. Trans. Am. Fish. Soc. 129:1082-1095.

Holey, M.E., Elliot, R.F., Marcquenski, S.V., Hnath, J .G., and Smith, K.D.
1998. Chinook salmon epizootics in Lake Michigan: possible contributing
factors and management implications. .1. Aquatic Animal Health. 10:201-210.

Johnson, DC. and J .G. Hnath. 1991. Lake Michigan Chinook salmon
mortality-1988. Michigan Department of Natural resources, Fisheries
Division, Technical Report 91-4, Lansing.

Jonas, J ., Kraft, C., and Margenau, T. 1996. Assessment of seasonal changes

in energy density and condition in age-0 and age-1 muskellunge. Trans. Am.
Fish. Soc. 125:203-210.

Kocik, J .F ., Jones, ML. 1999. Pacific salmonines in the Great Lakes basin. In
Great Lakes fisheries policy and management: a binational perspective, eds.
W.W. Taylor and CR Ferreri, pp. 455-488. East Lansing, MI: Michigan State
University Press.

Krebs, Charles J. 1989. Ecological Methodology. New York, Harper Collins
Publishers.

Lall, SP. 2000. Nutrition and Health of Fish. In Avances en Nutricion
Acuicola V., eds. L. E. Cruz-Suarez, D. Ricque-Marie, M. Tapia-Salazar, M.

A. Olvera-Novoa, and R. Civera-Cerecedo, pp. 13-23. Yucatan, Mexico:
Merida.

Lambert, Y. and Dutil, J -D. 1997. Can simple condition indices be used to

monitor and quantify seasonal changes in the energy reserves of Atlantic cod
(Gadus morhua)?. Can. J. Fish. Aquat. Sci. 54 (Suppl. 1): 104-112.

Madenjian, C.P., Elliot, R.F., DeSorcie, T.J. , Stedman, R.M., O'Connor,
D.V., and Rottiers, D.V. 2000. Lipid concentrations in Lake Michigan fishes:

seasonal, spatial, ontogenetic, and long-term trends. J. Great Lakes Res.
26:427-444.

Magnuson, J .J ., Meisner, J .D., and Hill, D.K. 1990. Potential changes in the
thermal habitat of Great Lakes fish after global climate warming. Trans. Am.
Fish. Soc. 125:821-830.

Morgan, I.J., McCarthy, I.D., and Metcalfe, NE. 2002. The Influence of

Life-History Strategy on Lipid Metabolism in Overwintering Juvenile Atlantic
salmon. J. Fish. Bio. 60:674-686.

81

Oliver, J .D., Holeton, GE, and Chua, KB. 1979. Overwinter mortality of
fingerling smallmouth bass in relation to size, relative energy stores, and
environmental temperature. Trans. Am. Fish. Soc. 108:130-136.

Pangle, KL. and Sutton, TM. 2005. Temporal changes in the relationship
between condition indices and proximate composition of juvenile Coregonus
artedi. J. Fish Bio. 66:1060-1072.

Pedersen, J. and Hislop, J .R.G. 2001. Seasonal variations in the energy
density of fishes in the North Sea. J. Fish Bio. 59:380-389.

Plante, S., Audet C., Lambert, Y. and Noue, J. 2005. Alternative Methods for
Measuring Energy Content in Winter Flounder. N. Am. J. Fish. Management.
25:1-6.

Sutton, S.G., Bult, T.P., and Haedrich, R.L. 2000. Relationship among fat
weight, body weight, water weight, and condition factors in wild Atlanic
salmon parr. Trans. Am. Fish. Soc. 129:527-538.

Trudel, M., Tucker, S., Morris, J .F.T., Higgs, D.A., and Welch, D.W. 2005.
Indicators of Energetic Status in Juvenile Coho and Chinook Salmon. N Am.
J. Fish. Management. 25:374-390.

82

Pefiod

FaHOO
Spr01
FaHO1
Spr02
FaHOZ

Spr03

Total

Sex

 

Location
West East
0 38
7 24
19 68
7 43
24 45
8 62
65 280

9 c3
25 13
13 18
51 35
30 20
42 27
31 39
192 153

Maturity Status Size Class*
Imm. Mat Sm. Med. Lg.
15 23 12 18 8
25 6 4 20 7
58 29 33 44 10
26 24 8 24 18
24 45 19 22 28
28 42 5 39 26
176 169 81 167 97

Total

38

31

87

69

70

345

* Fish were categorized as small if they measured <495mm in the spring or < 550mm in fall.

Fish were categorized as large if they measured >700mm in spring or >775mm in the fall.

Table 3.1. Summary of samples collected by location, sex, maturity status, and size

class.

83

Measurement Range of Mean Values Error Needed Sample Size

% lipid in whole fish 0.24%-7.53% i0.75% 39

% water in muscle 75.3%-80.4% i0.75% 19

Table 3.2. Sample size required for desired absolute error for two measures of proximate

composition for small, spring samples.

84

 

 

 

 

 

“E 16 _ °

3 14. . ° °

3 12 - ' z .

5 -. : . .

.6, 10 . o . ... 0": c:- o‘. ". .. o.

u 8 t 3"" 23': Far-1:19,: ' ° ° . '

3 - 12‘4” 1' *3". -' °

'9 6 ’ . 9 ’ Q..’ . O. .

.9, 00. . o ’0: 0’. ‘I. .. ‘0‘ .

'_l 4 L . 0:12.: .':.o.: : . ' o '0.

a? 2 . '. ‘ °. : . .
o .e ' 0 00% . .. 1 .
800 1000 1200 1400 1600 1800 2000 2200 2400 2600 2800 3000 3200

Calories per gram w hole-fish

 

 

 

 

 

18 - - w - - - - -

15 l ° .
f9 19)
h 14 ’ o o
3 12. , ' - °
3 o $ . :.. . o .
g 10 ’ ' ‘ ‘ 3"). . ' 0. o ' . . o
b ..' . . ’0'. .: : z . o .0 .0
8 8 - .0 . '; it: :0~: 0g: 3... o: ‘ ..'I. :9 . o 9 9
8. 6 . o . ; f . ."..°.ta :. . .’ o a...“ . :0 o.
D o . a": 0. a”.... .o.‘: . .. .
3 4 , . .M. .9 .:0 01“... o: . . ..o : .

o 0‘ .°. .. . . 3 . .

32 2 ’ . o .. ° ' . 0. ~ .

o . - .. O. f. O. . 2.ga A A . A+ A

0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7
Condition Factor (K)

20 - - - ~ 5 v r - - - - - e v - ~ + - -
4: 18l c)
“’3 16 . ' .
Q ’
g 14 - . . . <
3 12 . '. . - ‘
E .0 .0 .; . o
g” 10’ ‘ ~:‘}°.t 8.. 0"

o o .. : . f o 4

g 8 I . . o o f ofigfi \°' C o o O
:9 6 _ o . o 35. ’2..}.§:?.‘o.u.. o .
5- 4. .' o; ,, 2.33 A..." ‘p .. .
Be 2 r ° 0 g 0‘ :0 ‘0... . ° “

0 .0 '0'... 3 °

 

 

 

64 66 68 7O 72 74 76 78 80 82 84
% Water per gram w hole-fish

Figure 3.1. Scatterplots of whole-fish lipid content versus (a) caloric content r2=0.79,

p<0.001 (b) condition factor r2=0.07, p<0.001 (c) water content r2=0.68, p<0.001 (cont.)

85

20 vvvvvvvvvvvvvvvvvvafvwvvvvvfiv vvvvvvvvv

 

 

 

 

18- d)

.C

é? 16»

Q

'5 14. °

.c . .

3 12 ' Q

E O Q .8 . .

g 10 r 3 .0... (.0. . o. . . i

o O '

h 8 r : ?£:¢.?$.o . . .0. ° .1

a . ....:.i%$ '80.... . .

2 6L . . O 0"”..4.." I. . . .. ‘

.9” '.~¢s '4 o . . 1 o

.4 4r .0 . o a... . ’0 o O

o o . 0

°\ 2L 0 .. .. :.. .. . . Q ,4
0 AAAAAAA '— H '1: °

0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035 0.040
Hepatosomatic Index (HSI)

 

 

 

 

20 , f

18- e
5 . ) -
c": 16» J
m b
'6 14_ O O .
S o
3 12. o '
E .0 .. ’. 9‘

8
g10 . .... ... "{‘.. . .
' o.‘ . 5' .‘ .00

a”, 8. o o 0 . . .‘ §.. ”‘00 . 4
a '0 0.30 ‘w.. 2’.
u 6L ....0 .$ '00'0.:o‘. .
E. .0 .9 '0. on. ‘0'. . . '
._ 0 Q .‘ 0"... 0 .0 o
.1 4' 8.0 o o .00 .00. ’
°\° ’ .. O :. Q ‘

2h . . .0 . . O '1

0’ A 4. L - .-I.A0A..'. - '- A

12 14 16 18 20 22 24 26

% Protein per gram whole-fish

Figure 3.1 continued. Scatterplots of whole-fish lipid content versus (d) hepatosomatic

index r2=0.008, p<0.001and (e) protein content r2=0.09, p<0.001.

86

 

 

 

 

.: 18 a) .

1% 16-

0

'6 14. O . O

.c . Q

212- u '3 .

.0 ~... 0 ..

E 10’ ° . ° ’0 ..{..:.o .0... o...- .

S, 1.0.0‘Qo.’.. Q“: 0.. g. .

a, 8» .‘J, .I. .... 0

Q ' fl. ? 0.

'U 6.. ..o.

a :k...0 '0

3.: “aye-’- o.- -

\ D

° 253-. '
are
0 2 4 6 8 10 12 14 16 18 20 22 24

% Lipid per gram muscle

 

 

 

 

20 - , - - V - - - a
5 18, b) .
u".- 16 -
o l
T) 14 ~ ' ° 0
.C . . g
3 12 ' Q .
g . O egco.‘.‘ o...
L- 10 o 0 Q. 3'. .0. o. ’ . .
a i o . . o . o
'- 8 ‘ hm W ‘.'.. o o
d) i . .‘. ’0" .0 .° '0 0 . . o o . .
0. fl‘ .‘0 ' 0. $ 0 0 o
E 6 ' . .0 fi" 0 ° . o O
.9- !» . '«c' °
—‘ 4 ' .. ‘.oo .
O\° z. (:OQ .. . . O

2 ' . O. Q

01...... o .

0 2 4 6 8 10 12 14 16

°/0 Lipid per gram liver

Figure 3.2. Scatterplots of lipid content in whole-fish versus (a) lipid content in muscle

r2=0.61, p<0.001 (b) lipid content in liver r2=0.11, p<0.001 (cont.)

87

20 - - - - - - - - of - ViAfi

 

 

 

 

18? c)
'5) i
la: 16+-
0 r
6 Mi ° . °
.C i . o
z 12> - ' .

’ O. ' O.

0 o o '0 .u 0.

S 10* .0 .0 '.a’0.:°. .
2’ 8 . ,. ,.“'-“5‘-t':~s,- - ‘
a , . '8'” .3}: 011.. 1
U 6.. o o . .~:,‘X’O‘ .00
'Ei . é. fltgfl'o
._ ’ o ‘ 9 . I Q.
—.I 4 . .. x» ' Q
°\° . 0.. ‘

2 a“...

0 A A A A A 3:. "

60 62 64 66 68 70 72 74 76 78 80 82 84

% Water per gram muscle

 

 

 

 

20 - -- vvvvvvvvvvvvv Ar+ - - - - --
’ l

18» d) <
.5 . o
u? 16-
m D
15 14» '0'
.c i o . .
Z 12[ .‘0. ‘0. $ . 1

0

3 10+ . 3.0.53}. ,' 3. l
.7, a. . i :u-dtuog‘égu -~ .
Q ’ ' 3.1 ” '?a 4' ago' 0
E 6 . ° no 082‘fl:£.
.9- i . O ‘.. O
_l 4 . . .
O . . ..Q
°\ 2: ...::.I.. .

0' ' . of .'

55 60 65 70 75 80 85 90 95

% Water per gram liver

Figure 3.2 continued. Scatterplots of lipid content in whole-fish versus (c) water content

in muscle r2=0.50, p<0.001 and (d) water content in liver r2=0.01, p=0.19.

88

 

 

 

 

 

 

 

 

 

 

- Mean
77 I I MeantSE
.:
6‘9 6»
.9! l
.8 5
3 ’ l
E
:3 4»
a:
8 3»
.‘9
g .7 I
39
1 » I
O L
hitial 3 months 6 months
82
2 81’ <
O I
m —_
3
E 80+ .
E 79
g .
m I
m 73. .
a
9 77.
10
3 __
°\o 76 P
I
75 l __ .
74 -
Initial 3 Months 6 Months

Figure 3.3. Results from the winter simulation study showing means with standard errors
for (a) percent whole body lipid content for each treatment and (b) percent water in the

muscle tissue for each treatment.

89

500.-.. a -.--.z

 

—o— d = 50% Of p
400’ \ -.;- d=40% ofp
\ _ ..,-._.-.-A d = 60% Of P

300* ‘

200 ~

100-

 

Number of samples needed (n)

    

 

 

0 A + g - ‘ . ‘“ “1r“ , AMANJp ._ rum 7 ””7”!”
0.0 0.1 0.2 0.3 0.4 0.5

Proportion of the population in the critical range for nutritional stress (p)

 

Figure 3.4. Required sample size (n) to estimate the proportion of the population in the
critical range for nutritional stress when that true proportion in the population ranges
from 0.05 to 0.50. Sample size requirements are shown for three levels of desired
absolute error ((1) as 50%, 40%, and 60% of p. Lines illustrate the following example:
the true pr0portion of fish that exceeded the critical threshold, p, was 0.20 and the
estimate of the proportion of the population which exceeded the critical threshold was set
to fall between 0.1 and 0.3 (ie. dis set to 50% of p) 95% of the time, then a sample size

of 60 would be required.

90

0,3fi,--.--,--,-c,fc.-, no

 

0.7 .-
0.6 1
0.5 3.
0.4 i . 1‘
0.3 g l
0.2

ijw
AAL‘.—AA

0.1 _.

prop. of samples with >78% water
content In muscle

0.0 1 0

 

 

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
prop. of samples with >2.45% whole-fish lipid

 

Figure 3.5. Comparison of the pr0portion of the population with water content in the
muscle above the threshold of 78% with the proportion of the population with whole-fish
lipid content below the threshold of 2.45% for all combinations of size class and

collection periods (r2=0.94). A 1:1 Line is provided for reference.

91

CHAPTER 4
OPTIMAL ENERGY ALLOCATION FOR CHINOOK SALMON, ONC ORHYNCH US
TSA W YYSCHA, IN THEIR NATIVE AND A NOVEL ENVIRONMENT

Introduction

The amount of energy that a fish obtains and the allocation of that energy to
various processes can shape its life history decisions and its fitness. Bioenergetics
models have been used to calculate how consumption and growth are affected by a wide-
range of variables, including temperature, prey availability, and prey energy density
(Beauchamp et al. 1989; Railsback and Rose 1999; Galarowicz and Wahl 2003).
After meeting energy requirements for basic respiration and metabolic
processes, the organism can allocate surplus energy to increasing body size, energy
reserves or reproductive organs. The trade-off between allocating surplus energy to
growth versus maturation has been a major topic of discussion in life history theory (Roff
1992; Stearns 1992).

For many fishes, increasing body size early in life is important because it can
decrease predation risk and allow the fish to feed upon larger prey items (J onsson and
J onsson 1998). The cost of allocating all surplus energy to body growth, usually in the
form of protein, is that no energy would be allocated to energy storage in the form of
lipids (Rikardsen and Elliott 2000; Sutton et al. 2000; Morgan et al. 2002). While early
growth may be important to decrease vulnerability to predation, insufficient lipid reserves
can make fishes more likely to suffer mortality from disease or starve during periods of

low energy availability (Miranda and Hubbard 1994; Adams 1999). Over-winter

92

survival for many fish species has been shown to depend on adequate energy reserves
(Oliver et al. 1979; Henderson et al. 1988; Madenjian et al., 2000). Once fish begin the
maturation process, energy is allocated to gonad and gamete development and also to
secondary sexual characteristics, taking away from the energy that can be allocated to
somatic growth or lipid reserves.

Many parameters in bioenergetics models are dependent upon temperature and
size, thus the environment can greatly affect energy intake and utilization by fish. Also,
predictably changing environmental conditions can shape expectations of future energy
intake. Thus, in environments with large seasonal changes in prey availability or
temperature, fish would be expected to use an energy allocation strategy that accounts for
this variable environment. For example, many fishes increase lipid reserves before
winter, in anticipation of a period of cold temperatures and low food availability (J onsson
and Jonsson, 1998; Craig et al. 2000; Dempson et al. 2004). These factors can affect
energy allocation strategies. Garvey and Marschall (2003) investigated optimal energy
allocation strategies for largemouth bass, Micropterus salmonides, and found that
latitude, body size and consumption affected the Optimal energy allocation strategy.

Chinook salmon, Oncorhynchus tshawytscha, the focus of this study, provide an
interesting focus for a study of energy allocation strategies because of their mode of
reproduction and their geographical distribution. First, because Chinook salmon are
semelparous, the timing of maturation is especially critical. Second, Chinook salmon
inhabit regions with temperate climates and thus would be expected to experience

seasonal fluctuations in temperature and prey availability. Third, Chinook salmon have

93

been introduced into a wide-range of environments. In these novel environments, the
energy allocation strategy that was suited to their native environment may no longer be
optimal. One novel environment into which Chinook salmon have been introduced is
Lake Michigan. Chinook salmon were introduced through large-scale stocking efforts in
the late 19603 in order to help control the population numbers of the invasive alewife
(Alosa pseudoharengus) and create a recreational fishery (Hansen and Holey 2002;
Tanner and Tody 2002). After initial success, the population abundance of Chinook
salmon decreased substantially in the late 19805, probably due to increased mortality
stemming from nutritional stress and an epizootic of bacterial kidney disease (BKD), and
population abundance subsequently increased in the 19905 (Holey et al. 1998). This
fluctuation in abundance led to concerns that Chinook salmon in Lake Michigan were
energetically challenged because of the differences in enviromnental conditions between
their native environment and the Lake Michigan environment.

One useful approach for estimating energy allocation strategies is using a
dynamic programming model. These models are basically individual optimization
models of behavior and can be set up to look at how an individual would optimally
allocate energy to maximize some quantitative measure of fitness based on a set on initial
conditions and decision rules (Mangel and Clark 2000).

The goals of this paper are to (1) estimate optimal energy allocation strategies for
Chinook salmon in their native environment, the northeastern Pacific, and a novel

environment, Lake Michigan; (2) determine optimal age of maturation and (3) determine

94

whether Chinook salmon are allocating energy optimally in Lake Michigan, given the

assumptions underlying our modeling approach.

Methods

We developed a model that simulated energy intake, allocation of energy to three
compartments, and the effect of allocation on the growth and survival of Chinook salmon
from age 1 up to a specified age of maturation. The model then determined the ultimate
fitness, measured as egg mass. The optimization model varied the monthly allocation of
energy among lipid reserves, somatic growth, and (in the year of maturation) gonads. By
comparing alternate allocation schedules, this analysis allowed us to determine the energy
allocation strategy and the age at maturation that maximized the expected production of
eggs (mass of gonads). More specifically, we used a bioenergetics model (Stewart and
Ybarra 1991) to determine growth given a specified age of maturation and a set of
environmental conditions, and a dynamic programming model (Mangel and Clark 2000)

to determine the optimal allocation schedule for each possible age of maturation.

Bioenergetics Model

The bioenergetics model predicted the surplus energy available for allocation or
the energy deficit based on environmental temperature, fish mass, the proportion of
maximum consumption achieved, and salinity (either freshwater or saltwater). The

bioenergetics model worked on a daily time step to calculate the energy intake and costs

95

of metabolism for a female Chinook salmon to determine the energy surplus or deficit for

each day using the following equation:

Energy Surplus(or deficit) = Consumption — SDA — Egestion — Excretion — Respiration

where SDA is specific dynamic action and the units are calories per day (Kitchell et al.
1977). The models for both Lake Michigan and the Pacific were constructed identically
and used the parameter values from the Stewart and Ybarra (1991) model with a few
exceptions. Monthly temperature data were obtained from more recent archival data tag
work in Lake Huron (R. Bergstedt, United States Geological Survey, Hammond Bay
Laboratory, unpublished data) and the Pacific (D. Welch, Department of Fisheries and
Oceans, Nanaimo, British Columbia, unpublished data) (Fig 4.1). Chinook salmon in
Lake Michigan were assumed to have a similar thermal experience to Chinook salmon in
Lake Huron since the two lakes are connected. The Stewart and Ybarra model was
developed for freshwater, and metabolic costs are generally higher in saltwater.
Therefore, for the Pacific model, we increased metabolic cost by 15% to account for the
difference in salinity (Trudel et al. 2004). The percent of the prey items that were
indigestible and the energy content of prey was set to 9.0% and 1.28 kcal/g respectively
for Lake Michigan based on recent work by Madenjian et al. (2006) and 8.6% and 1.37
kcal/g respectively for the Pacific (Davis et al. 1998). We set annual instantaneous
mortality rates higher in the Pacific (0.3) compared with Lake Michigan (0.2) based on
survivorship data from the Pacific (Thedinga et al. 1998) and percent of stocked fish that
returned to the Strawberry Creek Weir on Lake Michigan (Peeters and Royseck 2003).

96

In our model, we determined annual proportions of maximum consumption (p-values) by
iteratively running the model (which finds the optimal allocation strategy for a specific
set of p-values) and adjusting the p-values until the model predictions of weight and
length were consistent with actual data for Chinook salmon collected in Lake Michigan
(Peters et al., in press) and in the Columbia Basin for a mixture of stocks at the
Bonneville Dam (Hooff et al. 1999a; Hooff et al. 1999b). For both models, the winter
proportion of maximum consumption was different than the rest of the year. For Lake
Michigan, we assumed that Chinook salmon do not eat during the winter, so the
consumption was 0 for the months of January, February, and March. This created a
period of three months when there was no energy intake for Lake Michigan Chinook
salmon and the model forced the energy deficit to be made up by mobilizing lipid
reserves. During the rest of the first modeled year (age one), proportion of maximum
consumption was set to 0.64. For the second year, the proportion of maximum
consumption during the feeding period increased to 0.68 and for years 3 and 4, the
proportion of maximum consumption increased to 0.72 to account for the larger size of
the fish and the increased metabolic demands. In the Pacific, juvenile consumption rates
have been shown to be reduced over winter, but are high enough to maintain positive
growth (Trudel et al. 1999). Based on this information, the proportion of maximum
consumption in the winter that allowed for positive growth was 0.33. For the rest of the
year, the proportion of maximum consumption which best fit known growth data was 0.6.
For the second year, proportion of maximum consumption was increased to 0.64 and in

years 3 and 4 it was increased to 0.68. Changing the proportion of maximum

97

consumption by more than 0.04 affected the growth of the modeled fish and the effect of

this change on the energy allocation strategy was evaluated using sensitivity analyses.
Other fixed constants input into the model included values for somatic tissue

energy density (0.74 kcal/ g), lipid energy density (9 kcal/ g) and energy density of eggs

(2.34 kcal/ g) (Patterson 2004).

Dynamic Programming Model

The dynamic programming model searched for the energy allocation strategy that
maximized the expected egg mass (the actual egg mass produced by a fish surviving to
reproduce multiplied by the probability of surviving to that stage) of a female Chinook
salmon. We viewed expected egg mass as a surrogate for fitness and refer to the energy
allocation strategy that maximized expected egg mass as the optimal strategy. Egg mass
at the end of the last time step is referred to as final egg mass. We found the optimal
strategies for environmental conditions experienced by Chinook salmon in two regions:
the north Pacific Ocean and Lake Michigan. Possible allocation strategies were defined
by the proportion of energy (above the energy needed for maintenance) that was allocated
to somatic growth, lipid reserves, and gonad development each month.

Allocation strategies were restricted by several allocation rules we assumed must
apply. First, when energy is allocated to somatic growth, lipid reserves, or gonad
development, it is permanent and can not be moved or reallocated. Second, energy
allocated to lipid reserves can be used to meet energetic needs during periods when the
metabolic costs are higher than the consumption, a situation which occurs in winter
months and during sexual maturation. Finally, if lipid levels dropped below 1%,

98

mortality occurred. Mortality was not size-dependent in this model. The bioenergetics
calculations were run on a daily time step, while the allocation of energy was done on a
monthly basis. The optimization considered monthly allocations to each component in
10% increments (e. g., 0%, 10%, 20%, etc), and compared all possible combinations of
allocations among the three components. The model started at the beginning of April
with an age 1 female Chinook salmon. Maturity was set at the beginning of each run of
the model to occur at either age two, three, or four, at the end of August. Therefore, fish
set to mature at age 2 were modeled for a total of 17 months, age 3 for 29 months, and
age 4 for 41 months. Gonads for immature female Chinook salmon constitute
approximately 0.55% of their body weight (A. Peters, Michigan State University,
unpublished data). Therefore, prior to April in the year of maturation, the model did not
allow fish to allocate any energy to gonad development, fish could only allocate energy
to somatic tissue or energy reserves. From April to August of the year of maturation, all
surplus energy could be allocated to gonad development and the energy needed for
metabolic costs could be supplied by mobilizing lipid reserves. Fitness in the last month
equaled the total mass of eggs. Since the model allowed energy to be allocated to eggs
beginning in April of the year of maturation, comparison of expected final fitness in April
were used to determine the optimal age of maturation.

To reduce computations, the model was constrained to limit the upper bound on
the proportion of body mass that could compose gonads or lipid reserves. The model
allowed the percent of the body mass composed of eggs to range from 0%-26%

(Campbell 2006). Data collected on 354 Chinook salmon in Lake Michigan showed a

99

range of lipid levels from 0.23% to 16.4% (A. Peters, Michigan State University,
unpublished data). Based on these data, the percent of the body composed of lipids was
allowed to range from 1%-16% of wet body weight.

Preliminary analysis showed that initial lipid levels made little difference in
determining optimal allocation strategy or the expected lifetime egg production. For
example, for a range of lipid levels from 1% to 10% for the Lake Michigan fish that
matured at age 3, the final egg mass only ranged from 1553 to 1556. In subsequent
analyses we set initial lipid level to 3%.

In addition to the expected egg mass at maturity, for each time step (month) the
model computed the percent of the whole fish composed of lipid, the percent of the
whole fish composed of eggs, the surplus energy available for allocation, and the
proportion of energy allocated to somatic growth, lipids or gonads.

After optimal energy allocation strategies were determined, simulations were run
to see how sensitive final egg mass was to changes in the model. Simulations were run to
explore how changes to the maximum proportion of consumption affected the optimal
energy strategy and the final egg mass. Also, simulations were run using the optimal
energy strategy under different temperatures. Finally, we looked at how small deviations
from the optimal allocation strategy affected final values. We ran a simulation in which
10% more of the available energy was allocated to somatic grth at the expense of
allocation to lipid reserves, compared to the optimal strategy. If the optimal somatic
allocation for a particular month was 100%, it was not changed. Then, we simulated the

opposite deviation, in which 10% more energy was allocated to lipid reserves at the

100

expense of somatic growth. We did not change the energy allocation for the final
months of the model when energy could be allocated to gonads because all of the model
runs showed 100% of the energy being allocated to gonads during the final months.

We computed the optimal energy allocation strategy for ages of maturity of 2, 3,
and 4 years. To determine the optimal age at maturation, expected final egg mass values
were compared between models with different ages of maturation. Comparisons were
made between the same time step for the different ages of maturation. For example, to
compare optimal age of maturity for a fish which matures at age 2 with a fish which
matures at age 3, we compared the expected final egg mass in April of year 2 for both the
age 2 and age 3 optimal strategy. Expected final egg mass (E17) for all months prior to
the last month was calculated as the product of Eff for the next month and the probability
of surviving that month.

Eff! = StEfltH
We also considered how mortality affected optimal age at maturation and final egg mass
by calculating Efl under annual mortality rates of 0.1, 0.2, 0.3 and 0.4.

Finally, we wanted to explore whether Chinook salmon in Lake Michigan allocate
energy in a way that corresponds closely to what we have estimated is optimal given the
model assumptions. We first compared mean lipid levels from Chinook salmon collected
in Lake Michigan in spring and fall from 2000 to 2003 (Peters et al. in press) to lipid
levels predicted for the optimal strategy in Lake Michigan. Second, we performed a
virtual transplant of a Pacific Chinook salmon into the Lake Michigan environment

(Munch and Conover 2002; Garvey and Marschall 2003). Specifically, we used the

101

optimal energy allocation strategy estimated by the model for a Chinook salmon in the
Pacific and grew the fish under the environmental conditions modeled for Lake

Michigan.

Results
Optimal energy allocation and model sensitivity

The optimal energy allocation for Chinook salmon in the Pacific and in Lake
Michigan differed. In the Lake Michigan optimal strategy for age 3 fish, all energy was
allocated to lipid reserves in the first month and then for the next six months a majority of
surplus energy was allocated to somatic growth (Table 4.1). In the months of November
and December more energy was allocated to lipid than to somatic growth. In the second
year, allocation to lipids was above 0.5 in July and then in October, November and
December, before feeding stopped in the winter (Table 4.1). There was no energy for the
fish modeled in the Lake Michigan environment to allocate during the winter months.
Beginning in April of the final year, all surplus energy was allocated to gonads. For a
fish modeled under the conditions of the Pacific environment and set to mature at age 3,
the first two months showed high allocation to somatic growth and then during the
months of June and July, there was a high amount of energy allocated to lipids (Table
4.1). During the first winter, the fish modeled under Pacific conditions put almost all of
their available energy into lipid reserves. As with the Lake Michigan individuals,
beginning in April of the final year, all surplus energy was allocated to gonads. In both

winters, Pacific fish had some energy available for allocation whereas the Lake Michigan

102

fish had an energy deficit (Fig. 4.2). Overall, the Lake Michigan fish experienced more
variation in the amount of energy they had to allocate or needed to make up by
mobilizing lipid reserves than the fish modeled in the Pacific environment (Fig. 4.2). The
final egg mass and percent of body weight composed of gonads were also different
between the fish modeled in the Lake Michigan environment and fish modeled in the
Pacific environment (Table 4.2). The fish modeled in the Pacific environment had a final
egg mass of 1700 grams compared to 1556 grams for the fish in the Lake Michigan
model (Table 4.2).

Final body weight was relatively similar for fish modeled in both environments
(Table 4.2). The optimal Lake Michigan strategy showed slightly higher body weights
than under the optimal Pacific strategy for all time steps except the first three months and
the final four months (Fig. 4.3). Lipid levels differed considerably between the fish
modeled in both environments. Lipid levels, measured as a percent of the total body
weight, were higher in the fish modeled under the Pacific conditions for 27 of the 29 time
steps (Fig. 4.4).

For fish modeled in both the Pacific and the Lake Michigan environment, altering
the optimal energy allocation strategy to one with 10% more energy allocated to somatic
growth than under the optimal strategy decreased final egg mass more than allocating
10% more energy to lipid reserves (Table 4.3). For the fish modeled in the Pacific
environment, increasing allocation to somatic grth by 10% decreased final egg mass to
1546 g, from the optimal of 1700 g (Table 4.3). The final egg mass which resulted from

allocating 10% more energy to lipid was 1655 g. For the fish modeled in Lake Michigan,

103

the increased allocation to somatic growth resulted in larger fish, but only 1515 g of eggs
at maturation, compared to 1556 g of eggs in the optimal strategy and 1536 g of eggs in
the strategy with an increased allocation to lipids (Table 4.3).

Fish that allocated 10% more energy to somatic growth than the optimal strategy
throughout the model, ended up with less energy available for allocation during the
second year. In the Pacific model, the modeled fish actually had an energy deficit during
the final winter months, whereas both the fish using the strategy which was found to be
optimal and those using the strategy which allocated 10% more energy to lipid reserves
than the optimal strategy still had an energy surplus in the winter months.

The model showed some sensitivity to the starting values for proportion of
maximtun consumption (p) used in the model. As would be expected, when p is
increased, the size of the modeled fish was larger and the final egg mass was also greater.
For the fish modeled under Lake Michigan conditions, final egg mass increased from
1342 g for a p value of 0.6 to 1770 g for a p value of .68. The optimal energy allocation
strategies also showed some variation depending on p values for the fish modeled under
Lake Michigan conditions. During the first 6 times steps and the final 10 time steps, the
optimal strategy is the same for the starting p values ranging from 0.6 to 0.66. The fish
that were modeled under a p value of 0.68 allocated more energy to lipid during the
second and third months than the fish modeled under the smaller p values (Figure 4.5)

For the fish modeled under Pacific conditions, variation in p values caused similar
changes in growth and expected egg mass. The optimal allocation strategy for fish

modeled with p values of 0.55 for the spring, summer, and fall months and 0.33 for the

104

winter months had a final egg mass of 1393 g. When the p values were increased to 0.65
for the spring, summer, and fall months and kept constant at 0.33 for the winter months,
the optimal energy allocation strategy resulted in a final egg mass of 2008 g (Table 4.5).
Most of the change in final egg mass was a result of final body size since the percent of
body mass composed of gonads changed only slightly for the range of p values, from
23.2% to 23.9%. Fish that were modeled with lower p values, had a faster initial
increase in percent lipids than the fish that were modeled with higher p values, although
by the end of the simulation, the fish with higher p values had higher final percent lipid
levels (Figure 4.6).

For the fish modeled under Lake Michigan conditions, when the optimal energy
strategy was used and temperatures were decreased by temperatures by 1° C for each
month, the resulting fish were larger, had more energy available for allocation, and a
higher final egg mass (1601 g) than the fish modeled under the temperature regime
obtained from the Lake Huron data tags (1556 g). Increasing temperature by 1° C for
each month did not change final egg mass (1556 g), but the resulting fish were smaller.
The pattern was similar for the fish modeled under Pacific conditions, in that decreasing
temperatures by 1° C for each month, resulted in larger fish which had more energy
available for allocation and a higher final egg mass (1759 g) than the fish modeled under
the temperature regime for the Northern Pacific (1700 g). For the fish modeled under
conditions with temperatures increased by 1° C each month, final egg mass decreased

substantially to 1557 g.

105

Optimal age at maturation

Under the conditions of the Lake Michigan model, expected final egg mass was
higher in April for fish maturing at age 3 when mortality was set to 0.1 or to 0.2 (Table
4.4). When mortality was increased to 0.3 or to 0.4, the expected final egg mass was
higher in April for fish maturing at age 2. For the fish modeled in the Pacific
environment, the expected final egg mass was higher in April for fish maturing at age 3
under mortality rates of 0.1, 0.2, and 0.3, but when mortality was increased to 0.4, the

expected final egg mass was highest at age 2 (Table 4.5).

Virtual transplant of Pacific strategy to Lake Michigan

When the optimal energy allocation strategy for fish modeled in the Pacific was
employed under Lake Michigan conditions, final egg mass decreased substantially from
1556 g to 1230 g for a fish which matured at age 3. Weight and length were also lower in
the virtually transplanted fish compared to the Lake Michigan fish (Table 4.2). The
virtually transplanted fish grew slowly and initially put a large amount of energy into
lipid reserves (Figs. 4.3 and 4.4).

Data for lipid levels of Chinook salmon that were actually sampled in Lake
Michigan during 2000-2003 were available for age 1, 2, and 3 fish in June and August
(Peters eta1., unpublished data). The lipid levels in the modeled Lake Michigan Chinook
salmon were closer to the lipid levels in the actual Lake Michigan fish than to the lipid

levels from the Chinook salmon modeled under Pacific conditions (Table 4.6). In the

106

first five time periods, the lipid levels for the fish modeled in Lake Michigan were lower
than the lipid levels from fish actually collected from Lake Michigan. In the final time
step, the lipid levels in the fish modeled in Lake Michigan were higher than the actual
lipid levels of Lake Michigan samples. The fish modeled under Pacific environmental
conditions had higher lipid values than either fish modeled under Lake Michigan

conditions or fish actually sampled from Lake Michigan in all time periods (Table 4.6).

Discussion

The model predicted different optimal energy allocation strategies for Lake
Michigan and the Pacific suggesting that the differences in temperature and consumption
are sufficient to affect how Chinook salmon allocate energy to maximize fitness. In both
cases, changes in the proportion of maximum consumption (p) had a relatively large
effect on final egg mass compared to changes in temperature. The qualitative result that
fish modeled in the Pacific environment put more of their energy available for allocation
to lipid reserves than the fish modeled in Lake Michigan was robust, however, as this was
true for all temperature regimes and p values we considered. Under the model
assumptions, temperatures were, on average, higher in the Pacific environment and there
was an increased metabolic cost in the model of 15% associated with living in a saline
environment. Therefore, for the fish modeled in the Pacific environment, rapid, early
grth resulted in high metabolic costs and decreased final egg mass. The fish modeled
in the Pacific environment invested more into lipids to avoid fast grth and the

accompanying increases in metabolic costs. The Pacific simulations that increased

107

energy allocation to somatic growth by ten percent and resulted in a decrease in final egg
mass which illustrates this cost. The fish modeled in Lake Michigan did not have the
increased metabolic cost of being in a saline environment and experienced cooler
temperatures, on average, so the optimal strategy given their parameters was to invest in
growth early.

Overall, both the temperature and the amount of energy available for allocation
are more variable in Lake Michigan than in the Pacific (Fig. 4.1 and Fig. 4.2). The
introduction of an organism into an environment that is more variable than its native
environment could result in energetic stress. One might predict that in such a variable
environment, it would be better to maintain a larger reserve of energy (lipids). Indeed,
Garvey and Marschall (2003) found that largemouth bass in northern environments that
experience large seasonal variation allocated more energy to lipids than largemouth bass
in southern environments with relatively constant energy availability and temperature. In
contrast, the modeled Lake Michigan fish grew faster than the modeled Pacific fish
during the first 16 months. If the models are accurately capturing the true energy
allocation strategies of Pacific and Lake Michigan fish, then it seems somewhat risky that
in an environment that varies as much as Lake Michigan, the lipid levels are lower in the
fish modeled in Lake Michigan than those of the fish modeled in the Pacific (Fig. 4.4).
This contrast could be an artifact of the way we modeled the effect of lipid depletion on
mortality or it be a bioenergetic consequence of Lake Michigan Chinook salmon dealing

with the novel environment.

108

In Lake Michigan and the Pacific, the observed age at maturity can range from 2-
6 years for female Chinook salmon. Our model predicted age 3 to be the optimal age at
maturity based on initial conditions for both Lake Michigan and the Pacific. The optimal
age at maturity was sensitive to changes in mortality rates. An increase in mortality rates
led to a decrease in optimal age at maturation. This finding fits with life history theory
which says that maturation should only be delayed when the risk of waiting for another
year to mature is offset by a large increase in fitness (Stearns, 1992). In this model, when
annual mortality is increased to 0.4, the high risk of mortality leads to a model prediction
for the optimal age of maturity to be age 2 in both environments.

The model produced two different optimal strategies, one for the environmental
conditions set for Lake Michigan and another one for the environmental conditions set
for the Pacific. If the differences in the modeled environments and the assumptions about
growth and the rules for energy allocation are valid, we can use the optimal strategy for
the Pacific environment to see how a fish adapted to the Pacific would grow in Lake
Michigan environmental conditions. The low final egg mass and size of the virtual
transplant shows that the optimal strategy estimated by the model for the Pacific
environment is a poor strategy for Lake Michigan environmental conditions. If Lake
Michigan Chinook salmon had not altered their energy allocation strategy, we would
expect them to have the same values as the transplant. The transplant and the fish
modeled in Lake Michigan using the Lake Michigan optimal strategy have quite different

final egg mass and lipid levels throughout the three years (Table 4.6). This outcome

109

supports the idea that Lake Michigan Chinook salmon have altered their energy
allocation strategy to better suit their novel environment.

The lipid values for the fish following the optimal energy allocation strategy in
the Pacific increase from 3% at time zero, to 5.5% in June and to 7.4% by August of the
first year. Data collected by Trudel et al. (2005) on 104 juvenile Chinook salmon in the
Pacific had lipid levels that ranged from 1-8% for a Chinook salmon from 200-1200
grams. The predicted lipid levels for Pacific Chinook salmon using the optimal strategy
are within this range. While this limited comparison is intriguing, data on actual lipid
levels for oceanic stage Pacific Chinook salmon larger than 1200 grams would provide a
better test of whether Pacific Chinook salmon allocation is more closely in accord with
our estimate of the optimal allocation strategy for that environment.

Future work on this model could involve two major changes. First, we could
change the mortality rules. Specifically, instead of having a knife-edge rule of mortality
occurring only when lipids drop below 1%, we could have a probability function which
has high probability of mortality with lipids close to 1% and declining probability of
mortality as lipids increase to 5%. Second, the current model does not allow for lipid
reserves to be converted into gonads, they are only drawn upon to satisfy metabolic costs.
Since final lipid values are above 1% for the optimal strategy in both the Pacific and in
Lake Michigan, allowing any extra lipids to be converted into gonads during the year of
maturation would increase final egg mass. However, because egg mass is already
approximately 23% of the body weight of the fish in the model, allowing lipids to be

converted into gonads might results in unrealistically large percentages of body weight as

110

gonads since data from Lake Michigan Chinook salmon showed a maximum of 21%
body weight in gonads. (A. Peters, Michigan State University, unpublished results).
Furthermore, the model does not currently include migration or spawning costs which
could draw upon lipid reserves.

This combination of a bioenergetic and dynamic programming model is especially
useful for understanding the trade-offs that fish are making under different conditions and
can inform us about how energy allocation affects fitness and life history traits. These
models can help to explain native distribution of organisms by showing how temperature,
length of growing season, and consumption can constrain growth or affect allocation and
timing of reproduction (Garvey and Marschall 2003). The ability to perform virtual
transplants is a powerful tool to evaluate energetic constraints for introduced species and

even to inform biologists about what species might be more likely to colonize a new area.

111

References

Adams, S. M. 1999. pp. 132-60 in, Lipids in freshwater ecosystems. M. T. Arts, and B. C.
Wainman, Editors. New York: Springer-Verlag.

Campbell, B., B.R. Beckman, W.T. Fairgrieve, J .T. Dickey, P. Swanson. 2006.
Reproductive Investment and Growth History in Female Coho Salmon. Trans Am Fish Soc
135: 164-173.

Craig, S.R., MacKenzie, D.S., Jones, G., and Gaitlin, D.M. 2000. Seasonal changes in
the reproductive condition and body composition of free-ranging red drum, Sciaenops
ocellatus. Aquaculture 120: 89-102.

Dempson, J .B., Schwarz C.J., Shears, M. and Furey, G. 2004. Comparative proximate
body composition of Atlantic salmon with emphases on parr from fluvial and lacustrine
habitats. J. Fish Bio. 64: 1257-1271.

Garvey, J .E. and L. Marschall. 2003. Understanding latitudinal trends in fish body size

through models of optimal seasonal energy allocation. Can J. Fish. Aquat. Sci. 60 : 93 8-
948.

Hansen, M.J. and ME. Holey. 2002. Ecological Factors Affecting the Sustainability of
Chinook and Coho Salmon Populations in the Great Lakes, Especially Lake Michigan.
Pages 155-180 K.D. Lynch, M.L. Jones and W.W. Taylor, editors Sustaining North
American Salmon: perspectives across regions and disciplines. American Fisheries
Society, Bethesda, Maryland.

Henderson, P. A., R. H. A. Holmes, and R. N. Bamber. 1988. Size-selective Overwintering
Mortality in the Sand Smelt, Atherina-boyeri risso, and its Role in Population Regulation.
Journal of Fish Biology 33: 221-33.

Holey, M. E., R. F. Elliot, S. V. Marcquenski, J. G. Hnath, and K. D. Smith. 1998. Chinook
salmon epizootics in Lake Michigan: Possible contributing factors and management
implications. Journal of Aquatic Animal Health 10: 201-210.

Hooff, R.C., Fryer, J. and Netto, J. 1999a. Age and Length Composition of Columbia
Basin Chinook, Sockeye, and Coho Salmon at Bonneville Dam in 1998. Columbia River
Inter-Tribal Fish Commission Technical Report 99-3. Portland, Oregon.

Hooff, R.C., Ritchie, A., Fryer, J. and Netto, J. 1999b. Age and Length Composition of
Columbia Basin Chinook, Sockeye, and Coho Salmon at Bonneville Dam in 1999.
Columbia River Inter-Tribal Fish Commission Technical Report 99-4. Portland, Oregon.
Kitchell, J .F., Steward, DJ. and Weininger D. 1977. Applications of a bioenergetics

112

model to perch (Percaflavescens) and walleye (Stizostedion vitreum). J. Fish. Res. Board
Can. 34: 1922-1935.

Mangel, M., and CW. Clark. 2000. Dynamic Modeling in Behavioral Ecology.
Princeton University Press, Princeton, New Jersey.

Madenjian, Charles P., Steven A. Pothoven, John M. Dettmers, and Jeffrey D. Holuszko.
2006. Changes in seasonal energy dynamics of alewife (Alosa pseudoharengus) in Lake

Michigan after invasion of dreissenid mussels. Canadian Journal of Fisheries and Aquatic
Science 63(4): 891-902.

Madenjian, C. P., R. F . Elliot, T. J. DeSorcie, R. M. Stedman, D. V. O'Connor, and D. V.
Rottiers. 2000. Lipid Concentrations in Lake Michigan Fishes: Seasonal, Spatial,
Ontogenetic, and Long-Tenn Trends. Journal of Great Lakes Research, 26:427-444.

Miranda, LE. and W.D. Hubbard. 1994. Length-Dependent Winter Survival and Lipid
Composition of Age-0 Largemouth Bass in Bay Springs Reservoir, Mississippi. Trans.
Am. Fish. Soc. 123: 80-87.

Munch, SB. and DO. Conover. 2002. Accouting for local physiological adaptation in
bioenergetic models: testing hypotheses for growth rate evolution by virtual transplant
experiments. Can J Fish Aquat Sci 59: 393-403.

Oliver, J. D., G. F. Holeton, and K. E. Chua. 1979. Overwinter mortality of fingerling
smallmouth bass in relation to size, relative energy stores, and environmental temperature.
Transactions of the American Fisheries Society 108: 130-136.

Patterson, D.A., MacDonald, J .S., Hinch, S.G., Healey, M.C., and AP. Farrell. 2004.
The effect of exercise and captivity on energy partitioning, reproductive maturation and
fertilization success in adult sockeye salmon. Journal of Fish Biology 64: 1039-1059.

Peters, A.K., Jones, M.L., Bence, J .R. and Honeyfield, D. in press.
Monitoring Energetic Status of Lake Michigan Chinook Salmon Using Water
Content as a Predictor of Whole-Fish Lipid Content. Journal of Great Lakes
Research.

Peeters, P. and Royseck, K. 2003. Harvest, Age, and Size at Age of Chinook and Coho
Salmon at Strawberry Creek Weir and Besadny Anadramous Fisheries Facility Fall 2003.
Wisconsin Department of Natural Resources Technical Report.

Roff, D.A. 1992. Evolution of life histories: theory and Analysis. Chapman and Hall,
New York.

Stearns, SC. 1992. The Evolution of Life Histories. Oxford : Oxford University Press.

113

Tanner, HA. and Tody, W.H. 2002. History of Great Lakes Salmon Fishery: A
Michigan Perspective. Pages 139-154. K.D. Lynch, M.L. Jones and W.W. Taylor,
editors Sustaining North American Salmon: perspectives across regions and disciplines.
American Fisheries Society, Bethesda, Maryland.

Thedinga, J .F., Johnson, S.W., and K.V. Koski. 1998. Age and Marine Survival of
Ocean-Type Chinook Salmon from the Situk River, Alaska. Alaska Fishery Research
Bulletin, 5: 143-148.

Trudel, M., Thiess, M.E., Morris, J .F.T, Candy, J .R., Beacham, T.D., Higss, D.A., and
D.W. Welch. 2006. Overwinter mortality and Energy Depletion in Juvenile Pacific
Salmon off the West Coast of British Columbia and Alaska. Abstract from oral
presentation presented at the 2nd North Pacific Anadramous Fishes Conference workshop.

Trudel, M., Tucker, S. Morris, J .F.T., Higgs, D.A., and D.W. Welch. 2005. Indicators of
Energetic Status in Juvenile Coho and Chinook Salmon. N Am J of Fisheries
Management. 25: 374-390.

Trudel, M., Geist DR, and Welch, D.W. 2004. Modeling the Oxygen Consumption
Rates in Pacific Salmon and Steelhead: An Assessment of Current Models and Practices.
Trans. Am. Fish. Soc., 133:326-348.

Trudel, M., Tucker, S. Morris, J .F .T., Higgs, D.A., and D.W. Welch. 2005. Indicators of

Energetic Status in Juvenile Coho and Chinook Salmon. N Am J of Fisheries
Management. 25: 374-390.

114

Apr
May
Jun
Jul

Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul

Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul

Aug

"/0 Lipid allocation
Pacific

10
30
70
90
30
100
10
100
40
100
100
60
60
60
60
100
10
70
20
100
100

0000000

% Lipid allocation
Lake Michigan

1 00
O
0

30
0
3O
40
100
80
0
0
0
3O
20
60
90
5O
40

100

100

100

00000000

Table 4.1 Optimal Lipid Allocation Strategy for Chinook salmon modeled in Lake

Michigan and in the Pacific beginning at age 1 in April and maturing at age 3 in

September.

Lake Michigan

Weight (g) 7001
Length (mm) 826
% Gonad 22.2

Final Egg Mass (g) 1556

Pacific

7165

805

23.7

1700

Virtual Transplant
4905
715
20.5

1230

Table 4.2. Final values at time of maturity for weight (g), length (mm), % gonad (per

gram whole fish) and final egg mass (a surrogate measure of fitness) for age 3 fish

modeled under Lake Michigan condition, Pacific conditions, and fish employing the

optimal strategy for the Pacific model, but transplanted to Lake Michigan conditions.

116

Pacific
Length (mm)
Weight (g)
% Gonad

Egg Mass (a)

Lake Michigan
Length (mm)
Weight (g)
% Gonad

Egg Mass (g)

Optimal Strategy
from Model

806
7164
23.7

1700

826
7001
22.2

1556

Additional 10%
Somatic Growth

895

8601

17.9

1546

890

8097

18.7

1515

Additional 10%
Lipid Reserves

753
6236
26.5

1655

760
5945
25.8

1536

Table 4.3. The effect of increasing energy allocation from the optimal strategy to a

10% in somatic grth and to a 10% increase in lipid reserves on length, weight, the

percent of the whole body composed of gonads and final egg mass for both fish

modeled under the Pacific and the Lake Michigan environmental conditions.

117

Age of Maturity Expected final egg mass (g) in April of Year 2

2 1086, 1070, 1050, 1016
3 1535,1191,1041, 912
4 1233, 976, 773, 612

Table 4.4. Expected final egg mass (g) in April of Year 2 for fish modeled in the
Lake Michigan environment with age at maturation set to 2, 3, and 4 years. Results

are shown under 4 different mortality regimes: 0.1, 0.2, 0.3, and 0.4.

118

Age of Maturity Expected final egg mass (g) in April of Year 2

2 1298, 1245, 1104, 1032
3 1488, 1302, 1139, 997
4 1356, 1058, 838, 663

Table 4.5. Expected final egg mass (g) in April of Year for fish modeled in the
Pacific environment with age at maturation set to 2, 3, and 4 years. Results are

shown under 4 different mortality regimes: 0.1, 0.2, 0.3, and 0.4.

119

Month

June-Year 1
Aug-Year 1
J une-Year 2
Aug-Year 2
J une-Year 3

Aug-Year 3

Modeled Pacific

5.5

7.4

11.2

15.1

13.1

11.9

Modeled LM

1.4

1.3

4.7

6.8

7.7

6.8

Actual LM Transplant

1.7

5.7

7.3

7.9

9.6

5.0

6.4

8.8

9.1

15.3

11.0

9.8

Table 4.6. Lipid levels (percent lipid per gram whole fish) during June and August of

years 1-3 for fish modeled under Pacific conditions, Lake Michigan conditions, actual

fish collected and analyzed from Lake Michigan, and for a virtual transplant from the

Pacific to Lake Michigan.

120

a;
l
1

 

  

 

—<>'—Pac'ific
14 , fix". -—-—- Lake Michigan
{5 12-
9 10
3 .
S c fir)
8 8'
E
#3 6»
4r ur ~‘_‘__.
2

 

 

 

Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar

Figure 4.1. Temperature values (°C) for the Lake Michigan and Pacific environments

used in the model (unpublished data from Roger Bergstedt and Dr. David Welch).

121

40

 

-0— Pacific ' N
-I- Late Michigan I \

30r

10-

Monthly Energy (cal)

-10 .

 

 

 

 

-20

 

Apr Jun Aug Oct Dec Feb Apr Jun Aug Oct Dec Feb Apr Jun Aug

Figure 4.2. Monthly surplus energy (calories) available for allocation (above horizontal
line) or monthly energy deficit (below horizontal line) for Chinook salmon modeled in

Pacific and in Lake Michigan and set to mature at age 3.

122

8000 - a . . .

 

—‘>— Pacific
7000 L -a- Lake Mchigan ,’ J
----<»- Virtual Transplant I
6000 - //
A 5000 " ’r ,.0 ‘
3 ’ “'0
E .I/ up,
5: 4°°° ' ’ .. l
g 3000 ' I' J— J‘ . ~I’ 6:0" 1
/ o...°...or "
2000 [ /
/ _,o---<>--<>---o-"°".
,o'--°'
1000 l ,’ -'
00".0
0 ”"3""

 

 

 

Apr Jun Aug Oct Dec Feb Apr Jun Aug Oct Dec Feb Apr Jun Aug

Figure 4.3. Weight (g) of Chinook salmon modeled in the Pacific and in Lake Michigan,

the Pacific, and a virtual transplant from the Pacific to Lake Michigan.

123

 

18 . f r . . . . . . . . fl 4 . .
-°— Pacific

15 . -'- Lale Michigan
---<>~ Virtual Transplant

    

14-
12r

a \
. s c ‘ l \ .'
o ‘ ' ‘. o “
n
0' ‘0 ' ' : ‘\ a. ‘
10 n . a I '. a , . q
u o c -
: u c ' n . : O
. I o c -‘
, a 0 I _

% Lipid (per gram body weight)

 

 

 

Apr Jun Aug Oct Dec Feb Apr Jun Aug Oct Dec Feb Apr Jun Aug

Figure 4.4. The percent of lipid (per gram body weight) of Chinook salmon modeled in

the Pacific, Lake Michigan, and a virtual transplant from the Pacific to Lake Michigan.

124

 

20 I I I I U I T I I I r I T I 1’

 

 

 

 

—o—06

1 .
fl 8 -1J- 0.62
'5) 15. r --...... 0.64
.5 5...: +0.66
: 14. .9. g -----+ 0.68
.0 ,"~ .1 'u
g 12-
E 10-
U)
'22 .. .
'0 .
'5. 6
3
°\° 4r

2 .

0

0 May Jul Sept Nov Jan Mar May Jul Sept Nov Jan Mar May Jul

Figure 4.5. The percent lipid (per gram body weight) over time for 4 different values of

maximum consumption values for fish modeled in the Lake Michigan environment.

125

 

20

-o— 0.575. 0.33

18r —-- 0.6.0.33 54-.

16L ...o... 0525,0333 /.p' .35?“
_. - 0.65.0.33 ‘ ' " ' of If!

. d: , . ‘\ I ..
14 . . / .' . ° ‘ ' \N 1
cl ‘ O l 0
p x , , a,
12 ' c I" o 1
' o W ' . c .

lq 6 ' 0’
10? ‘ Ia'" "
I '
L ' l P
6 , 5
I l
4.

to May Jul Sept Nov Jan Mar May Jul Sept Nov Jan Mar May Ju

% Lipid per gram body weight

 

 

 

Figure 4.6. The percent lipid (per gram body weight) over time for 4 different values of

maximum consumption for fish modeled in the Lake Michigan environment.

126

lilllllllllllllllllllllllllllllll

3802845

   

, A
-----i-