.. sink.

Jr.
.

‘3”.

QC... i .1!

”3%.," JAW.

kaﬁawwﬁu
.

 

”(rat 2

L

. . «amt.
.;.m.rMR§. . . . .
=nE.\n: \LU . . V . ‘ _ . ._
a.” , , , A . . . 3m. 9m.
. . , , ‘ . . 4 .aamwﬁ?
. ‘ . . fumﬁu,‘

5 .

 

 

\J ‘ .
ﬁning
xﬁumﬁw...‘
Aﬁﬁ..2..
3???”
3.. . :
nx‘nwthh?
rim.» . r: . ‘ .
r...“ uhﬁﬂfzi .. .

 

‘i
game:

 

(on! @337 J€.i.,::\.,.?n.uba . 91:47“ 0.3. aft-#1

51.... ‘51.“. Du ‘ ‘ a any... ..

 

_ .
w
_

Wniwwudmﬂeumum: w ,, V. 41...)». 3... .

 

VIII. ‘ 0". |

i "“1"?
‘-._ v S
‘-‘a

a(.J

, -’ r:
6%;17000

 

LIBRARY
Michigan State
U niversity This is to certify that the

 

 

 

dissertation entitled

AN EVALUATION OF THE ROLE OF TOP PISCIVORES
IN THE FISH COMMUNITY OF THE MAIN BASIN OF
LAKE HURON

presented by

Norine E. Dobiesz

has been accepted towards fulfillment
of the requirements for the

Ph.D. degree in Fisheries and Wildlife

 

 

0 Major Professor’s Signature

Uuné, 1%) &003

 

Date

MSU is an Affirmative Action/Equal Opportunity Institution

_ ﬂ" -—-—‘.—- Fﬁv -‘ ﬁ- —

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

 

we I 12902

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 CJCIRC/DaIBDquBS-DJS

 

AN EVALUATION OF THE ROLE OF TOP PISCIVORES
IN THE FISH COMMUNITY OF THE MAIN BASIN OF LAKE HURON

By

Norine E. Dobiesz

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

2003

ABSTRACT

AN EVALUATION OF THE ROLE OF TOP PISCIVORES
IN THE FISH COMMUNITY OF THE MAIN BASIN OF LAKE HURON

By

Norine E. Dobiesz

Stocking of hatchery-reared ﬁsh has been widespread in Lake Huron since the
mid-1960’s, representing the majority of recruitment for several key predator populations
including the introduced Chinook salmon and the native lake trout. With recruitment
dominated by hatchery plants, natural limitations on recruitment may not be able to
prevent predator populations from exceeding the capacity of the forage base. Exceeding
forage ﬁsh capacity can reduce predator growth, negatively affect predator survival, and
delay or impair predator reproductive capabilities. The purpose of my research was to
improve our understanding of the forage demand by the key predators in Lake Huron.
This was accomplished by analyzing the temporal and spatial characteristics of the
caloric content of Lake Huron ﬁsh species, using bioenergetics models coupled with age-
structured stock assessment models to estimate annual population consumption,
projecting future forage demand under different management scenarios; and
parameterizing a functional response model for the dominant predator, Chinook salmon.

The key predators in the open waters of the main basin of Lake Huron are burbot,
lake trout, Chinook salmon, and walleye. Estimates of their combined forage demand

averaged nearly 36 million kg annually between 1996 and 1998. During this time, lake

trout and Chinook salmon were the major consumers, accounting for 74% of the total
consumption of prey ﬁsh by the key predators. Based on estimates of prey abundance,
consumption by the key predators may be approaching prey capacity, supported by recent
evidence of declines in predator growth. Projections of forage demand resulting from
various management actions suggest that changes to Chinook salmon stocking and
reductions in sea lamprey-induced mortality have signiﬁcant effects on predator forage
demand.

A functional response model relates the number of prey eaten to prey abundance.
We used this model to explore how changes in prey abundance affect consumption and
growth. Our functional response model suggested that variations in total consumption
and growth have been only weakly tied to measured prey abundance. Age 1-4 Chinook
salmon were feeding above 60% of their maximum rate of consumption and variations in
prey abundance explained little of the variation in observed growth. Model ﬁtting results
suggest that the decline in Chinook salmon growth between 1974 and 1998 cannot be
explained by variations in prey abundance so observed declines in growth must be related
to other factors. We noted differences in weight—at-age 1 followed a cohort through its
life span such that ﬁsh that weighed less at age I consistently weighed less throughout
their life span than ﬁsh whose weight at age 1 was higher. Another possible explanation
for our model results is that the assumed relationships and constants we used were
substantially in error, and there is actually a stronger relationship between predator

consumption and prey availability.

To my Dad, the man who taught me to ﬁsh.

Norbert Dobiesz

April 1, I930 — September 16, 2002

ACKNOWLEDGMENTS

This work was supported by the thoughtful inputs and scientiﬁc efforts put forth by many
participants. First, I would like to thank my committee members, Dr. Angela Merti g, Dr.
Richard Kobe, Dr. Michael Jones, and Dr. James Bence, for their insights and editorial
reviews. Next, I also thank those agencies who supplied ﬁsh samples for the energy
density study and data for the various models used in these analyses or who funded parts
of my research including the Michigan Department of Natural Resources, the
Chippewa/Ottawa Resource Authority, the Ontario Ministry of Natural Resources, the
USGS Great Lakes Science Center, and the Great Lakes Fishery Commission. My
thanks also goes to the members of the Lake Huron Technical Committee for helping test
the pilot version of the Consumption Projection Model computer program. Finally, I
would like to thank my family, especially my parents and my sister, for their endless

support and encouragement.

TABLE OF CONTENTS

 

 

 

 

 

LIST OF TABLES - . _ - VIII

LIST OF FIGURES ....... ......... - - XIII

CHAPTER 1

INTRODUCTION ..... ....... g .............. - - 1
Energy Density ................................................................................................................. 4
Bioenergetics .................................................................................................................... 5
Estimating Consumption .................................................................................................. 5
Consumption Projection Model Computer Program ....................................................... 6
Functional Response Model ............................................................................................. 7
Literature Cited ................................................................................................................ 8

CHAPTER 2

ENERGY DENSITY OF KEY PREDATORS AND THEIR PREY IN LAKE

HURON - -- _ - ‘ . - - = 11
Introduction .................................................................................................................... I 1
Methods .......................................................................................................................... 14
Results ............................................................................................................................ 18
Discussion ...................................................................................................................... 21
Literature Cited .............................................................................................................. 30

CHAPTER 3

RECENT AND PROJECTED ESTIMATES OF FORAGE FISH CONSUMPTION

BY KEY PREDATORS IN THE MAIN BASIN OF LAKE HURON ------- - ..... 49
Introduction .................................................................................................................... 49
Methods .......................................................................................................................... 52
Results ............................................................................................................................ 62
Discussion ...................................................................................................................... 68
Literature Cited .............................................................................................................. 77

CHAPTER 4

PARAMETERIZATION OF A FUNCTIONAL RESPONSE MODEL FOR
CHINOOK SALMON IN THE MAIN BASIN OF LAKE HURON. - _- - _ “-96

 

Introduction .................................................................................................................... 96
Methods .......................................................................................................................... 99
Results .......................................................................................................................... 105
Discussion .................................................................................................................... 109
Appendix ...................................................................................................................... I I 4
Literature Cited ............................................................................................................ I I 9

vi

APPENDIX A
DESCRIPTIVE DATA FOR FISH SAMPLES USED IN ENERGY DENSITY

 

 

 

ANALYSIS ----- - - u ....... - -- - - - 144
Literature Cited ............................................................................................................ 145
APPENDIX B
PARAMETER VALUES USED IN BIOENERGETICS MODELS ------ - 153
Simulation length ......................................................................................................... I53
Actual and preferred water temperature ..................................................................... 154
Diet composition .......................................................................................................... I54
Prey Energy Density .................................................................................................... 155
Predator Growth .......................................................................................................... I 58
Predator spawning losses ............................................................................................ 159
Predator Energy Density ............................................................................................. I 60
Literature Cited ............................................................................................................ 161
APPENDIX C
DATA AND ASSUMPTIONS USED FOR PROJECTIONS OF CONSUMPTION
............... _- - _ - _- ................. 175
Mortality rates ............................................................................................................. I 75
Weight-at-age ............................................................................................................... I 76
Diet and gross conversion efﬁciency ........................................................................... 177
Recruitment .................................................................................................................. I 77
Size regulations ............................................................................................................ 1 78
Literature Cited ............................................................................................................ 179
APPENDIX D
SURVEY INSTRUMENT AND DESCRIPTIVE ANALYSIS OF RESULTS ....... 194
Introduction .................................................................................................................. I94
Objectives ..................................................................................................................... I94
CPM Training Session ................................................................................................. 195
Survey Instrument ........................................................................................................ 196
Survey responses .......................................................................................................... 197
Conclusion ................................................................................................................... 201
Literature Cited ............................................................................................................ 202

vii

LIST OF TABLES

Table 2.1 - Mean and standard deviation of wet weight (kg), and percent water content
for predator and prey species collected in Lake Huron during 1996-1997. Samples were
collected in Saginaw Bay (B) and the northern (N), central (C), and southern (S) regions
of Lake Huron. Months represent the numerical value for each month a sample was
collected, with January as month 1. ................................................................................. 36

Table 2.2 — Water content (mean with :1 sd) for ﬁsh samples processed in the bomb
calorimeter and those that were only ground and dried. .................................................. 37

Table 2.3 — Regression model results for the relationship of energy density to percent
water content. ................................................................................................................... 38

Table 2.4 — Extra sums of squares criterion for models 1, 2, and 3 where SSE, and SSEf
are the residual sum of squares from the reduced and the full model, respectively, and df,
and d]? are the degrees of freedom in each model. ........................................................... 38

Table 2.5 — Intercepts and slopes for ﬁnal model (model 3) used to predict energy density

from percent water using the equation: E = 0t + B W, where E is energy density in J 0g“
wet weight and W is the percent water content. Model 3 assumed a common slope for all
predators but allowed intercepts to vary, and allowed a different linear relationship for
prey than predators (different intercept and slope) but assumed the same relationship for
all prey species. ................................................................................................................ 39

Table 2.6 —AN OVA main effects (month and region) and covariate (predator wet weight)
for each sampled Lake Huron species. Least square means are shown with one standard
error in parentheses. Missing entries represent effects that could not be tested due to
insufﬁcient samples. ......................................................................................................... 40

Table 3.1 — Mean energy density used for sensitivity analyses. Lake Huron energy
density was determined in Chapter 2. These represent the mean basin-wide values. ‘*’
represent species that used seasonally or regionally varying energy density in the
bioenergetics models. Published energy density values represent the mean value for all
noted references for the species. ...................................................................................... 83

Table 3.2 -— Gross conversion efficiency estimated from bioenergetics models ............. 84

viii

Table 3.3 — Estimates of mean consumption in millions of kg for 1996-1998 ................ 84
Table 4.1 — Symbols used in the chinook salmon functional response model. .............. 122

Table 4.2 — Equations used in chinook functional response model. Descriptions of
variables are shown in Table 4.1. ................................................................................... 123

Table 4.3 — Values of assumed constants used in the functional response model and the
sensitivity analyses. “Base model” denotes the functional response model with four
estimated search rate parameters (Model 4, by prey type and predator age). Other
scenarios represent the conﬁgurations for sensitivity analyses. All assumed constants
used in the “Base model” are listed with subscript indicators and values. Sensitivity
analyses scenarios list only those constants that were changed in the scenario ............. 124

Table 4.4 — Likelihood ratio tests for all combinations of model configurations. ......... 125

Table 4.5 — Model hypotheses and estimated search rate parameter(s) on the log scale
with asymptotic standard errors for parameter estimates shown in parentheses. The
search rate parameters (afa) control the overall search rates for the predator on a prey

species after adjusting for predator and prey sizes. The ﬁrst subscript is the prey type-
speciﬁc scalar for alewife (i=1) and rainbow smelt (i=2). The second subscript is the
predator age grouped by age 1 (a=1) and ages 2-4 (a=2). Some models ignored one or
more of these subscripts and these are represented by dashes in place of a value for the
subscript. ........................................................................................................................ 126

Table 4.6 — Results of the sensitivity analysis using alternate values for assumed
quantities (Table 3). Estimated parameter values are shown with asymptotic standard
error in parentheses. For the search rate parameters (aid), the ﬁrst subscript is the prey

type-speciﬁc scalar for alewife (i=1) and rainbow smelt (i=2) and the second subscript is
the predator age grouped by age 1 (a=1) and ages 2-4 (a=2). ....................................... 127

Table A.1 - Number of samples from different statistical districts and lake regions. For
the regional analysis of energy density, statistical districts were grouped to represent a
particular lake region. The grouping of these statistical districts coincides with the
regional lake trout populations. ...................................................................................... 146

Table A.2 — Species name abbreviations. These 3-letter codes are used in some tables
and ﬁgures when the full species name did not ﬁt into a table or ﬁgure. ...................... 146

Table A3 -- Fish sample characteristics by month. Number of samples, N, are given as
the total number of samples (top) and the number of samples processed in the bomb
calorimeter (bottom, in parentheses). ............................................................................. 147

Table A4 - Fish sample characteristics by statistical district. Number of samples, N, are
given as the total number of samples (top) and the number of samples processed in the
bomb calorimeter (bottom, in parentheses). ................................................................... 148

Table A5 -- Fish sample characteristics by gender. Number of samples, N, are given as
the total number of samples (t0p) and the number of samples processed in the bomb
calorimeter (bottom, in parentheses). ............................................................................. 148

Table A6 - Descriptive statistics by lake region and month for all ﬁsh samples collected.
Means are shown with standard deviations in parentheses. ........................................... 149

Table B.1 — Estimated Lake Huron water temperatures on the frist day of each month,
based on N OAA/GLERL reports (Grumblatt 1976; McCormick 1996; Nalepa et al. 1996;
Johengen et al. 2000) ...................................................................................................... 163

Table 8.2 - Water temperatures on the ﬁrst day of each month as experienced by
predators in Lake Huron during bioenergetics modeling. Estimated water temperatures
are used (Table B. 1) except when the preferred water temperature is exceeded. It was
assumed that predators would reside in their preferred water temperature or in lower
temperatures when the preferred temperature is not available. Shaded cells represent
preferred water temperatures .......................................................................................... 164

Table B.3 — Diet composition of Lake Huron predators by age class. Values represent the
proportion by weight of each prey item in the diet. ....................................................... 165

Table B.4 — Energy density of Lake Huron prey species used in this implementation of
the Wisconsin model. Data were derived from samples collected in Lake Huron (see
Chapter 2) except for invertebrates and “other ﬁsh”, which were not sampled. Mean
energy density for invertebrates (Cummins and Weychuck 1971; Stewart et al. 1983;
Stewart and Binkowski 1986) and for “other ﬁsh” (Stewart et al. 1983; Rudstam et al.
1995) was derived from published values. ..................................................................... 166

Table 3.5 — Predator starting weights (grams) as used in the bioenergetics models. The
ending weights were the starting weights for the next age class. For age 2 walleye, no
value is given for the maintenance period since these represent immature individuals that
are not spawning. Therefore, the end weight for age 2 walleye was the starting weight for

age 3. For chinook salmon, starting weight for age 0 ﬁsh was used as in Stewart and
Ibarra (1991) ................................................................................................................... 167

Table B.6 — Regression results for the ﬁnal model used for each predator. Predators with
no mass cutoff showed no evidence of a relationship between weight and energy density.
Predators with a mass cutoff value were best deﬁned with one model below the cutoff

and another above the cutoff (see Figure B4). ............................................................... 168

Table B.7 - Physiological parameters used in the Wisconsin bioenergetics models for
Lake Huron predators. The equations (Eq) and parameters (e. g., CA, FA, etc.) refer to
bioenergetics models as presented by Hewett and Johnson (1995). .............................. 169

Table C.1 -- Assumptions used during the projection period. These are default
assumptions in the Consumption Projection Model software but the user may change

them. ............................................................................................................................... 180
Table C.2 - Natural mortality rates used in projections of consumption ....................... 181
Table C.3 — Fishing mortality used in projections of consumption. .............................. 182

Table C.4 — Fishing and maturation proportions for chinook salmon used in projections of
consumption. The chinook salmon model operates with two time periods within a year
consisting of the ﬁrst seven months (prior to a “pulse” harvest and maturation process)
followed by the remainder of the year. .......................................................................... 182

Table C.5 - Sea lamprey-induced mortality for lake trout and burbot used in projections
of consumption, before applying the scaling factor (Table C.6). ................................... 183

Table C.6 — Sea lamprey-induced mortality scaling factor for projection periods. After
2015 the last value of 0.1601 was used for all other years ............................................. 184

Table C.7 - Predator weight—at-age (kg) used in projections of consumption. Burbot
weight-at-age was obtained from a von Bertalanffy growth model ﬁtted to Michigan
Department of Natural Resources (MDNR) data. Lake trout weight-at-age was obtained
from MDNR spring gill new surveys. Walleye weight-at-age was estimated from 1985-
1995 Lake Huron creel data. Both the Saginaw Bay and southern region walleye
populations used the same weight-at-age values ............................................................ 185

Table C.8 - Diet composition for the projection period. ............................................... 186

xi

Table C.9 — Age-speciﬁc gross conversion efﬁciencies used during the projection period.
........................................................................................................................................ 187

Table C. 10 — Number of recruits assumed for projection period. .................................. 188

Table C.11 - Lake trout movement matrix used during the projection period. This matrix
deﬁnes the percent of ﬁsh stocked in each stocking location that become resident in each
lake region. ..................................................................................................................... 189

Table C. 12 — Lake trout stocking matrix used during the projection period. This matrix
identiﬁes the number of ﬁsh stocked at each location by year. Values after 2001 are
estimates of the numbers to be stocked. No stocking was reported in MH6, 0H3, or 0H4
during the projection period. .......................................................................................... 190

Table C. 13 - Lake trout recreational ﬁshery minimum size limits during the projection
period .............................................................................................................................. 191

Table C. 14 — The von Bertalanffy growth model parameters used to estimate length-at-
age (mm) for northern and central lake trout during the projection period .................... 191

Table C.15 — Actual mean length (mm) and standard deviation used to estimate the
adjustment factor on recreational ﬁshing mortality for northern and central lake trout
during the projection period. .......................................................................................... 192

Table C. 16 - Size limit adjustment factor on recreational ﬁshing mortality of northern
and central region lake trout. The recreational ﬁshing mortality (Table C3) is multiplied
by the adjustment factor to simulate the effect of hooking mortality related to the
enforcement of the minimum size limit regulations during the projection period. ........ 193

Table D.l. Answers to survey questions. The number of answers is shown in hold. ..203

xii

LIST OF FIGURES

Figure 2.1 — Statistical districts grouped into lake regions (north, central, south, and
Saginaw Bay) for the regional analysis of energy density. Statistical districts in the US
use MH labels while Canadian waters are labeled with OH (Smith et al. 1961). ............. 41

Figure 2.2 -- Relationship between mean percent water content and mean energy density
for all sampled Lake Huron species. ................................................................................. 42

Figure 2.3 — Relationship between energy density and percent water content for predators
and prey in Lake Huron. [Note: the x-axis origins are not continuous from zero.] ......... 43

Figure 2.4 — Mean regional energy density of the key predators and their prey in Lake
Huron during 1997. Least squares means are shown with one standard error. ................ 44

Figure 2.5 — Mean seasonal energy density of the chinook salmon and lake trout in Lake
Huron, 1996-1997. Least squares means are shown with one standard error. ................. 45

Figure 2.6 — Mean seasonal energy density of the primary prey species in Lake Huron
during 1996-1997. Least squares means are shown with one standard error ................... 46

Figure 3.1 — Statistical districts in the US and Canadian waters of Lake Huron (Smith et
al. 1961) grouped into lake regions. Statistical districts, used in sampling to denote
location, are shown as MH- (Michigan waters) or OH- (Ontario waters) ........................ 86

Figure 3.2 - Estimated gross production of key predators in the main basin of Lake
Huron from 1984-1998 ...................................................................................................... 87

Figure 3.3 — Estimated total consumption by key predators in the main basin of Lake
Huron from 1984-1998 ...................................................................................................... 88

Figure 3.4 — Estimated chinook salmon consumption and population biomass in the main
basin of Lake Huron from 1968-1998. .............................................................................. 89

Figure 3.5 — Estimated biomass of key predators in the main basin of Lake Huron from
1984-1998 .......................................................................................................................... 90

xiii

Figure 3.6 — Estimated annual consumption by lake trout and walleye by lake region,
1984-1998 .......................................................................................................................... 91

Figure 3.7 — Comparison of estimated current key predator consumption in the main
basin of Lake Huron to estimated consumption by pre-collapse lake trout. ..................... 92

Figure 3.8 — Diet composition of the key predators in the main basin of Lake Huron.
Proportion of each prey type in the diet represents the mean by weight for 1989 — 1999.
........................................................................................................................................... 93

Figure 3.9 — Comparison of estimated key predator consumption to estimated combined
alewife and rainbow smelt biomass in the main basin of Lake Huron from 1984-1998. .94

Figure 3.10 - Projected consumption by key predators in the main basin of Lake Huron
through 2020 under three management scenarios. ............................................................ 95

Figure 4.1 —Combined alewife and rainbow smelt abundance and standing stock biomass
for the main basin of Lake Huron from 1974-1998. ....................................................... 128

Figure 4.2 — Annual age-speciﬁc chinook salmon growth (top panel) and instantaneous
growth (bottom panel) from weight-at-age data. ............................................................ 129

Figure 4.3 — Relationship between observed chinook salmon growth determined from
weight-at-age data and the combined alewife and rainbow smelt abundance between 1974
and 1998. ......................................................................................................................... 130

Figure 4.4 - Predator size preference for the prey species. The values of the function
shape variables are given in Table 4.3. The Default curve was used for both prey species
in the model ﬁtting process. The adjusted curves were used to test the sensitivity of the
size preference function to differences in prey weight for a given size category. .......... 131

Figure 4.5 — Relationship between age-speciﬁc maximum consumption (Cmax) and
chinook salmon length (mm) used to determine handling time in the functional response
model. .............................................................................................................................. 132

Figure 4.6 — Observed and predicted growth with search rate parameter related only to
prey type (Model 2) or related to both prey type and predator age (Model 4) ................ 133

xiv

Figure 4.7 -- Observed and predicted growth with search rate parameter related only to
predator age (Model 3) or related to both prey type and predator age (Model 4). .......... 134

Figure 4.8 — Relationship between the estimated age-speciﬁc consumption of prey
biomass (kg) and combined alewife and rainbow smelt abundance. .............................. 135

Figure 4.9 - Estimates of consumption from Model 4 using incremental prey abundance
and two levels of ﬁxed predator size representing high (1974) and low (1984) growth
periods. The vertical dashed line represents the lowest observed prey abundance between
1974-1998 ........................................................................................................................ 136

Figure 4.10 — Consumption by a cohort and weight at age 1. Consumption = 39.76 W +

21.49 with R2 = 0.7696, where W is weight-at-age 1 (kg). Consumption is shown for 21
full cohorts over 1974—1998 (1995 was last cohort). .................................................... 137

Figure 4.11— Comparison of age-speciﬁc consumption from the bioenergetics models
(Chapter 3) and the functional response model ............................................................... 138

Figure 4.12— Sensitivity of estimated search rate parameters to ﬁxing Cmax (ﬁrst bar),
increasing Cmax by 20% (second bar), decreasing Cmax by 20% (third bar), modifying
size preference for weight of prey ﬁsh (fourth bar), and adjusting values to actual mid-
year, day 182 (ﬁfth bar). Each grouping represents one search rate parameter (afa) as

deﬁned in Table 4.5 ......................................................................................................... 139

Figure 4.13— Results of the sensitivity analysis showing the minimum and maximum
values of Pmax for each age. The lower level of each bar represents the minimum Pmax
while the upper point represents the maximum value for each sensitivity analysis. The
dashed lines represent the minimum and maximum values of Pmax estimated from Model
4 using the base values of all assumed constants (Table 4.3). ........................................ 140

Figure 4.14 — Estimated proportion of maximum consumption (Pmax) related to chinook
salmon abundance (top panel) and combined alewife and rainbow smelt abundance
(bottom panel). ................................................................................................................ 141

Figure 4-15 — Annual estimated gross conversion efﬁciency (calculated from
bioenergetics models, Chapter 3 and Appendix B) for age 1-4 chinook salmon in the
main basin of Lake Huron. .............................................................................................. 142

Figure A.1 — Mean energy density (A) and mean percent water (B) of each species
collected with error bars representing one standard deviation. The percent coefﬁcient of

XV

variation for each graphed variable is shown in the table on the right. Note, graphs use
non-zero origin. Abbreviated species names are shown in Table A.2. .......................... 152

Figure B] — Alewife seasonal energy density in 10g1 wet weight. Samples were
available from June through September but only from the central and southern regions
(A). To approximate a seasonal energy density pattern, missing months were estimated

as the proportional difference from published values of alewife energy density from other
Great Lakes (B). .............................................................................................................. 171

Figure B.2 — Rainbow smelt seasonal energy density in J og‘l wet weight. Samples were
available for J anurary and from May through August; samples from all lake regions were
pooled (A) as results from Chapter 2 showed no signiﬁcant differences between regions.
To approximate a seasonal energy density pattern, missing months were estimated as the
proportional difference from published values of rainbow smelt energy density from
other Great Lakes (B). ..................................................................................................... 172

Figure B.3 — Linear relationships between predator weight and energy density used in this
implementation of the Wisconsin model. Where two different relationships were
employed, the mass cutoff separating the two lines is indicated below the title ............. 173

Figure D.1. Questionnaire for evaluating the program Consume .................................. 204

xvi

Chapter 1

Introduction

Among the Great Lakes, Lake Huron is ranked as the second largest in surface
area and the third largest in volume (Beeton 1984). Native peoples have inhabited the
Lake Huron basin and ﬁshed its waters since the Wisconsin ice sheet retreat
approximately 12,000 years ago (Spangler and Peters 1995). Settlers in the area ﬁshed
primarily for food but by the mid 18008 commercial ﬁshing developed causing
nearshore ﬁshing to move farther offshore (Spangler and Peters 1995). With
technological advances in capture gear, ﬁshing vessels, and preservation techniques,
Lake Huron commercial ﬁsheries rapidly grew throughout the early 19005.

Lake Huron supported one of the world’s largest freshwater ﬁsheries before the
19503, primarily in lake trout, with average annual commercial yields of 2.4 million kg
from 1912 to 1940 (Ebener et al. 1995). A downward trend in annual catch during the
early 19003 to the mid-1930s was attributed to invasion of exotic species and pollution
from industrial deve10pment near the littoral zone (Berst and Spangler 1973).
Overﬁshing coupled with the negative effects of sea lamprey Petromyzon marinus
predation initiated a collapse in Lake Huron piscivore populations during the 19405
(Christie 1974; Mills et al. 1993; Eshenroder et al. 1995). Prior to the collapse of the
ﬁshery, Lake Huron’s top predators were lake trout Salvelinus namaycush and burbot
Lota lota but by the mid-19605 these piscivores were rare in the upper Great Lakes
(DesJardine et al. 1995; Eshenroder and Bumham-Curtis 1999). Today, chinook
salmon Oncorhynchus tshawytscha, lake trout, walleye Stizostedion vitreum, and

burbot are considered the major predators in the main basin and their primary forage

l

ﬁsh consist of the exotic prey ﬁsh alewife Alosa pseudoharengus and rainbow smelt
Osmerus mordax.

Restoration efforts in Lake Huron have focused on rebuilding piscivore
populations, controlling exotic alewife, restoring self-sustaining stocks of lake trout,
and promoting the recreational ﬁshery (DesJardine et al. 1995; Great Lakes Fishery
Commission 2001). Methods include extensive stocking of salmonines, a sea lamprey
reduction program, and control of ﬁshing effort. In the late 19503, management
agencies began chemical treatment of streams to control sea lamprey abundance and
reduce mortality on native ﬁsh species (Smith and Tibbles 1980). Lake trout stocking
began in 1969 with 31,000 ﬁsh and increased to 3.3 million in 1992 (Ebener et al.
1995). Other top predators such as chinook salmon and walleye have been stocked
since 1968 (Ebener et al. 1995). Today, stocked predators form an important part of
the ecosystem and primarily consume exotic prey species (Christie 1974; Kitchell et
al. 1994; Eby et al. 1995).

In Lake Huron, stocking of hatchery-reared salmon and trout provides
substantial recreational, social, and economic beneﬁts. In 1991 the commercial
ﬁsheries in Lake Huron achieved landed harvests of $3.4 million (US) and $6.9
million (CAN) (Dann 1994). Similarly, in 1990 - 1991 the economic value of all
Great Lakes’ recreational ﬁsheries was estimated to be approximately $1.34 billion
(1991, US) and $0.26 billion (1990, CAN) in US and Canadian waters respectively
(Bence and Smith 1999). In Lake Huron alone, US recreational ﬁshing effort was
estimated at 2,113,000 ﬁshing days while Canadian effort was more than double that

at 4,579,000 ﬁshing days (Bence and Smith 1999). Communities bordering the lake

beneﬁt from monies spent by recreational users including expenditures for food,
lodging, or other related activities. Indirect economic value can be attributed to a
healthy lake ecosystem and its functions (Costanza et al. 1997) such as fresh water
storage (Edwards and Abivardi 1998) and nutrient recycling by organisms in the lake
(Kraft 1993).

The restoration of naturally reproducing piscivore stocks has met with limited
success. Natural reproduction is occum’ng in some stocks but hatchery-reared ﬁsh
constitute the majority of recruitment (Ebener et al. 1995). With predator abundance
predominantly controlled through stocking, an important natural linkage between
predator abundance and prey availability may be disrupted. Therefore, management
actions that alter predator abundance could result in predator consumption outreachin g
the forage ﬁsh capacity. However, the effects of ﬁshery management actions on
predator-prey dynamics are unknown (Stewart et al. 1981; Kitchell et a1. 1994). In
Lake Huron, predator forage demand and the effects of changes in prey ﬁsh abundance
on predator growth are not well understood.

The purpose of my research is to examine how forage demand by the key
predators responds to management actions such as changes in stocking or the
reduction of sea lamprey-induced mortality. I addressed these questions through four
distinct steps: (1) determination of the caloric content of Lake Huron ﬁsh species; (2)
estimation of prey consumption for an average predator using bioenergetics models;
(3) estimation of consumption by extrapolating individual predator consumption to a
predator population, and projection of predator consumption under different

management scenarios; and (4) parameterization of a functional response model for

the dominant predator, chinook salmon, to explore how changes in prey abundance
affect consumption and growth. Estimates of the caloric content, or energy density, of
the predators and prey are an important input into bioenergetics models which in turn,
estimate consumption of prey by an individual ﬁsh given its growth. Stock assessment
models then expand this consumption to a population and account for prey
consumption by ﬁsh that die during the model time step. Results from the ﬁrst three
steps were consolidated into a computer program that allows ﬁsheries managers to
project future consumption of prey by the key predators under varying management

actions. The following paragraphs address each of these steps.

Energy Density

Several studies have explored the seasonal and annual cycles of energy density
of ﬁsh species in Lake Michigan (Foltz and Norden 1977; Flath and Diana 1985;
Stewart and Binkowski 1986), Lake Ontario (Rand et al. 1994), and Lake Superior
(Vondracek et al. 1996; Johnson et al. 1999) but corresponding data are generally
lacking for Lake Huron. Further, some studies of forage demand have borrowed
energy density from other species (e.g., Hurley 1986; Madon and Culver 1993;
Rudstam et al. 1995) or from the same species in other lakes (LaBar 1993). However,
energy densities may not be interchangeable since ﬁsh condition and thus energy
content varies with changes in the ﬁsh community, food density, and climatic
conditions (Rand et al. 1994). In Chapter 2, I describe the process I used to estimate
the energy content of Lake Huron ﬁsh species and the statistical analyses (AN OVA

and ANCOVA) used to determine how energy content varies regionally and

seasonally. Appendix A contains supplemental material about the samples used to

determine energy content.

Bioenergetics

Bioenergetics models relate an individual organism’s assimilation and
utilization of energy from food, partitioning that energy into growth, metabolism, and
waste losses (Adams and Breck 1990; Ney 1993). These models require energy
budgets to consist of balanced inputs and outputs (Hewett and Johnson 1995). Here,
growth integrates the feeding rate over time so short-term variability in food
availability, temperature, etc. is minimized. Fish growth is denoted as an increase in
body weight, which is the simplest measure to obtain for an energy budget. In
Appendix B I describe the Lake Huron-speciﬁc parameters used in the Wisconsin
Model, a widely used bioenergetics model (Hewett and Johnson 1995), to estimate
year- and age-speciﬁc consumption for an average predator. Values of consumption
and growth from these models were used to estimate the gross conversion efﬁciencies
(GCE) of the Lake Huron predators. These GCEs become an important input into the

estimation of consumption as outlined in the next section.

Estimating Consumption

Balancing predator forage demand and prey ﬁsh availability is a major concern
for Great Lakes ﬁshery managers. Estimates of recent consumption provide insight
into the effects of stocking practices and other management actions on predator forage

demand. In Chapter 3, I describe how age-structured population models, using the

production-conversion efﬁciency approach (Ney 1993), extend consumption by an
individual predator to estimates of consumption by a population. Projecting future
predator consumption under different management scenarios allows managers to
compare the potential effects of management initiatives on predator-prey dynamics.
Assumptions regarding mortality rates, weight-at-age, diet composition, and GCE,
needed to project predator consumption for the period 1999-2020 are also outlined in

Chapter 3.

Consumption Projection Model Computer Program

Prior to this dissertation, preliminary results of predator consumption were
contained in the “No Name model”, which was used to assess the overall consumption
of prey ﬁsh by predators in the main basin of Lake Huron using a series of eight linked
spreadsheets. While the “No Name model” could be amended with new data and
additional calculations, correctly updating the series of spreadsheets was cumbersome,
often requiring numerous changes to one or more spreadsheets. Furthermore, to
compare multiple management scenarios required a copy of the entire suite of
spreadsheets for each scenario. Updating these spreadsheets introduced errors common
to spreadsheet manipulation (e. g., copying cells or losing cell formulas).

As part of an ongoing research program to improve our understanding of
predator consumption, I created the Consumption Projection Model (CPM). This
computer program is a user-friendly replacement for the “No Name model” that greatly
simpliﬁes the process of projecting consumption under multiple management scenarios

(Dobiesz 2003). The CPM employs a user-friendly Microsoft Windows-based interface

that allows users to quickly and easily obtain and compare consumption projections
resulting from various management actions. For projections period, CPM uses
assumptions regarding key population attributes (Appendix C). This computer program
was distributed to ﬁsheries managers during a training session. Participants were also
asked to complete a short survey to determine the usefulness and ease-of—use of the

CPM (Appendix D).

Functional Response Model

The amount of prey eaten and the composition of the diet depend upon prey
availability in ways that are unknown or only partially understood. The functional
response model provides a framework for relating the number of prey eaten per unit
time to prey density (Holling 1959; Murdoch 1973). Predation mortality as predicted
from functional response models and estimated predator consumption from
bioenergetics models provide two ways to view the effects of a consumer on their
forage base. Functional response models relate the number of prey eaten to prey
abundance. Extending the model to multiple prey species provides insight into how
prey consumption changes as the composition of the forage base changes. Similarly,
bioenergetics models provide a method of estimating consumption by a single predator
that may be extended to an entire population. Consumption estimates from
bioenergetics models can be compared to functional response estimates. In Chapter 4,
I describe the parameterization of a Type II functional response for chinook salmon,
the dominant key predator in Lake Huron, and compare the results from the functional

response and bioenergetics models.

Literature Cited

Adams, S. M. and J. E. Breck. 1990. Bioenergetics. In Schreck, CB. and PB. Moyle
(eds.), Methods for ﬁsh biology. American Fisheries Society, Bethesda,
Maryland, p. 389-415.

Beeton A. M. 1984. The World’s Great Lakes. Journal of Great Lakes Research 10:
106-113.

Bence, J. R. and K. D. Smith. 1999. An overview of the recreational ﬁsheries of the
Great Lakes. In Taylor, W. W. and P. Ferreri (eds.), Great Lakes Fisheries Policy
and Management: A Binational Perspective, Michigan State University Press.

Berst, A. H. and G. R. Spangler. 1973. Lake Huron — the ecology of the ﬁsh community
and man’s effects on it. Great Lakes Fishery Commission, Tech Report. 21. P 41.

Christie, W. J. 1974. Changes in the ﬁsh species composition of the Great Lakes.
Journal of the Fisheries Research Board of Canada 31: 827-854.

Costanza R., R. d'Arge, R. de Groot, S. Farber, M. Grasso, B. Hannon, K. Limburg, S.
Naeem, R. V. O'Neill, J. Paruelo, and J. Raskin. 1997. The value of the world‘s
ecosystem services and natural capital. Nature 387: 253-260

Dann, S. L. 1994. The life of the lakes: A guide to the Great Lakes ﬁshery. Michigan
Sea Grant Extension Bulletin #E-2440, East Lansing, Michigan.

DesJardine , R. L., T. K. Gorenflo, R. N. Payne and J. D. Schrouder. 1995. Fish-
community objectives for Lake Huron. Great Lakes Fishery Commission Special
Publications 95-1, pp 38.

Dobiesz NE. 2003. Computer Projection Model (installable software and documentation
included). Available for download at the anonymous FTP site at glpd.fw.msu.edu
in the directory CPM V1.0.

Ebener, M. P., J. E. Johnson, D. M. Reid, N. P. Payne, R. L. Argyle, G. M. Wright, K.
Kruger, J. P. Baker, T. Morse and J. Weise. 1995. Status and future of Lake
Huron ﬁsh communities. In Munawar, M., T. Edsall and J. Leach (eds.), The
Lake Huron ecosystem: Ecology, Fisheries and Management. Ecovision World
Monograph Series, SPB Academic Publishing, Amsterdam, The Netherlands,
pp. 125-170.

Eby, L. A., L. G. Rudstam and J. F. Kitchell. 1995. Predator responses to
prey population dynamics: An empirical analysis based on lake

trout growth rates. Canadian Journal of Fisheries and Aquatic
Sciences 52(7): 1564-1571..

Edwards P. J., and C. Abivardi. 1998. The value of biodiversity: Where ecology and
economy blend. Biological Conservation 83 (3): 239-246.

Eshenroder, R. L. and M. K. Bumham-Curtis. 1999. Species Succession and
Sustainability of the Great Lakes Fish Community. In Taylor, W. W. and P.
Ferreri (eds.), Great Lakes Fisheries Policy and Management: A Binational
Perspective, Michigan State University Press.

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

Flath, L. E. and J. S. Diana. 1985. Seasonal energy dynamics of the alewife in
southeastern Lake Michigan. Transactions of the American Fisheries Society
114: 328-337.

Great Lakes Fishery Commission. 2001. Strategic Vision of the Great Lakes Fishery
Commission for the First Decade of the New Millennium. Ann Arbor, MI. 40 p.

Hewett. S. W. and B. L. Johnson. 1995. Fish Bioenergetics Model 3. University of
Wisconsin Sea Grant Institute, WIS-SG-91-250.

Holling, C. S. 1959. The components of predation as revealed by a study of small-
mammal predation of the European pine sawfly. The Canadian Entomologist 91:
293-320.

Hurley, D. A. 1986. Growth, diet, and food consumption of walleye: an application of
bioenergetics modeling to the Bay of Quinte, Lake Ontario, population. In C. K.
Minns, D. A. Hurley, and K. H. Nicholls [eds.] Project Quinte: point-source
phosphorus control and ecosystem response in the Bay of Quinte, Lake Ontario.
Canadian Special Publication of Fisheries and Aquatic Sciences 86: 0706-6481.

Kitchell, J. F., L. A. Eby, X. He, D. E. Schindler and R. A. Wright. 1994. Predator-prey
dynamics in an ecosystem context. Journal of Fish Biology 45(Suppl. A): 209-
226.

Kraft, C. E. 1993. Phosphorus regeneration by Lake Michigan alewives in the mid-
19703. Transactions of the American Fisheries Society 122 (5): 749-755.

Labar G. W. 1993. Use Of Bioenergetics Models To Predict The Effect Of Increased
Lake Trout Predation On Rainbow Smelt Following Sea Lamprey Control.
Transactions of the American Fisheries Society 122 (5): 942-950.

Madon, S. P. and D. A. Culver. 1993. Bioenergetics Model for Larval and Juvenile
Walleyes: An In Situ Approach with Experimental Ponds. Transactions of the

American Fisheries Society 122: 797-813.

Mills, E. L., J. H. Leach, J. T. Carlton and C. L. Secor. 1993. Exotic species in the Great
Lakes: A history of biotic crises and anthropogenic introductions. Journal of
Great Lakes Research 19: l-54.

Murdoch, W. W. 1973. The functional response of predators. Journal of Applied
Ecology 10: 335-342.

Ney, J. J. 1993. Bioenergetics modeling today: Growing pains on the cutting edge.
Transactions of the American Fisheries Society 122: 736-748.

Rand P. S., B. F. Lantry, R. O’Gonnan, R. W. Owens, D. J. Stewart. 1994. Energy
density and size of pelagic prey ﬁshes in Lake Ontario, 1978-1990: implications
for salmonine energetics. Transactions of the American Fisheries Society 123(4):
519-534.

Rudstam, L. G., P. E. Peppard, T. W. Fratt, R. E. Bruesewitz, D. W. Coble, F. A. Copes
and J. F. Kitchell. 1995. Prey consumption by the burbot (Lota lota) population

in Green Bay, Lake Michigan, based on a bioenergetics model. Canadian Journal
of Fisheries and Aquatic Sciences 52: 1074-1082.

Smith B. R. and J. J. Tibbles. 1980. Sea Lamprey (Petromyzon marinus) in Lakes
Huron, Michigan, and Superior: History of invasion and control, 1936-78.
Canadian Journal of Fisheries and Aquatic Sciences 37: 1780-1796.

Spangler, G. R. and J. H. Peters. 1995. The Lake Huron ecosystem: Ecology, Fisheries
and Management. In Munawar, M., T. Edsall and J. Leach (eds.), Ecovision
World Monograph Series, SPB Academic Publishing, Amsterdam, The
Netherlands. Pp. 103-124.

Stewart, D. J ., J. F. Kitchell and L. B. Crowder. 1981. Forage ﬁshes and their salmonid
predators in Lake Michigan. Transactions of the American Fisheries Society.
110: 751-763.

Chapter 2

Energy Density of Key Predators and Their Prey in Lake Huron

Introduction

Lake Huron once supported one of the world’s largest freshwater ﬁsheries,
primarily in lake trout with average commercial yields of 2.4 million kg from 1912 to
1940 (Ebener et al. 1995). Overﬁshing coupled with the negative effects of sea
lamprey Petromyzon marinus predation initiated a collapse in Lake Huron piscivore
populations during the 1940’s (Christie 1974; Mills et al. 1993; Eshenroder et al.
1995). By the mid-1960’s native piscivores were rare in the upper Great Lakes and
management agencies began chemical treatment of streams to control sea lamprey
abundance and improve the lake ecosystem for salmonines (DesJardine et a1. 1995).
Since that time, restoration efforts have focused on rebuilding piscivore populations,
controlling exotic alewife, and promoting the recreational ﬁshery (DesJardine et al.
1995; Great Lakes Fishery Commission 2001).

Stocked salmon and trout provide substantial recreational, social, and
economic beneﬁts (Dann 1994; Bence and Smith 1999) and play an important role as
top predators in the lake ecosystem, primarily consuming exotic prey species (Christie
1974; Kitchell et al. 1994; Eby et al. 1995). However, with predator abundance
predominantly controlled through stocking, an important natural linkage between
predator abundance and prey availability may be disrupted. Such a situation may have
occurred in Lake Michigan. As stocked chinook salmon abundance increased in Lake
Michigan, their primary prey, alewife, increased in abundance (Madenjian et al. 2002).

Bioenergetics models suggested that chinook salmon predation on alewives caused

11

 

substantial annual alewife mortality there (Stewart et al. 1981; Stewart and Ibarra
1991). A subsequent trophic-dynamic modeling effort (Jones et al. 1993) suggested
that alewife might be driven to very low abundance in Lake Michigan at the salmonine
stocking levels of the 19803 and early 19903. During the late 19803 and early 19903,
chinook salmon in Lake Michigan experienced substantially elevated natural mortality
rates (Benjamin and Bence, in press), which may have been the result of a disease.

Bioenergetics models have been used in the Great Lakes for various purposes,
including estimation of predator forage demand (Stewart et al. 1981 and 1983; Eby et
al. 1995; Negus 1995), projection of changes in predator consumption with changes in
predator abundance (LaBar 1993; Negus 1995), prediction of predator-prey dynamics
(Jones et al. 1993), and examination of nutrient cycling within aquatic food webs (He
et al. 1993; Kraft 1993). Estimates of prey consumption can be calculated from
bioenergetics models (e. g., Kitchell et al. 1977; Stewart et al. 1981; Hewett and
Johnson 1995), which typically require the energy density of predators and prey as
input. For instance, the Wisconsin model (Hewett and Johnson 1995) requires input of
predator and prey energy density. While the production-conversion efﬁciency (Ney
1990) method, a simple method of estimating prey consumption, does not directly use
energy density data, id does require an estimate of the gross conversion efﬁciency
(GCE). Typically, this GCE is estimated through application of the more complex
bioenergetics models that do require energy density information.

Determining energy density is a time-consuming process that includes
collecting, grinding, drying, and bomb calorimetry of individual ﬁsh (Braﬁeld 1982).

Therefore, measurements of energy density are often not available for a particular

12

 

species or from a particular lake. Consequently, energy density values used in
bioenergetics models are borrowed from the literature for other species with similar
physiology (e. g., Hurley 1986; Madon and Culver 1993; Rudstam et al. 1995) or from
the same species occupying other lakes (LaBar 1993). However, energy densities may
not be interchangeable because prey ﬁsh condition varies with changes in the ﬁsh
community, food density, and climatic conditions (Rand et al. 1994).

The energy density for various predator and prey species within the Great
Lakes has been determined (Cummins and Wuycheck 1971; Rottiers and Tucker 1982;
Vondracek et al. 1996; Johnson et a1 1999) but data are generally lacking for species
from Lake Huron. Additionally, studies of Lakes Michigan, Superior, and Ontario
species have identiﬁed seasonal, regional, and annual variations in energy density
(Foltz and Norden 1977; Flath and Diana 1985; Hurley 1986; Rand et al. 1994;
Vondracek et al. 1996; Johnson et al. 1999; Madenjian et al. 2000). Because the Great
Lakes are interconnected and share many of the same predator and prey species, we
hypothesize that the energy density of Lake Huron species should be very similar to
that found in the other Great Lakes. Seasonal patterns in energy density often
observed for the introduced prey species, alewife and rainbow smelt, should also be
evident in Lake Huron. However, most studies did not ﬁnd strong trends for the
predator species so it seems more likely that these trends will also be missing from
Lake Huron species.

Our objectives were to (1) determine the energy density of Lake Huron
predators and prey; (2) identify seasonal and regional energy dynamics in these

species; and (3) evaluate the relationship between energy density and percent water

13

content. This study did not span multiple years and could not detect long-term
fluctuations in energy density. However, these data represent a fairly comprehensive
view of energy density for the primary predators and prey in Lake Huron not
previously available as well as provide an important baseline for comparison with

future energy density data.

Methods

From June 11, 1996 to September 24, 1997, the Michigan Department of
Natural Resources, the Chippewa/Ottawa Treaty Fishery Management Authority, the
Ontario Ministry of Natural Resources, and the Biological Research Division (USGS)
collected 707 ﬁsh representing the major predator and prey species in Lake Huron.
The predator species sampled were lake trout Salvelinus namaycush, burbot Lota lota,
chinook salmon Oncorhynchus tshawytscha, and walleye Stizostedion vitreum. Prey
species included alewife Alosa pseudoharengus, rainbow smelt Osmerus mordax,
bloater Coregonus hoyi, slimy sculpin Cottus cognatus, and ninespine stickleback
Pungitius pungitius. Each ﬁsh was placed intact in a plastic bag and then frozen.
Identiﬁcation tags placed with each ﬁsh included information on collector name, site,
date, time of day, length, and weight. Prior to grinding and drying the samples, length
and weight of each ﬁsh were assessed in the lab, and gender and maturity were
recorded. Each collection site was identiﬁed with a statistical district (Figure 2.1). To
analyze regional variation, the statistical districts were consolidated into four lake

regions: northern, central, southern, and Saginaw Bay (Figure 2.1).

14

An alternative collection procedure was sometimes applied to forage ﬁsh since
their small size did not always allow for accurate measurement of weight. Groups of
small forage ﬁsh of the same species and from the same collection site were either sorted
by size interval into separate bags or grouped together if the collector did not have time to
sort by size class. An identiﬁcation tag was placed in the bag with the same information
outlined above. To minimize weight loss, water was added to each bag, which was then

frozen or placed on ice until a freezer was available.

Fish collection spanned only one year; therefore we were not able to estimate
between-year differences, but we did estimate regional and seasonal variations in energy
density. Furthermore, we did not obtain sufﬁcient numbers of stickleback or sculpin to
statistically analyze variations in their energy density. These prey species do not
contribute signiﬁcantly to the diets of Lake Huron predators, with the exception of

burbot.

Energy density was evaluated for a sub-sample of 203 ﬁshes chosen to provide

coverage across the regions, months, and ﬁsh lengths. Each ﬁsh was ground, and
approximately 28 g of slurry was dried at 60-70°C to a constant mass. Approximately 1
g of each dried sample was processed in a bomb calorimeter (Braﬁeld 1982) to determine
the caloric content of the sample. The resulting energy density was expressed as cal-g‘l

dry weight and then converted to J og'1 wet weight using the water content of each
sample. The same grinding and drying process was applied to the remaining 504 ﬁsh, but

these samples were not processed in the calorimeter due to limited time and manpower.

15

While energy density was measured directly for only 203 ﬁsh, percent water,
which is predictive of energy density (Kitchell et al. 1977; Rottiers and Tucker 1982;
Hartman and Brandt 1995), was measured for all 707 ﬁsh. We used the data from the
203 ﬁsh for which energy density was measured directly to develop linear regression
models that predicted energy density from percent water. These models were used to
predict energy density for all 707 ﬁsh. These predicted energy densities were then used

in subsequent analyses.

To relate energy density (J og‘l wet weight) to percent water, four different
models were examined: (1) a single regression grouping all species; (2) a regression in
which predators and prey formed two groups with separate intercepts and slopes; (3) a
regression in which prey species were grouped together while predators were identiﬁed
with separate intercepts and a shared slope (here burbot and lake trout were grouped
together). We selected these models for consideration based on an initial examination of
scatter plots of energy density versus wet weight, which suggested different linear
relationships between predators and prey, more subtle difference in the level (intercept)
of the regression lines for predators, and little difference in the relationships between lake

trout and burbot or among prey species.

(SSEr —SSEf)/(pr _pf) ~
SSE]. /pf N (P'"Pf””f (1)

 

where SSE, and SSEf are the residual sum of squares from the reduced and the full
model, respectively, and p, and pf are the number of parameters in each model. The

models progress from the most reduced form with the fewest parameters to the most

16

complex form with the most parameters. Therefore, the extra sums of squares test (Neter
et al. 1996): was applied to models 1 and 2, and models 2 and 3 to determine whether the
added parameters were statistically different from zero (p < 0.05).

To explore regional and seasonal variations in energy density, an analysis of
variance (AN OVA) was conducted for each species using the energy density values that
were predicted from the percent water content. To avoid mixing measured energy
densities and estimated energy densities, the energy densities as predicted from the
models were used for all ﬁsh. Additionally, preliminary analysis showed that most of the
uncertainty in energy density for a particular category of ﬁsh stems from among ﬁsh
variation in percent water and not from the uncertainty in estimating the expected energy
density given the percent water.

The main effects were region (a) and month (,6). A ﬁsh’s size inﬂuences its

energy density so wet weight (w) was used as a covariate. The full model was

dij = p+a,. +8]. +aﬂij + 7/(w,j — w..) +80. (2)

where dij was the estimated energy density in J og'l wet weight for the ith region and jth
month; aﬂij was the interaction between region and month; and 7 was the coefﬁcient for
the linear regression of dij on Wij- Differences between levels of the main effects were

tested using Tukey’s pairwise comparison. Fish samples were not available for all

regions or months so subsets of the full model (2) were used as needed (Table 2.1).

 

Results
Linear relaiionshio between water content and energy density

Simple means for water content and energy density were calculated using all ﬁsh
of a species for which these were measured. Mean water content ranged from 44.7% to
81.9% (Table 2.2) and mean energy density was inversely related to the mean percent
water content among species (Figure 2.2). Forage ﬁsh species had higher water content
and lower energy density than predator species.

In all cases there were strong negative relationships between energy density and
percent water content (Table 2.3). The model that allowed separate intercepts by
predator species (with a single slope) and a single linear relationship for a combined prey
group (model 3, Figure 2.3) provided a signiﬁcantly better ﬁt to the data than the model
that only recognized one predator and one prey group (model 2), or the model that

assumed a single linear relationship for all species (model 1) (Table 2.4). This model

(Table 2.5) was then applied to all 707 samples to estimate the energy density in J og‘I
wet weight from percent water. Relatively few of these samples fell outside the

regression ranges (Table 2.2), and most of these lay close to the modeled ranges.

_A_r;2_tlvsis of energy density by species

The full ANOVA model could not be applied for all species due to variations
among species in data available for particular regions or months (Table 2.1). For the
models used in this analysis, the interaction between region and month was either not
estimable or not a signiﬁcant effect. Hence, this interaction was not included in any of

the ﬁnal models. The following results are presented ﬁrst by predator and then by prey

l8

species. Simple means are reported when the main effects were not signiﬁcant. Least
square means, which are adjusted for the other factors in the model, are reported when
the statistical model included other effects (i.e., regions, season, or wet weight as a
covariate).

Energy density of burbot did not vary regionally or seasonally and wet weight

was not a covariate (Table 2.6). The overall mean energy density of burbot was found to
be 5,630.0 Jog’l wet weight. Although the seasonal effect was not signiﬁcant, there is
some suggestion that burbot energy density was higher in March and October, averaging
5,825.2 J-g'1 wet weight, and lower from May through August, averaging 5,585.3 J cg"l
wet weight.

Chinook salmon samples were obtained in May through October. In this time
frame, neither regional (Figure 2.4) nor seasonal (Figure 2.5) differences in energy
density were detected but wet weight was a signiﬁcant covariate (Table 2.6). The mean
energy density of a 1.56 kg chinook salmon was 6,451.6 J og‘l wet weight.

Energy density of lake trout was found to vary regionally (Figure 2.4) and
seasonally (Figure 2.5); wet weight was also found to be a covariate (Table 2.6). Lake
trout exist in regionally distinct stocks with different characteristics such as age
composition and size-at-age (Sitar et al. 1999; Eschenroder et al. 1995). Energy density
of northern lake trout (6,767.5 Jog"1 wet weight) was statistically different from the

energy density of central (8,956.5 Jog"1 wet weight, p=0.0222, df=143) and southern

(8,378.3 J og'l wet weight, p=0.0026, df=143) lake trout. However, there was no

difference between the energy densities in the central and southern regions. Lake trout
energy density is lowest in April (6,232.82 J-g'l wet weight) and increased during the

19

summer (9,478.3 Jog'1 wet weight), dropping slightly through October (Figure 2.5). July
was the only month that was statistically different from all other months sampled.
The majority (44 out of 45) of walleye were taken from Saginaw Bay during the

months of August, September, and October. Mean energy density in August was 4,637.6
Jog'l wet weight, but this value was based on a single sample that came from the central
region of Lake Huron. Energy density was higher in September (6,564.2 Jog'1 wet

weight) and lower in October (6,305.9 J-g'l wet weight), but these differences were not
statistically signiﬁcant (p=0.2570, df=41). Wet weight was found to be a signiﬁcant
covariate (Table 2.6).

Alewife energy density was found to vary by region (Figure 2.4) and by month
(Figure 2.6) with wet weight as a covariate (Table 2.6). An interaction between month

and region could not be estimated due to lack of samples. Mean energy density of central
region alewife was lower (4,400.4 J-g'1 wet weight) than that of the southern region

alewife (5,138.1 J og'l wet weight); no samples from the northern region were available.
Alewife taken in the months of June through September were analyzed for seasonal

trends. Energy density did not differ between June and July, averaging 4,191.1 and

4,368.9 Jog'1 wet weight respectively. However, energy density in August was

statistically different from June (p<0.0001, df=175) and July (p<0.0001, df=175), with a

mean of 5,255.9 Jog'1 wet weight. The single sample from September was 5,260.9 J-g'1
wet weight.

For bloater, only the month (Figure 2.6) and wet weight covariate component of
the model were used because all but two samples came from the northern region (Table
2.6). Samples were obtained from January, March, May, and June. In the ANOVA, the

20

main effect of month was driven by the two June samples. With these two samples

eliminated from the AN OVA, month was not a signiﬁcant effect. While there appears to
be a large difference in mean energy density between May (6,020.2 Jog“1 wet weight) and

June (2,410.3 Jog'l wet weight), the limited number of June samples (N=2) makes this
difference very uncertain. A further uncertainty is that both June samples came from the
southern region but all other samples (N=34) came from the northern region. Additional
data are required to determine if the differences between May and June mean energy
densities are related to seasonal trends or regional differences.

Energy density of rainbow smelt varied seasonally (Figure 2.6) but not regionally

(Table 2.6). Rainbow smelt mean energy density in July was lower, 4,611.7 J og‘l wet
weight, than in May, June, or August. Only energy density during May and August were
statistically different from July.

There were only three stickleback samples available for our study. The mean
energy density of these samples was 5,194.2 J-g’l wet weight. Similarly, only one

sculpin sample was dried, with an estimated energy density of 4,635.5 J og‘l wet weight.

Discussion

Fish communities in the upper Great Lakes share many of the same species
and the hydrological connection between Lakes Huron and Michigan have led some
to consider them a single waterbody (Beeton and Saylor 1995). We anticipated that
energy densities of the Lake Huron species we sampled would be similar to
conspeciﬁcs from other Great Lakes, and most similar to those observed in Lake

Michigan. This was true to some extent. For instance, our estimates of mean

21

rainbow smelt energy density fell between those observed in Lakes Michigan (Rand
et a1. 1994) and Superior (Vondracek et al. 1996). Similarly, our estimates of burbot
mean energy density were comparable to those found in Lake Superior (Johnson et al.
1999). However, the energy densities for all other Lake Huron species were lower
than those reported from the other Great Lakes.

Energy density of salmonid predators from Lake Huron was lower than

published energy density from other Great Lakes. The mean reported lake trout energy

density was 10,294 J og'l wet weight (Rottiers and Tucker 1982; Johnson et al. 1999;

Madenjian and O’Connor 1999) while the mean energy density for chinook salmon was

6,678 J-g'l wet weight (Cummins and Wuycheck 1971; Rottiers and Tucker 1982).
These published energy densities are 20% and 13% higher than those found for Lake
Huron lake trout and chinook salmon. One exception to these patterns is the estimated
energy density of lean lake trout in Lake Superior (Johnson et al. 1999), which is about
5% lower than lake trout from Lake Huron. However, variations between lake trout
phenotypes (Henderson and Anderson 2002) and signiﬁcantly colder temperatures in
Lake Superior could account for these differences.

One possible reason for the low energy density of salmonids is the poor
condition of Lake Huron alewife, which had the lowest mean energy content of the
major prey species. Alewife is a major constituent in the diets of many top predators
including lake trout (43% of ages 1-6 and 66% for ages 7+) and chinook salmon
(73% for ages 2+). Lake Huron alewife exhibited much lower energy content than
estimates from Lakes Michigan and Ontario for earlier time periods (Rottiers and

Tucker 1982; Flath and Diana 1985; Rand et al. 1994). A pattern of declining alewife

22

 

energy density can be seen in these results (Figure 2.8). Our results ﬁt the pattern of
declining energy density and represent a continuation of that trend. Rand et al. (1994)
hypothesized reasons for the declining alewife energy density they observed
including density-dependent effects caused by an increasing alewife population or
lower lakewide productivity. During our sampling in 1996-1997, alewife abundance
in Lake Huron (Figure 2.7) was declining from a peak in 1994 making density-
dependent effects a less likely cause for their low energy content. Another possibility
is changes in benthic macroinvertebrates abundance that could limit consumption by
adult alewife. Diporeia, a macrobenthic organism with a high lipid content (Guiguer
and Barton 2002), is a primary constituent in alewife diets. Sampling in 1972 found
that it was abundant throughout Lake Huron but was virtually absent from the
southern portions of the main basin by 2000 (T. Nalepa, Great Lakes Environmental
Research Laboratory, Pers. Comm.). While there is some evidence that Diporeia
were declining in the shallow areas of Lake Huron during 1996 (Nalepa et al. 2003),
they were still abundant at a site in the middle of the southern basin (EPA monitoring
data).

A preponderance of lower energy density prey may be responsible for lower
predator growth in Lake Huron. When prey are energy-dense, fewer prey are required
to sustain predator growth. Conversely, predators must increase their daily ration of
low energy prey to maintain growth (Brett and Groves 1979). Chinook salmon
represent a major demand on forage and their abundance was increasing during our
sample collection period of 1996-1997. At the same time, diet information for chinook

salmon was difﬁcult to obtain due to the large number of void stomachs (J. Johnson,

23

Michigan Department of Natural Resources, Pers. Comm). While the cause of this is
not evident or easily explained by low prey abundance, it is possible our results partly

reﬂect low rates of predator feeding.

Regional Pgttems in Energy Density

Lake trout are known to exist in spatially separated subpopulations
(Eshenroder et al. 1995) and our results indicated that mean energy density of lake
trout varied by subpopulation (i.e., lake region). Although the lake regions are
contiguous, there is a measurable north-south gradient in water temperature that
appears to impact the growing seasons such that lake trout in the northern regions
grow slower (Eshenroder et al. 1995). Many physiological functions that affect
energy absorption, respiration, and growth depend on temperature (Brett and Groves
1979). Therefore, for lake trout populations that segregate by location but have
similar diet composition, we would expect a gradient in energy density with lower
values in the northern region, higher in the southern region, and intermediate in the
central region. Our analysis found this gradient of decreasing energy density with
latitude for lake trout (Figure 2.4). However, this regional relationship was not
strongly evident in any other species we sampled.

Lake trout energy density has been shown to be directly related to increasing
lipid concentrations (Rottiers and Tucker 1982) and lipids play a key role in
determining predator condition (Adams 1999). Madenjian et a1. (2000) found
variations in lipid concentration between nearshore and offshore lake trout with total

length < 600 mm but no variation in lipid concentration for larger lake trout. Similar

24

 

to the Lake Huron lake regions we used, their samples included sites in the
southeastern, northwestern, and mid-lake regions of Lake Michigan. They did not
ﬁnd a pattern of declining energy density with decreasing latitude. However, their
northernmost site was not as far north as the northern region from which our samples
came. Further more, the pronounced north-south cline in lake trout growth in Lake
Huron is not nearly as evident in Lake Michigan (Bence and Ebener 2002). The

mean energy density of lake trout in the northern region of Lake Huron was 8,190
Jog‘l wet weight, a value closer to the estimate of lean lake trout energy density in

Lake Superior of 7,788 J og'l wet weight (Johnson et al. 1999). This suggests that the
northern part of Lake Huron is more similar to Lake Superior than the lower northern

reaches of Lake Michigan.

Seasonal Patterns in EnergvDensity

Seasonal differences in energy density have been attributed to changes in diet
composition, energy density of food consumed, and reproductive status. Studies of
temperate ﬁshes have found the highest energy density values in the fall (October and
November) and the lowest in early spring (March to May) (Flath and Diana 1985 ;
Hayes and Taylor 1994; Rand et a1. 1994; Jonas et al. 1996; Encina and
GranadoLorencio 1997; Foy and Paul 1999; Pedersen and Hislop 2001). During the
winter or spawning seasons, many ﬁsh cease feeding, living off of stored energy
reserves, or dramatically reduce feeding due to colder water temperature and its
effects on digestion and metabolism (Adams 1999). Consequently, we expected low

energy density values at the beginning of the year, increasing through the fall months.

25

 

We detected these seasonal patterns for alewife, bloater, lake trout, rainbow smelt,
and walleye. However, walleye data were essentially limited to September and
October so an actual seasonal pattern cannot be determined.

Rainbow smelt energy density in Lake Huron also varied by month although
our samples were limited to the months of May through August (with a limited
number of samples from January). In Great Lakes, Foltz and Norden (1977) found
that the energy density of smelt in Lake Michigan increased from June to October,
and Vondracek et a1. (1996) noted an initial decrease in Lake Superior smelt energy
density in the spring, leading to an increase through September. In our samples,
smelt energy density was highest in May, declining slightly through June and July,
then increasing in August (Figure 2.6). If this August increase were to continue
through the fall months, this pattern would again place Lake Huron smelt energy
densities between those of Lakes Superior and Michigan. However, in Lake Oahe,
South Dakota, Bryan et al. (1996) found rainbow smelt energy density was the
highest in July, decreasing through the remainder of summer, suggesting that the
seasonal patterns in rainbow smelt energy density can vary across lakes.

Our chinook salmon samples were obtained from May through October making
it difﬁcult to suggest the pattern of energy density throughout the winter. However,
with the exception of the May value, energy density increased from June to October
(Figure 2.5) in accordance with other observed seasonal patterns, although the pattern
was not statistically signiﬁcant using our model (Table 2.6). The least square means
were roughly constant during the summer months and only increased during September

and October. However, there were only four ﬁsh sampled during this time period.

26

While the pattern we observed was not particularly strong, it is similar to the pattern
found in Lake Michigan where energy content was higher in the fall and lower in the

following spring (A. Peters, unpublished data). Similarly, chinook salmon energy
density increased from 4,940 J og'l wet weight in May to 5,987 J-g‘l wet weight in

July, and to 6,824 Jog'l wet weight in September in Lake Oahe South Dakota (Bryan et
al. 1996). This pattern may have been obscured in our data due to our relatively small
sample sizes and large variations among individual ﬁsh. Alternatively, it is possible
that energy density did not increase over the summer in our study because of poor

feeding condition for chinook salmon in Lake Huron during 1997.

Predicting Energy Density from Percent Water Content

There are many reasons for monitoring changes in energy density of both
predator and prey species. First, declines in ﬁsh growth have been attributed to
reductions in the nutritional content of prey (Boisclair & Leggett 1989; Anthony et al.
2000) resulting from lower energy density. Second, population abundance may also be
impacted when declines in energy density adversely affect growth, reproduction, and
survival of individuals (Henderson and Wong 1998; Holey et al. 1998; Adams 1999).
For example, in the late 19803, chinook salmon abundance in Lake Michigan was
reduced by over 50%. This was attributed to nutritional stress caused by low prey
availability thought to have been initiated by poor overwinter survival of alewife that
entered winter with low lipids levels (Holey et al. 1998). Last, bioenergetics models

require estimates of energy density for both predator and prey species. These models

27

 

provide important beneﬁts to fishery managers in terms of understanding predator
forage demand (e. g., Stewart et al. 1983; Rudstam et al. 1994; Rand and Stewart 1998).

Determining energy density can be costly in terms of manpower and money
because samples must be processed using bomb calorimetry with equipment not
normally available to ﬁshery managers (Hartman and Brandt 1995), and energy density
can vary by season, by location, and over time, requiring frequent and ongoing
sampling. One simpler alternative to estimate energy density makes use of the strong
negative relationship between percent water content and energy density, which has
been observed in many ﬁsh species (Kitchell et al. 1977; Rottiers and Tucker 1982;
Hartman and Brandt 1995; Jonas et al. 1996; Schreckenbach et al. 2001). This
relationship held true in Lake Huron and we further noted that predator species had a
lower percentage of water and higher energy density than prey species (Figure 2.2).
Since processing a ﬁsh sample to determine water content is less expensive than
determining its energy content, these measures could be done more often than direct
measures of energy content. While our estimates of the energy density — percent water
relationship (Table 2.5) were similar to estimates from studies in the other Great Lakes
(Rottiers and Tucker 1982; Johnson et al. 1999), the estimated energy density for many
species is lower reﬂecting recent Lake Huron conditions.

The primary focus of this study was to determine energy density for the major
predator and prey species in Lake Huron for use in bioenergetics models (Chapter 3).
In our analyses, we found temporal and spatial differences in energy density that
varied within the lake and across the Great Lakes. This suggests that borrowing

energy density values from other studies may not provide the most accurate or

28

contemporaneous data. Our results point the way towards some pragmatic
approaches to more frequent evaluations of energy status. First, water content can be
used to predict energy density, although the validity of the relationship for a speciﬁc
location should periodically be checked. Second, in large lakes energy density may
not vary much spatially for widely ranging species (e. g., chinook salmon), or might
demonstrate predictable spatial patterns (e. g., lake trout) so that less comprehensive
spatial sampling might be sufﬁcient. However, since these observations are based on
our limited sampling of energy density in Lake Huron, further sampling and energy
density analyses could help determine the validity of these patterns. Last, the
observed decline in alewife energy density and the effects this may have on predator

condition indicates the need for frequent monitoring of energy density.

29

 

Literature Cited

Adams SM. 1999. Ecological role of lipids in the health and success of ﬁsh
populations. In M.T. Arts and B. C. Wainman (eds.), Lipids in freshwater
ecosystems, Springer: New York.

Anthony, J. A., D. D. Roby, and K. R. Turco. 2000. Lipid content and energy density of
forage ﬁshes from the northern Gulf of Alaska. Journal of Experimental Marine
Biology and Ecology 248: 53-78.

Beeton, A.M. and J .H. Saylor. 1995. Limnology of Lake Huron. From: The Lake
Huron Ecosystem: Ecology, Fisheries and management. Eds. M. Munwar, T.
Edsall, J .Leach. pp 1-37.

Bence, J. R. and K. D. Smith. 1999. An overview of the recreational ﬁsheries of the
Great Lakes. In Taylor, W. W. and P. Ferreri (eds.), Great Lakes Fisheries Policy and
Management: A Binational Perspective, Michigan State University Press.

Bence, J. R. and M. P. Ebener (eds). 2002. Summary status of lake trout and lake
Whiteﬁsh populations in 1836 treaty-ceded waters of Lakes Superior, Huron and
Michigan in 2000, with recommended yield and effort levels for 2001. Technical
Fisheries Committee, 1836 Treaty-Ceded Waters of Lakes Superior, Huron and
Michigan.

Benjamin, D. M. and J. R. Bence. 2000. Statistical catch-at-age assessment of Chinook
salmon in Lake Michigan, 1985-1996. Michigan Department of Natural
Resources, Fisheries Division, Fisheries Research Report.

Boisclair, D. and W. C. Leggett. 1989. Among population variability of ﬁsh growth: 11
Inﬂuence of prey type. Canadian Journal of Fisheries and Aquatic Sciences 46:
468-482.

Braﬁeld, A. E. and M. J. Llewellyn. 1982. Animal Energetics. Chapman and Hall: New
York.

Brett J. R. and TDD. Groves. 1979. Physiological energetics, p 280-345. In W. S.
Hoar, D. J. Randall, and J. R. Brett [eds.] Fish Physiology. Academic Press: New
York , NY.

Bryan, S.D., C.A. Soupir, W.G. Duffy, and CE. Freiburger. 1996. Caloric densities of

three predatory ﬁshes and their prey in Lake Oahe, South Dakota. Journal of
Freshwater Ecology 11(2): 153-161.

30

 

Christie, W. J. 1974. Changes in the ﬁsh species composition of the Great Lakes.
Journal of the Fisheries Research Board of Canada 31: 827-854.

Cummins, K. W. and J. C. Wuycheck. 1971. Caloric equivalents for investigations in
ecological energetics. Mitteilungen Internationale Vereinigung fur theoretische
und angewandte Limnologie 18.

Damn, S. L. 1994. The life of the lakes: A guide to the Great Lakes ﬁshery. Michigan
Sea Grant Extension Bulletin #E-2440, East Lansing, Michigan.

DesJardine , R. L., T. K. Gorenﬂo, R. N. Payne and J. D. Schrouder. 1995. Fish-
community objectives for Lake Huron. Great Lakes Fishery Commission Special
Publications 95-1, pp 38.

Ebener, M. P., J. E. Johnson, D. M. Reid, N. P. Payne, R. L. Argyle, G. M. Wright, K.
Kruger, J. P. Baker, T. Morse and J. Weise. 1995. Status and future of Lake
Huron ﬁsh communities. In Munawar, M., T. Edsall and J. Leach (eds.), The
Lake Huron ecosystem: Ecology, Fisheries and Management. Ecovision World
Monograph Series, SPB Academic Publishing, Amsterdam, The Netherlands,
pp.125-170.

Eby, L. A., L. G. Rudstam and J. F. Kitchell. 1995. Predator responses to prey
population dynamics: An empirical analysis based on lake trout growth rates.
Canadian Journal of Fisheries and Aquatic Sciences 52(7): 1564-1571.

Encina, L. and C. GranadoLorencio. 1997. Seasonal variations in the physiological
status and energy content of somatic and reproductive tissues of chub. Journal of
Fish Biology 50(3): 51 1-522.

Eshenroder, R. L., N. R. Payne, J. E. Johnson, C. Bowen H, and M. P. Ebener. 1995.
Lake trout rehabilitation in Lake Huron. Journal of Great Lakes Research 21
(Supplement 1): 108-127.

Flath, L. E. and J. S. Diana. 1985. Seasonal energy dynamics of the alewife in
southeastern Lake Michigan. Transactions of the American Fisheries Society
114: 328-337.

Foltz, J. W. and C. R. Norden. 1977. Seasonal changes in food consumption and energy
content of smelt in Lake Michigan. Transactions of the American Fisheries
Society 106(3): 230-234

Foy, RJ. and A.J. Paul. 1999. Winter feeding and changes in somatic energy content of
age-0 Paciﬁc herring in Prince William Sound, Alaska. Transactions of the
American Fisheries Society 128: 1193-1200.

31

Great Lakes Fishery Commission. 2001. Strategic Vision of the Great Lakes Fishery
Commission for the First Decade of the New Millennium. Ann Arbor, MI. 40 p.

Guiguer, K. R. R. A. and D. R. Barton. 2002. The trophic role of Diporeia (Amphipoda)
in Colpoys Bay (Georgian Bay) benthic food web: a stable isotope approach.
Journal of Great Lakes Research 28(2): 228-239.

Hartman, K. J. and S. B. Brandt. 1995. Estimating energy density of ﬁsh. Transactions
of the American Fisheries Society 124: 347-355.

Hayes, DB. and W.W. Taylor. 1994. Changes in the composition of somatic and
gonadal tissues of yellow perch following white sucker removal. Transactions of
the American Fisheries Society 123:204-216.

He X., J. F. Kitchell, S. R. Carpenter, J. R. Hodgson, D. E. Schindler, K. L. Cottingham.
1993. Food-web structure and long-term phosphorus recycling - a simulation-

model evaluation. Transactions of the American Fisheries Society 122 (5): 773-
783.

Henderson, B.A. and D.M. Anderson. 2002. Phenotypic differences in buoyancy and
energetics of lean and siscowet lake charr in Lake Superior. Environmental
Biology of Fishes 64: 203-209.

Henderson, B. A. and J. L. Wong. 1998. Control of lake trout reproduction: role of
lipids. Journal of Fish Biology 52: 1078-1082.

Hewett. S. W. and B. L. Johnson. 1995. Fish Bioenergetics Model 3. University of
Wisconsin Sea Grant Institute, WIS-SG-91-250.

Holey M.E., R.F. Elliott, S.V. Marcquenski, J .G. Hnath, K.D. Smith. 1998. Chinook
salmon epizootics in Lake Michigan: Possible contributing factors and
management implications. Journal of Aquatic Animal Health 10: 202-210.

Hurley, D. A. 1986. Growth, diet, and food consumption of walleye: an application of
bioenergetics modeling to the Bay of Quinte, Lake Ontario, population. In C. K.
Minns, D. A. Hurley, and K. H. Nicholls [eds.] Project Quinte: point-source
phosphorus control and ecosystem response in the Bay of Quinte, Lake Ontario.
Canadian Special Publication of Fisheries and Aquatic Sciences 86: 0706-6481.

Johnson T. B., D. M. Mason, S. T. Schram, J. F. Kitchell. 1999. Ontogenetic and
seasonal patterns in the energy content of piscivorous ﬁshes in Lake Superior.
Journal of Great Lakes Research 25 (2): 275-281.

Jonas, J .L., C.E. Kraft, and TL Margenau. 1996. Assessment of seasonal changes in
energy density and condition in Age-0 and Age-1 muskellunge. Transactions of
the American Fisheries Society 125: 203-210.

32

Jones, M. L., J. F. Koonce, and R. O’Gorman. 1993. Sustainability of hatchery-
dependent salmonine ﬁsheries in Lake Ontario: The conﬂict between predator

demand and prey supply. Transactions of the American Fisheries Society 122:
1002-1018.

Kitchell, J. F., J. J. Magnuson, and W. H. Neill. 1977. Estimation of caloric content for
ﬁsh biomass. Environmental Biology of Fishes 2(2): 185-188.

Kitchell, J. F., L. A. Eby, X. He, D. E. Schindler and R. A. Wright. 1994. Predator-prey
dynamics in an ecosystem context. Journal of Fish Biology 45(Suppl. A): 209-
226.

Kraft, C. E. 1993. Phosphorus regeneration by Lake Michigan alewives in the mid-
19703. Transactions of the American Fisheries Society 122 (5): 749-755.

Labar G. W. 1993. Use of bioenergetics models to predict the effect of increased lake
trout predation on rainbow smelt following sea lamprey control. Transactions of
the American Fisheries Society 122 (5): 942-950.

Madenjian, C.P., G. L. Fahnenstiel, T.H. Johengen, T.F. Nalepa, H.A. Vanderploeg,
G.W. Fleischer, P.J.Schneeberger, D.M. Benjamin, E.B. Smith, J .R. Bence, E.S.
Rutherford, D.S. Lavis, D.M. Robertson, D.J. Jude, and MP. Ebener. 2002.
Dynamics of the Lake Michigan food web, 1970-2000. Canadian Journal of
Fisheries and Aquatic Sciences 59: 736-753.

Madenjian, GP. and D.V. O’Connor. 1999. Laboratory Evaluation of a Lake Trout
Bioenergetics Model. Transactions of the American Fisheries Society 128: 802-
814.

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

spatial, ontogenetics, and long-term trends. Journal of Great Lakes Research
26(4): 427-444.

Madon, S. P. and D. A. Culver. 1993. Bioenergetics Model for Larval and Juvenile
Walleyes: An In Situ Approach with Experimental Ponds. Transactions of the
American Fisheries Society 122: 797-813.

Mills, E. L., J. H. Leach, J. T. Carlton and C. L. Secor. 1993. Exotic species in the Great
Lakes: A history of biotic crises and anthropogenic introductions. Journal of Great

Lakes Research 19:1-54.

Nalepa, T. F., Fanslow, D. L., Lansing, M. L., and Lang,G. A. 2003. Trends in the
benthic macroinvertebrate community of Saginaw Bay, Lake Huron, 1987-1996:

33

responses to phosphorus abatement and the zebra mussel, Dreissena polymorpha.
Journal of Great Lakes Research 29: 14-33.

Negus, M. T. 1995. Bioenergetics modeling as a salmonine management tool applied to
Minnesota waters of Lake Superior. Northern American Journal of Fisheries
Management 15:60-78.

Neter, J ., M.H. Kutner, C.J. Nachtsheim, and W. Wasserman. 1996. Applied Linear
Regression Models. 3rd ed. Chicago: Irwin, p 260-267.

Ney, J. J. 1990. Trophic economics in ﬁsheries: Assessment of demand—supply
relationships between predators and prey. Reviews in Aquatic Sciences 2: 55-81.

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

Rand P. S., B. F. Lantry, R. O’Gorman, R. W. Owens, D. J. Stewart. 1994. Energy
density and size of pelagic prey ﬁshes in Lake Ontario, 1978-1990: implications

for salmonine energetics. Transactions of the American Fisheries Society 123(4):
519-534.

Rand PS. and DJ. Stewart. 1998. Prey ﬁsh exploitation, salmonine production, and
pelagic food web efﬁciency in Lake Ontario. Can. J. Fish. Aquat. Sci. 55:318-
327.

Rottiers, D. V. and R. M. Tucker. 1982. Proximate composition and caloric content of
eight Lake Michigan ﬁshes. US. Fish and Wild]. Serv. Tech. Paper 108, 9 p.

Rudstam, L. G., P. E. Peppard, T. W. Pratt, R. E. Bruesewitz, D. W. Coble, F. A. Copes
and J. F. Kitchell. 1995. Prey consumption by the burbot (Lota lota) population

in Green Bay, Lake Michigan, based on a bioenergetics model. Canadian Journal
of Fisheries and Aquatic Sciences 52: 1074-1082.

Schreckenbach K., R. Knosche, and K. Ebert. 2001. Nutrient and energy content of
freshwater ﬁshes. Journal of Applied Ichthyology 17(3): 142-144.

Sitar, S. P., J. R. Bence, J. E. Johnson, M. P. Ebener, and W. W. Taylor. 1999. Lake
trout mortality and abundance in southern Lake Huron. North American Journal
of Fisheries Management 19: 998-900.

Smith, S. H., H. J. Buettner, and R. Hile. 1961. Fishery statistical districts of the Great
Lakes. Great Lakes Fishery Commission Technical Report No. 2.

Stewart, D. J ., J. F. Kitchell, and L. B. Crowder. 1981. Forage ﬁshes and their salmonid

predators in Lake Michigan. Transactions of the American Fisheries Society 110:
751-763.

34

Stewart, D. J ., D. Weininger, D. V. Rottiers, and T. A. Edsall. 1983. An energetics
model for lake trout, Salvelinus namaycush: application to the Lake Michigan
population. Canadian Journal of Fisheries and Aquatic Sciences 40: 681-698.

Stewart, D. J. and M. Ibarra. 1991. Predation and production by salmonine ﬁshes in
Lake Michigan, 1978-88. Canadian Journal of Fisheries and Aquatic Sciences 48:
909-922.

Vondracek, B., B. D. Giese, and M. G. Henry. 1996. Energy density of three ﬁshes from
Minnesota waters of Lake Superior. Journal of Great Lakes Research 22:757-764.

35

Table 2.1 - Mean and standard deviation of wet weight (kg), and percent water content

for predator and prey species collected in Lake Huron during 1996-1997. Samples were

collected in Saginaw Bay (B) and the northern (N), central (C), and southern (S) regions

of Lake Huron. Months represent the numerical value for each month a sample was

collected, with January as month 1.

 

 

 

Wet Weight (kg) Percent water
Species Location Months N Mean Std dev Mean Std dev
Predators
Burbot N,C,S 3,5,6,8,10 86 1.40 0.86 76.33 3.06
Chinook
salmon N,C,S 5,6,7,8,9,10 96 1.56 2.02 75.14 4.41
Lake
Trout N,C,S 1 ,3,4,5,6,7,10 153 1.63 1.21 68.94 4.89
Walleye C,B 8,9,10 45 1.42 0.74 71.74 2.23
Prey
Alewife C,S 6.7.89 181 0.02 0.02 80.89 3.91
Bloater N,S 1,3,5,6 36 0.15 0.07 76.07 5.05
Rainbow
smelt N,C,S 1,5,6,7,8 106 0.03 0.05 77.39 5.15
Sculpin N 7 1 < 0.003 --- 79.26 ---
Stickleback C 7 3 0.00 0.00 69.11 16.85
36

Table 2.2 — Water content (mean with :1 sd) for ﬁsh samples processed in the bomb

calorimeter and those that were only ground and dried.

 

Fish processed with bomb

Fish ground and dried only

 

 

 

 

 

calorimetry (N=504)
(N=203)
Outside
Species Percent water Percent water of model
N congnt N cont_e_nt range
Burbot 25 75.2 i 1.4 61 76.8 :I: 0.5 3.3%
(65.9 - 81.6) (69.8 - 82.9)
iChinook 49 73.1 i 0.9 47 77.2 i 1.0 14.9%
salmon (64.0 - 79.5) (67.1 - 86.9)
Lake trout 25 67.8 i 1.9 128 69.2 i 0.7 3.1%
(55.3 - 78.1) I (60.0 - 85.7)
Walleye 25 71.3 :1: 0.7 20 72.3 i 0.9 15.0%
(67.2 - 74.8) (68.1 - 76.4)
i
Alewife 26 81.9 i 1.2 1155 80.7 J; 0.5 5.2%
(74.8 - 90.6) I (63.0 - 91.0)
I
Bloater 25 76.1 i- 1.6 11 75.9 i 3.3 9.1%
(66.9 - 85.9) (68.3 - 88.2)
Scuplin --- I 1 79.3 ----
Rainbow smelt 25 78.9 i 1.7 ' 81 76.9 i 1.0 7.4%
(68.7 - 92.9) (64.3 - 93.5)
Stickleback 3 77.2 i 9.2 I 1 44.7 ----
(72.3 - 83.1) i

 

37

Table 2.3 — Regression model results for the relationship of energy density to percent

water content.

 

 

Model R2 F p
(1) Overall 0.9318 F 120,: 2745.6 <.0001
(2) Grouped as predators or prey 0.9415 F 3199: 1067.28 <.0001

(3) Prey group and separate predator intercepts, 0.9573 F 5,197: 884.11 <.0001
with burbot and lake trout combined

 

Table 2.4 — Extra sums of squares criterion for models 1, 2, and 3 where SSE, and SSEf
are the residual sum of squares from the reduced and the full model, respectively, and df,

and dff are the degrees of freedom in each model.

 

Model SSE, df, SSEf dff F p

 

Overall vs. groups 4.94E+07 201 4.24E+07 199 16.49 <.0001
(model 1 vs model 2)

Groups vs. prey group with

separate predator intercepts, 4.24E+07 199 3.09E+07 197 36.59 <.0001
with burbot and lake trout

combined

Qnodel 2 vs model 3)

 

38

 

Table 2.5 — Intercepts and slopes for ﬁnal model (model 3) used to predict energy density

from percent water using the equation: E = Cl + B W, where E is energy density in Jog‘1
wet weight and W is the percent water content. Model 3 assumed a common slope for all
predators but allowed intercepts to vary, and allowed a different linear relationship for
prey than predators (different intercept and slope) but assumed the same relationship for

all prey species.

 

 

Species or group Intercept Slope
(a) G3)
Prey 26,442.37 -275.13
Burbot and lake trout 32,077.70 -346.49
Chinook salmon 31,609.71 -346.49
Walleye 31 ,294.05 -346.49

 

39

Table 2.6 —ANOVA main effects (month and region) and covariate (predator wet weight)

for each sampled Lake Huron species. Least square means are shown with one standard

error in parentheses. Missing entries represent effects that could not be tested due to

insufﬁcient samples.

 

 

Parameter Alewife Bloater Burbot Chinook Lake trout Rainbow Walleye
salmon Smelt
Overall Mean 4,187.3 5,563.2 5,630.0 5,575.0 8,189.7 5,151.0 6,435.3
Region:
df 1, 175 -- 2, 83 2, 87 2, 143 2, 98 --
F 9.35 [a] -- 0.69 [b] 0.34 4.92 0.86 [a]--
p-value 0.0026 -- 0.5060 0.7095 0.0086 0.4266 --
North -- -- 5693.6 5,736.7 6,767.5 -217.6 --
(257.7) (324.8) (334.8) (726.1 )
Central 4,400.4 -- 5834.5 5,481.7 8,956.5 385.1 --
(239.1) (231 .9) (299.1) (798.5) (396.3)
South 5,138.1 -- 5517.9 5,820.8 8,378.3 0 --
(322.6) (153.4) (360.3) (431 .3)
Month:
(11 3, 175 3, 30 4, 79 5, 87 6, 143 4,100 2,41
F 11.11 3.92 1.44 1.24 2.51 3.63 [d] 3.97
p-value <0.0001 0.0179 0.2301 0.2972 0.0242 0.0083 0.0266
January -- 5,213.9 -- -- 8,253.9 4758.7 --
(440.3) (776.2) (402.5)
March -- 5,244.6 7,402.8 -- 8,426.6 -- --
(720.9) (796.9) (1,011.1)
April -- -- -- -- 6,232.8 -- --
(1,401.6)
May -- 6,020.2 5,166.4 6,372.5 7,812.8 6096.3 --
(266.1 ) (394.9) (432.6) (371 .2) (407.5)
June 4,191.1 2,410.3 4,820.5 5,145.4 7,534.8 5048.1 --
(132.9) (1 1 1 1.0) (393.1) (334.8) (543.4) (296.9)
July 4,368.9 -- -- 5,327.3 9,478.3 461 1.7 --
(1 59.06) (434.3) (492.6) (258.9)
August 5,255.9 -- 4,597.4 5,483.1 -- 5516.5 4637.6
(180.37) (712.6) (280.5) (232.3) (709.7)
September 5,260.9 -- -- 5,439.9 -- -- 6564.2
(932.9) (466.9) (130.9)
October -- -- 7,406.4 6,310.3 8,499.5 -- 6305.9
(841 .9) (667.9) (380.3) (182.4)
Covariate
df 1.175 1,30 1,87 1,143 1,100 1,41
F 12.26 4.75 [c] 28.35 58.53 15.80 [a] 4.66
p-value 0.0006 0.0373 <0.0001 <0.0001 <0.0001 0.0368
Overall mean 0.017 0.151 1.559 1.635 0.030 1.417

lab wet wt

[a] Not enough samples for analysis of this main effect
[b] The main effect of month was not significant so the model was refit without it.

[c] Covariate not signiﬁcant; model reﬁt without it

[(1] The main effect of region was not significant so the model was refit without it.

40

 

Region

Northern
Region 0H2
Central /
ﬂ 0H3
Southern

Saglnaw
Bay F

Figure 2.1 — Statistical districts grouped into lake regions (north, central, south, and
Saginaw Bay) for the regional analysis of energy density. Statistical districts in the US

use MH labels while Canadian waters are labeled with OH (Smith et al. 1961).

41

Mean energy denisty (Jig wet weight)

8,500 a

8.000 7 Lake trout
7,500 -
7,000 1
6500 _j Walleye‘

6,000 *

5,500 ~

 

Chinook salmon‘ Burbot

Bloater Stickleback

 

5,000 — Rainbow smelt
‘ ‘Sculpin
4,500 ~
. ‘ Alewife
4,000 . . - +4 . . . . . r . . . a
65 70 75 80 85

Mean percent water

Figure 2.2 -- Relationship between mean percent water content and mean energy density

for all sampled Lake Huron species.

42

 

 

 

 

 

 

11,000 - A. Prey species
10,000 ~
9,000 -
8 .
3 8,000
g 7,000 -
E 6,000 -
2
8 5,000 -
> 4,000
,9,’ 1
3,000 ~
.5
2'00" ‘ —Predicted
1’000 ‘ A Observed
o as
65 70 75 80 85 90
Percent Water Content
“Doc ‘1 IMF-1...: . 1:161 B. Predator species
10,000— '3' ‘~ _, _ 0.0
.__ _
9,000 ~ ‘~ .
on ' ~ .
g 8,000 1 ’t. -_
‘5‘ I‘- D
g 7,000 - I 'I
O Ajf‘g" ' q
9%. 6,000 ‘ ‘78, A 1’17,“
g 5,000 — . '* ht. ,
>. 4 000 - Predicted burbot and lake trout “ - a" D
g ' - - - - Predicted chinook salmon " . ~
t5 3’000 I '"~---- Predicted walleye m.
2,000 - 0 Observed burbot and lake trout
1 000 , 0 Observed chinook salmon
0 A Observed walleye
58 63 68 73 78

Percent Water Content

Figure 2.3 - Relationship between energy density and percent water content for predators

and prey in Lake Huron. [Note: the x-axis origins are not continuous from zero.]

43

”if?

g

5,000 —

Mean energy denslty (J/g wet welght)
9

 

Alewife

12,000 a

—A

0,000 -

5%

Mean energy denslty (J/g wet welght)
.0,

 

4,000 -
2,000 -
0 ’1
Burbot
Cl Northern

 

 

 

 

A. Prey species

   

 

 

 

 

 

 

Bloater

 

 

l

Smelt Sculpin Stickleback

B. Predator species

 

Chinook Lake trout Walleye

Central

I Southern El SaginawBay

Figure 2.4 - Mean regional energy density of the key predators and their prey in Lake

Huron during 1997. Least squares means are shown with one stande error.

A. Chinook Salmon

; ----------- .---.

 

 

0 I I l I r I I I I I l l

JFMAMJJASOND

B. Lake Trout
11,000 h

A10,000 -
9,000 — - -. ,

8,000~{'_- i”"
7,000 —
6,000 -
5,0001
4,000 4
3,0001
2,000 -
1,000 -

O l | I l l | l T I H I l

JFMAMJJASOND
Months

Energy denisty (J/g wet weight

 

 

Figure 2.5 — Mean seasonal energy density of the chinook salmon and lake trout in Lake

Huron, 1996-1997. Least squares means are shown with one standard error.

45

Energy density (J/g wet Energy density (J/g wet

Energy density (J/g wet

weight)

weight)

weight)

7,000 -
6,000 ~
5,000 -
4,000 1
3,000 ~
2,000 1
1,000 1

 

A. Alewife

.-
-.-.-
.-
.

 

7,000 -
6,000 —
5,000 1
4,000 -
3,000 1
2,000 -
1,000 -

 

TI I I I I I I I

J F M A M J J A
B.Bloater

T I I

SOND

 

8,000 1
7,000

6,000 ‘
5,000 1
4,000 -
3,000 ~
2,000 ~
1,000 ~

 

I I I I I I I I I I I I

FMAMJJASOND

R ‘AﬁOHA

J

C. Rainbow Smelt

if" xix/”M

 

I I I I I I I I I I I I

F M A M J J A S O N D
Months
- - . - lnterpolated

J

—0— Data

 

Figure 2.6 — Mean seasonal energy density of the primary prey species in Lake Huron

during 1996-1997. Least squares means are shown with one standard error.

46

111

6.0E+09

1

5.0E+09 . .. j

111411111
It

4.0E+09

1

'1
I
I g \ ‘v.
"'4. 1 "
1»: I I .
3.0E+09 . , . . .
....‘,
‘ 1 5.!

111

2.0E+09

Prey species abundance

1.0E+09

 

111L1 rite-1

 

0.0E+00 l I ' l l T T r I Y T T I I I i r 1 1 1 r i r Tﬁ
1974 1 978 1982 1986 1990 1994 1 998

Year
—o—Alewife - -e~ - Rainbow smelt

Figure 2.7 — Annual alewife and rainbow smelt abundance in the main basin of Lake

Huron during 1974-1998.

47

8,000 —

7,000 -

6,000 ~

5,000 1

 

4,000 -
(01)
3,000 —

2,000 -

Energy density Jlg wet weight

1 .000 -

 

0 T I r r
1965 1970 1975 1980 1985 1990 1995 2000

Last year of sampling

 

I T I

Figure 2.8 — Pattern of declining alewife energy density within the Laurentian Great
Lakes. Mean energy density from each study is plotted against the last year of the study

where (a) Rottiers and Tucker 1982; (b) Flath and Diana 1985; (c) Rand et al. 1994; (d)

this study.

48

Chapter 3

Recent and projected estimates of forage ﬁsh consumption by key predators
in the main basin of Lake Huron

Introduction

In non-managed systems, interactions between predator and prey populations can
potentially regulate the abundance of predator populations. However, in a hatchery-
dependent system, balancing predator forage demand and prey ﬁsh availability becomes a
major concern for ﬁshery managers. In Lake Huron, for example, overﬁshing and
parasitism by introduced sea lamprey Petromyzon marinus were the principal causes for
the collapse of the ﬁshery during the 19403 (Smith 1972; Eshenroder and Bumham-
Curtis 1999). Management reacted with efforts to control sea lamprey and by stocking
hatchery-reared chinook salmon Oncorhynchus tshawytscha and lake trout Salvelinus
namaycus (Ebener et al. 1995). Although some salmonines reproduce naturally (Ebener
et al. 1995), stocked ﬁsh make up a majority of the recruitment for all species and nearly
all recruitment of native lake trout (Eshenroder et al. 1995). This hatchery-dependent
system may have disrupted the natural feedbacks between predator abundance and the
dynamics of their prey, raising the possibility of overreaching the productive capacity of
the prey ﬁsh base (Kitchell and Crowder 1986; Eby et al. 1995). An inadequate forage
base may lead to declines in predator growth, delays in reproduction, and reduced
survival (e. g. Oglesby 1977; Boisclair and Leggett 1989; Rand et a1. 1994; Holey et al.

1998).

49

Understanding predator forage demand requires knowledge of individual
consumption rates and population dynamics. The consumption rate of an individual
ﬁsh can be estimated from gastric evacuation rates (e. g., Swenson and Smith 1973),
laboratory feeding experiments (e. g., Boisclair and Sirois 1993; Elliott and Hurley
2000), or by applying bioenergetics models (e.g., Stewart and Ibarra 1991; Rand et al.
1994). While there are drawbacks to each of these methods (Ney 1990),
bioenergetics models have been widely used to promote understanding of predator
consumption (e.g., Stewart et al. 1983; Hurley 1986; Negus 1995) and have often
been found to be representative of actual consumption given appropriate input
variables (Rice and Cochran 1984; Petersen and Ward 1999; Schaeffer et al. 1999;
Madenjian et al. 2000)

A bioenergetics model provides a method for estimating food consumption
utilizing a conceptual model that relates water temperature to consumption and
growth. An individual organism’s assimilation and utilization of energy from food is
partitioned into energy for growth (B), metabolism, and waste losses (Adams and

Breck 1990; Ney 1993; Hewett and Johnson 1995)

dB
—=C— R+F+U
Bdt ( ) (I)

where consumption and respiration (C, R) are temperature and size dependent while
egestion and excretion (F, U) are functions of consumption. Using this methodology,
growth integrates the feeding rate over time so short-term variability in food
availability, temperature, etc. are minimized. Fish growth is denoted as an increase in

body weight, which is the simplest measure to obtain for an energy budget. For a

50

given temperature and ﬁsh size, the energy budget can be solved to determine the
amount of food eaten to produce the observed growth.

Bioenergetics models have been applied to Lake Michigan salmonine predators to
establish the importance of these predators and their impact on prey communities (e. g.,
Kitchell et al. 1977; Stewart et al. 1981; Stewart and Ibarra 1991; Hansen et al. 1993;
Kitchell at al. 1994). The application of bioenergetics models also aided the
parameterization of the SIMPLE model (Jones et al. 1993), which played a role in the
decision to reduce stocking in Lake Ontario (Lange et al. 1995). There have also been
many other applications of bioenergetics models including estimating walleye
consumption (Hurley 1986), evaluating trends in forage ﬁsh predation (Eby et al. 1995),
and investigation of PCB, DDE, and mercury dynamics in Lakes Ontario (Borgmann and
Whittle 1992) and Michigan (Madenjian et al. 2000) lake trout.

Data from bioenergetics models must be coupled with predator mortality and
growth data to extend consumptive demand from an individual to a population. One such
approach is the production-conversion efﬁciency method (Ney 1990, 1993) that
incorporates estimates of predator production and gross conversion efﬁciency (GCE).
Using gross production instead of abundance at the start of the year, allows consumption
to be estimated for ﬁsh that live only a portion of the year. The GCE provides a measure
of how well an animal converts ingested food into new tissue (Brett and Groves 1979)
and typically declines as ﬁsh body size increases (Adams et a1. 1982). While it can be
determined experimentally (e.g., Kelso 1972; Edsall et al. 1999), it is often estimated

from bioenergetics models.

51

Fishery managers have identiﬁed chinook salmon, lake trout, walleye Stizostedion
vitreum, and burbot Lora Iota as the key open water predators in the main basin of Lake
Huron. While these predators are prominent in the lake (Ebener et al. 1995), information
about their consumption levels is lacking. The objectives of this paper are to (1) estimate
annual consumption of forage ﬁsh by the major predators in the open waters of the main
basin of Lake Huron; (2) compare this forage demand to recent prey availability and
historical consumption; and (3) project future consumption levels resulting from various
possible management actions. Estimates of recent consumption are useful for
understanding patterns of consumption and forage demand (e.g. Kitchell and Crowder
1986; Eby et al. 1995; Negus 1995) and provide an important basis for evaluation of
future management actions. Projecting predator consumption under different
management scenarios provides valuable insights into the effects of management
initiatives (LaBar 1993). Several alternative management scenarios are projected to

explore how various management actions affect predator forage demand.

Methods

We estimated consumption by the key predator populations using the
production-conversion efﬁciency approach (Ney 1990, 1993). Estimates of age-
speciﬁc population abundance and mortality rates from age-structured population
models, together with information on weight-at-age, were used to estimate production.
Production estimates were then divided by gross conversion efﬁciency (GCE) estimates

to compute year- and age-speciﬁc consumption. We estimated age-speciﬁc GCEs from

52

bioenergetics models (Hewett and Johnson 1995 version 3.0b) using Lake Huron

speciﬁc data on ﬁsh growth, diet, energy density, and water temperature.

StockLAssessment Models

Age-structured population models have been developed for each of the key
predator populations in the main basin of Lake Huron between 1974 and 1998. These
models included one for burbot in the main basin, one for chinook salmon in the main
basin, three for lake trout corresponding to a northern, central, and southern region of
the main basin, and two for walleye, one for Saginaw Bay and one for the main basin
south of Saginaw Bay (Figure 3.1). Modeling lake trout across three lake regions and
walleye between the main basin and Saginaw Bay was necessary because these
populations exhibit differences in survival, growth, and/or diet composition that
required separate stock assessment models. The critical information needed for each
predator population was year- and age-speciﬁc abundance and mortality rates. These
were obtained from the parameters of existing age-structured population models (Bence
and Dobiesz 2000). The parameters of these models were estimated by ﬁtting them to
available ﬁshery and survey data. Parameters included abundance-at-age in the initial
year of the time period being modeled, recruitment each year and additional parameters
determining mortality rates needed to project population dynamics over time (Bence

and Dobiesz 2000; Bence and Ebener 2002; McLeish et al. In preparation).

53

Lake trout, walleye, and burbot
Lake trout, walleye, and burbot population models operate with annual steps.
Starting with initial numbers for the ﬁrst year modeled, numbers-at-age (N), except for

the youngest age, were updated by:

Z a).

N

a+l,y+1

= Nme (2)

using species-speciﬁc year (y) and age (a) ranges. For lake trout and burbot, total

mortality (2) is broken into components for background natural mortality (M), sea

lamprey-induced mortality (L), and ﬁshing (F):

Za’y = Ma + Lao, + FM (3)

For walleye, only background natural mortality and ﬁshing components are included:

Abundance estimates from the population models along with mortality and growth rates
were used to calculate gross production over time for each species. Gross production
each year is estimated as the sum of yield, biomass of ﬁsh that die from other causes, and

change in standing stock biomass. For burbot, walleye, and lake trout, biomass (Bay) is
the product of number- and weight-at-age for each age and year. Gross production (Pay)

is calculated on an age- and year-speciﬁc basis accounting for population abundance,
mortality rates, and estimates of individual growth rate (Bence and Dobiesz 2000). It
represents predator biomass produced through the year including losses due to natural

mortality and harvest.

54

l
— B”) + BM Z” G— —Za,y

a,y

P -(B

a,y — a+l,y+l

(exp(Ga,y — Z”) — 1) (5)

where la y is the instantaneous mortality rate for a given age and year and Ga, y is the

instantaneous growth rate estimated by

Gay =1n(Wa+l,y+l /W0.y) (6)

where Wa, y is weight-at-age for year y. Ga, y was assumed to be constant over years for
burbot, northern lake trout, southern lake trout, and walleye. Age- and year-speciﬁc
values were used for central lake trout where weight-at-age was found to vary with
time. The instantaneous growth rate cannot be estimated for the last age from the

weight-at-age data; therefore, Ga, y was assumed to be zero for the last age group.

Weight-at-age estimates for burbot were obtained by ﬁtting a von Bertalanffy
curve to mean weights for ages 3 —17 (Jim Johnson, Michigan Department of Natural
Resources, Pers. Comm.) (Appendix B). For the lake trout models, weight—at-age was
estimated from data collected during spring gill net surveys conducted by the Michigan
DNR (Appendix B). Walleye weight-at-age was estimated from 1985-1995 Lake Huron

creel data (Appendix B).

Chinook Salmon
The population model for chinook salmon uses two time periods within a year
consisting of the ﬁrst seven months (prior to a “pulse” of harvest and maturation) and

then the remainder of the year (Bence and Dobiesz 2000). The annual update equation is:

55

N = NMe—M (1 - Pp,a,,) (1 — Pm,a,y) <7)

a+Ly+l
where PF, a, y and Pm, a, y are the proportions of ﬁsh that die due to fishing or maturation

respectively. The numbers at the end of the ﬁrst time period (prior to the pulse) and at

the beginning of the second time period (immediately following the pulse) are given by:

* __7_M
e 12
a,y,i a,yui (8)

5

,, ———M
N : Na,y,i e 12 (1 — PFa,y) (1 — Pm,a,y) (9)

a, y,i+1
Here Na,”- indicates the numbers for period i, and the “*” indicates if the numbers are for

the end rather than the beginning of the period. In calculations of harvest numbers and

return of mature ﬁsh it is assumed that ﬁshing mortality occurs prior to maturation.

Gross production by chinook salmon was calculated for two intervals — pre-
harvest and post-maturation. Annual production is the sum of production over these two

intervals for a given year. Biomass of age-a ﬁsh at the start of interval i in year y is

Ba y,i : Na,y,i ><Wa,y,i

9

(10)

N61,)“ and Wa, y, ,- are the number- and weight-at-age at the start of the interval.

Production lost to natural mortality during each period is

= B . M [1/Ga’y,,.—M)][exp(G

,y.i a,y.t a,y,i

Dd _M)ti-1] (11)

where M is the natural mortality rate, Gay, i is the instantaneous growth rate applying

to interval i, and t,- is the proportion of a year represented by the interval. For chinook

56

salmon, t is 7/12 and 5/12 for the pro-harvest and post-maturation intervals

respectively. The instantaneous growth rate is estimated by

_ 1n (“2, y,i+1 I W1, y,i

a,y,i _ t (12)
r

 

When the ith interval is the last interval in the year, Wa,y,,-+1=Wa+1,y+1, 1 .

Instantaneous growth rate for the last age of the post-maturation interval (age 5+) was
assumed to be zero. Chinook salmon weight-at-age was determined by ﬁtting a von
Bertalanffy model to weight-at-age while allowing the parameter representing the

asymptotic maximum length (Lao) to vary over time (McLeish et al., In preparation).

The growth and population models were ﬁt simultaneously to produce age- and year-
speciﬁc weight-at-age information (Appendix B).

Total production over the entire year is given by:

a,y =ZAa,y,i +ZDa,y,i (13)
z 1

where A0,»,- represents the change in age-speciﬁc standing stock biomass during period

i. This can be reexpressed as:

(14)

a,y = Aa,y -Ra,y +ZDa,y,i

where Ra, y are the pulse removals of fish (in biomass) due to ﬁshing and maturation

between the two periods of natural mortality.

57

Bioenergetics Models

The Wisconsin bioenergetics model (Hewett and Johnson 1995, updated to
V3.0b) was used to estimate consumption for an average individual ﬁsh by employing the
option to ﬁt consumption as a function of change in weight. Separate models were used
to represent each predator population in parallel with the stock assessment models. Lake
Huron—speciﬁc values were used for diet composition, energy density of predators and
prey, and water temperature (Appendix B). Age-speciﬁc gross conversion efﬁciency
(GCE) for each predator population was obtained by dividing annual growth estimates
from weight-at-age information (Appendix B) by estimates of consumption from the

bioenergetics models.

Es_tjr_n_§ting Recent Consumption

The production-conversion efﬁciency method (Ney 1990, 1993) was used to
compute estimates of prey consumption by each predator population. Age- and year-
speciﬁc gross production was divided by age-speciﬁc GCE and summed over all ages to
obtain total consumption. Consumption on a prey species basis was computed by
multiplying total consumption by estimates of the proportion (by weight) that each prey
species contributes to the diet. This approach, along with several key assumptions
regarding future changes (Appendix C), was also used to project consumption under

several typical management scenarios.

58

Sensitivity Analysis for Estim_ates of Recent Consumption

To approximate Lake Huron conditions, our bioenergetics models used Lake
Huron-speciﬁc values for water temperature, diet composition, and energy density of the
predators and prey (Appendix B). Since energy density is a time-consuming and difﬁcult
quantity to measure, values are often borrowed from other lakes or other time periods
within the same lake. We tested the sensitivity of the annual consumption estimates to
two types of variation in energy density. First, we explored the effects of using
regionally-varying energy density versus a basin-wide mean. Our previous research
found a unique pattern of regionally-varying energy density for lake trout, with lake trout
from the northern region having signiﬁcantly lower energy density than lake trout in the
central or southern regions (Chapter 2). We tested the importance of using regional lake
trout energy density against the basin-wide mean by running each lake trout bioenergetics
model with the basin-wide mean energy density and all other data held constant. From
these runs we obtained the GCEs and then estimated population level consumption.

Second, we examined the impact of using published energy density of predator
and prey species from other Great Lakes instead of data obtained from Lake Huron.
Energy density has been found to vary between ecosystems related to variations in the
ﬁsh community, food density, and climatic conditions (Rand et al. 1994; Foy and Paul
1999; Pazzia et a1. 2002). We noted that energy density of predators and prey in Lake
Huron was lower than that found in other Great Lakes (Chapter 2). We tested the
importance of using Lake Huron-speciﬁc values of energy by rerunning each
bioenergetics model using published mean energy density for both predator and prey

species from other Great Lakes (Table 3.1).

59

Estimating Historic Lake Trout Consumption

As a benchmark for comparison of current consumption levels, we estimated the
consumption by lake trout during 1912 to 1940. Our basic approach was to estimate a
level of recruitment and associated population abundance that could have produced an
equilibrium yield equal to the average observed yield during the period (Bence and
Dobiesz 2000). In these calculations we assumed that the ﬁshery was stable. We
proposed an overall mortality rate that would not cause a collapse, and subtracted an
estimate of natural mortality to obtain the corresponding ﬁshing mortality rate. We then
used numerical algorithms built into Microsoft® Excel Solver to adjust the recruitment
level until the equilibrium yield equaled the observed average. This produced estimates
of equilibrium abundance, biomass, and related quantities. With estimates of historical
diet composition and GCEs (Appendix B), the production-conversion efﬁciency method
(Ney 1990, 1993) was used to estimate annual consumption by lake trout. While burbot
likely played an important predatory role historically, we lacked information to estimate

their abundance and consumption levels.

Projecting Future Consumption

Projecting consumption under various management scenarios can improve our
understanding of how these actions may affect prey consumption by the key predators.
We project annual consumption for three possible scenarios. The “baseline” scenario
includes changes to lake trout stocking and harvest regulations resulting from the 2000
consent decree for waters ceded by the 1836 Treaty of Washington (United States v.

Michigan 2000), and the 20% reduction in chinook salmon stocking that began in 1999.

60

All other factors affecting abundance and consumption are held constant, at 1998 levels.
The “sea lamprey reduction” scenario includes all “baseline” factors but reduces sea
lamprey-induced mortality on lake trout and burbot assumed to result from enhanced
efforts to control sea lamprey in the St Marys River. With chinook salmon dominating
prey consumption, especially in recent years, it is informative to explore a scenario
resulting in decreased chinook salmon abundance. The last scenario, “chinook stocking
reduction”, adds a 50% reduction in chinook salmon stocking beginning in the year 2002

to the “sea lamprey reduction” scenario.

To estimate future consumption, several assumptions were made regarding
mortality rates, weight-at-age, diet composition, and GCE, during the projection period,
1999 through 2020. Natural mortality rates (excluding sea lamprey-induced mortality)
for the projection period were constant and set to the value used in the last year of the
assessment models. Several different assumptions were used for estimating ﬁshing
mortality during the projection period. For southern walleye and burbot a single source
of ﬁshing mortality was set to the value of the last year of data; for Saginaw Bay walleye
the average of the last three years was used. All three lake trout models and the chinook
salmon model contained commercial and recreational ﬁshing mortality calculated as the
product of selectivity and ﬁshing intensity. The chinook salmon and southern lake trout
models used constant selectivity from the last year of data and time varying ﬁshing
intensity as the average of the last three years. For northern and central lake trout,
selectivity and ﬁshing intensity were allowed to vary over time in the stock assessment
models, while the average of the last three years was used for projections. For chinook

salmon, the maturation proportion was set to the estimates for the last year in the

61

assessment model. Sea lamprey-induced mortality was applied to the burbot model and
all three lake trout models. For the projection period, this mortality source was adjusted
by a scalin g factor (Schleen et al. 2002) intended to reﬂect the reduction of sea lamprey
abundance resulting from treatment of the St Marys River. The scaling factor did not

apply to the “baseline” scenario.

For the projection period, weight-at-age was assumed ﬁxed for the assessment
models that did not use time-varying weight-at-age (northern and southern lake trout,
burbot, and Saginaw Bay and southern region walleye), while the mean of the last three
years was used for those models using time-varying weight-at-age (chinook salmon and
central lake trout). Diet composition and GCE were assumed constant for estimates of

recent and projected consumption.

Results
Gross conversion efﬁciency and gross production

Patterns in production for the main basin (Figure 3.2) are roughly similar to those
of consumption (Figure 3.3). Gross conversion efﬁciency (GCE) decreased with age and
varied among species and predator populations (Table 3.2). Burbot made up a larger
portion of consumption than of production because of relatively low GCEs (Table 3.2).
Consumption by chinook salmon increased faster than production because of a shift

toward lower GCEs (resulting from slower growth) over time.

62

Preﬂttor Biommnd Mean Consumption for 1984-1998

For the period in which all predators were modeled (1984-1998), estimated annual
mean total consumption in the main basin was 32 million kg. Consumption has varied
about this mean with a minimum of 26 million kg in 1984 and a maximum of 41 million
kg in 1998 (Figure 3.3). In the main basin, the major consumers are lake trout and
chinook salmon. On average their forage demand represents 71% of the total demand by
the major predators. From 1984 to 1998, chinook salmon increasingly dominated the
total consumption of prey in the main basin (Figure 3.3). Consumption data shown here

include consumption of “other prey” by Saginaw Bay walleye that are not typically part

 

of the main basin forage mix.

Trends in consumption, production, and biomass for chinook salmon from 1968
through the mid-19803 show their biomass and consumption of prey increasing (Figure
3.4). Consumption by chinook salmon increased by an average of 13% per year between
1968 and 1987, reaching 15.4 million kg from a low in 1968 of 514,000 kg (Figure 3.4).
Chinook salmon biomass also increased from nearly zero (933 kg) in 1968 to 4.3 million
kg in 1998 (Figure 3.4). Consumption, production, and biomass leveled off and then

declined in the late 19803 and early 19903, then began increasing during the mid-19903.

Lake trout biomass in the main basin declined from 4.1 million kg to 2.1 million
kg from 1984 to 1998 (Figure 3.5). Most of this decline occurred in the southern region
of the main basin (Figure 3.1). In 1984 biomass estimates for the northern region and the
central region were 505,000 kg and 604,000 kg respectively. Modest gains occurred in
the central region through 1998 with biomass estimated at 734,000 kg while the northern

region experienced modest losses with biomass estimated at 469,000 kg.

63

Annual consumption by lake trout (Figure 3.3) followed biomass (Figure 3.5)
declines and was exacerbated because the decline occurred in the area where lake trout
growth is highest. However, consumption was offset because the remaining biomass was
concentrated in younger, faster growing ﬁsh. Total consumption by lake trout declined
from a high of 11.0 million kg in 1984 to 6.5 million kg in 1998 (Figure 3.3).
Consumption increased from 1991-1994 by an average of 7% per year caused by
relatively strong year classes in 1992-1993 in the southern basin (Figure 3.6).

Consumption declined after 1994 by an average of 9% per year.

Mean prey consumption by burbot (1984-1998) was estimated at 4.8 million kg
annually (Figure 3.3). The temporal pattern was a product of our assumption that
recruitment was constant. However, with limited information on burbot, assumptions
needed to determine their population dynamics are open to question (Bence and Dobiesz
2000); therefore, we can only evaluate burbot consumption for general trends as data is
lacking to determine speciﬁc consumption rates. Their overall forage demand seemed to

lie between that of lake trout and walleye (Figure 3.3).

During 1984-1998, Saginaw Bay walleye biomass increased from 1.2 million kg
to 2.0 million kg while the biomass of southern walleye declined from 1.2 million kg to
514,000 kg. Mean walleye biomass (1984-1998) for the main basin was 2.6 million kg.
Total consumption by both the Saginaw Bay and southern region walleye was at its
maximum in 1986 with 5.3 million kg consumed (Figure 3.6). Consumption declined
since that time by approximately 4% per year, with total consumption of 3.6 million kg in

1998.

Sensitivity Analysis Using Basin-Wide Energy Density

Using basin-wide energy density, total annual lake trout consumption increased
1% in the northern region and decreased 2.5% in the central, and 1.5% in the southern
lake regions. This was accompanied by a mean reduction in the age-speciﬁc GCEs of
1.5% in the northern region and a increase in GCEs of 7% and < 1% in the central and
southern regions. The effect of using basin-wide energy density across the lake trout
regions was to increase the energy density of northern and central lake trout by 10% on

average but decrease the energy density of southern lake trout by 2%.

Sensitivity Analysis Usingggblished Energy Densitgr

For burbot and walleye, literature mean energy densities produced lower age-
speciﬁc annual consumption estimates by 5% and 17% respectively. The estimated
energy density for Lake Huron burbot was higher than the value used in Rudstam et al.
(1995), which was estimated from Atlantic cod (Table 3.1). Since we used the same
bioenergetics parameters as used in Rudstam et al. (1995), the lower estimated
consumption is primarily related to the energy density differences. The reduction in
walleye consumption using literature mean energy denisty was larger and likely
attributed to a higher mean energy density for Lake Huron walleye compared to that
used by Kelso (1972) (Table 3.1). However, the Lake Huron estimate of walleye
energy density was based on a limited number of samples over a brief two month
period and may not be representative of their energy density over the entire year.
Given the results using literature energy density, this may have caused our walleye

consumption estimates to be higher than expected.

65

Both chinook salmon and lake trout consumption estimates were higher using
the literature values for energy density of predators and prey as compared to Lake
Huron-speciﬁc values. Our estimated energy density for salmonines in Lake Huron
(Chapter 2) was substantially lower than most of the reported literature values (Table
3.1). The higher energy density value from the literature accounted for a 4% and 9%
increase in consumption by lake trout and chinook salmon respectively. However,
these differences are relatively small compared to other uncertainties in the

bioenergetics and age-structured models.

Comparison To Historic Consumption by Lake Trout

The historical (1912-1940) average yield was 1.8 million kg in the main basin
(Baldwin et al. 1979). The model matched this yield when recruitment slightly exceeded
5 million yearling recruits per year. Historical lake trout consumption during this period
was estimated at 38.2 million kg (Bence and Dobiesz 2000). While total consumption by

the major predators has approached this level, it was exceeded only in 1998 (Figure 3.7).

Consumption of Forage Species

Recent mean consumption (1996-1998) of alewife and rainbow smelt by all of the
major predators was 29.0 million kg (Table 3.3). Over 90% of consumption by lake trout
and chinook salmon was composed of alewife and rainbow smelt (Figure 3.8). In
contrast, the burbot diet included substantial amounts of invertebrates and sculpin
resulting in a lower reliance on alewife and rainbow smelt, which accounted for only

40% of their diet (Table 3.3). The walleye diet consisted of a high proportion of “other”

66

food items representing feeding inside Saginaw Bay on a mix of prey species not
normally found in the main basin of Lake Huron (Table 3.3). However, 63% of prey
consumption by walleye included alewife and rainbow smelt, with nearly 80% of this
amount attributed to alewife.

We compared estimated consumption by the key predators (1984-1998) with the
combined alewife and rainbow smelt swept area biomass obtained from the US
Geological Survey fall bottom surveys (Figure 3.9). This survey method expands the .
numbers of ﬁsh in each trawl from the actual area trawled to the numbers of ﬁsh in all US
waters. This expansion is based on the area swept by the trawls in different regions of the
lake at a series of depth strata, accounting for the total bottom area in each region and
depth station. Consumption of alewife and rainbow smelt by the key predators was
substantially lower than the swept area prey biomass estimates until the early 19903.
From this period on, prey consumption by the key predators more closely approached the

swept area estimates.

Projected Consumption Under Several Management Scenarios

Projections approximated a steady-state for the period 2010-2020 (Figure 3.10).
Each projected value below represents the mean annual consumption by all of the key
predators during 2010-2020. In the “baseline” scenario, the mean consumption for this
period was 43 million kg. Mean consumption for the “sea lamprey reduction” scenario
was 52 million kg, representing an increase in consumption of 21% over the “baseline”
scenario consumption levels. This increase is exclusively related to reducing sea

lamprey-induced mortality on lake trout and burbot. Mean consumption for the “chinook

67

stocking reduction” scenario was 38 million kg. In this scenario, mean consumption
decreases from the “sea lamprey reduction” scenario by approximately 26% and

decreases from the “baseline” scenario by 10%.

Discussion

Intensive stocking of Paciﬁc salmon began in Lake Michigan in the 19603,
because these fast growing species were expected to consume substantial numbers of
exotic prey ﬁsh and to be attractive to anglers (Tody and Tanner 1966). These
expectations have been borne out across the Great Lakes as the number of stocked
chinook salmon has increased substantially from the 19603 through the 19803 (e. g. Lange
et al. 1995). In recent years chinook salmon have become the dominant predator in
Lakes Michigan (Stewart et al 1981; Stewart and Ibarra 1991), Ontario (Rand and
Stewart 1998), and Minnesota waters of Lake Superior (Negus 1995) as well as Lake
Huron. This domination is partly due to the substantial numbers that have been stocked
(e.g., Bence and Smith 1996), but it also results from their rapid growth rates, especially
relative to native predators (e. g., Stewart and Ibarra 1991). We found that chinook
salmon accounted for 54% of the prey biomass consumed in the main basin in Lake
Huron (1996-1998), although their biomass was roughly equal to that of each of the other
species (Figure 3.3). Chinook salmon have also been found to consume a
disproportionate amount relative to their biomass in other lake ecosystems (Hill 1997).

Lake trout are the dominant native predator accounting for 20% of consumption
of prey biomass in Lake Huron (1996-1998); however, total consumption by lake trout

declined from a high in 1984 to its lowest level in 1998. Lake trout consumption

68

followed biomass declines and was exacerbated because the decline occurred in the area
where lake trout growth is highest (southern region), but offset because the remaining
biomass was concentrated in younger, faster growing ﬁsh..

Walleye are also important predators, especially in Saginaw Bay and southern
Lake Huron, with estimated total walleye abundance exceeding that of lake trout in 1998.
However, known movement of walleye between Saginaw Bay and the southern main
basin was not included in the population models so walleye abundance may be
overstated. Lack of spawning habitat (Colby et al. 1994), lower water temperatures
(Christie and Regier 1988), and predation on larval walleye by alewife (Fielder 2002) are
possible reasons for the lack of walleye in the main basin. Similarly, burbot may be a
more important predator in the main basin but we lacked the data to estimate burbot

abundance with high conﬁdence.

Consumption and GCE Estimates Across Lakes

Consumption by and GCE of lake trout have been estimated for Lakes Michigan
(Stewart et al. 1983; Stewart and Ibarra 1991), Ontario (Rand and Stewart 1998), and
Superior (Negus 1995). Estimated GCEs for Lake Ontario lake trout ages 1-3 averaged
12.7% (Rand and Stewart 1998), substantially lower than Lake Huron estimates
(approximately 18%). However, GCEs from both lakes were nearly identical for older
ages. Rand and Stewart (1998) used slower growth rates in young lake trout and higher
energy density for all lake trout, which likely explains these differences. In contrast,
GCEs estimated for Lake Michigan lake trout were 24.4 — 7.4% for ages 1-9 (Stewart et

al. 1983), nearly identical to our GCE estimates (Table 1). Similarly, the GCE of 9.2%

69

for a 3 kg Lake Superior lake trout (Negus 1995) was within the range of GCEs for 3
kg lake trout in all Lake Huron regions (7.3 - 9.3%). Information on lake trout GCEs
appears to be restricted to the Great Lakes, but the estimated GCE for hatchery brown
trout averaging 300 g was 16% (Elliott and Hurley 2001), comparable to Lake Huron
lake trout of the same size with mean (across lake regions) GCE of 18.3%.

Chinook salmon GCE and consumption have also been studied across the Great
Lakes (Stewart and Ibarra 1991; Negus 1995; Rand and Stewart 1998). For chinook
salmon ages 1-4, the Lake Huron mean GCEs across the two growth periods (Table l)
were lower by 16 % (ages 1-2) and 40 % (ages 3-4) than Lake Ontario GCEs that
ranged from 29 — 12% (Rand and Stewart 1998). However, the Lake Huron GCE for
age 0 chinook salmon was 15 % higher than the Lake Ontario estimate (Rand and
Stewart 1998). In Lake Superior, a 3 kg chinook salmon (Negus 1995) also had a
higher GCE (23.4%) than a similar sized ﬁsh in Lake Huron (GCE = 17.8%). With
similar growth rates between Lake Ontario and Lake Huron (except for age 2), lower
GCEs imply that higher consumption is needed to obtain the same growth.
Additionally, chinook salmon energy densities used in the Lake Michigan and Lake
Ontario models (Stewart and Ibarra 1991; Rand and Stewart 1998) were higher than our
Lake Huron value and inﬂuenced our GCE estimates. Since our energy density is
based on a one-year sample (Chapter 2), the estimated GCEs may reﬂect a time of
higher feeding due to poor prey ﬁsh condition during the sampling period. Using these
GCE estimates, we may be overstating chinook salmon forage demand over the entire
time frame we modeled. Generally, GCE estimates from other Great Lakes were at

least 15% higher while GCE estimates from Lake Oahe, South Dakota were

70

approximately 14% lower (Hill 1997) than Lake Huron. A sensitivity analysis using
GCEs that were i 15% of the Lake Huron values produced mean annual consumption
estimates for the period 1996 — 1998 that were 13% lower and 15% higher. This
analysis indicates that our chinook salmon consumption estimates may be off by i- 20%
but not by orders of magnitude.

Both Lake Huron walleye populations had GCEs that were comparable to the
annual mean GCEs in the Bay of Quinte, Lake Ontario (Hurley 1986) and experimental
feeding studies on walleye in Lake of the Woods, Minnesota (Swenson and Smith Jr.
1973). Hartman and Margraf (1992) used bioenergetics models to estimate annual
consumption of Lake Erie walleye during 1986-1988. Their estimates of individual
consumption ranged from 1,642 g for age 2 ﬁsh to 3,486 g for age 6. Estimates from
Lake Huron walleye populations were similar to Lake Erie for ages 2-3 but higher for
ages 4-6. Individual walleye from the Saginaw Bay population consumed an average
of 1,643 g for age 2 ﬁsh to 4,329 g for age 6 ﬁsh while individual walleyes in the
southern population consumed slightly less at 1,499 g to 3,837 g (ages 2-6). In Lake
Erie, walleye diets were dominated by clupeids (Hartman and Margraf 1992) whereas
diets of the Lake Huron open water walleye populations were dominated by alewife.
Since the growth rate for age 2-6 walleye was similar between Lake Erie and Lake
Huron, the differences in water temperature, diet, and energy density used in the

bioenergetics models account for the differences in consumption.

71

Effects of energy densgy on consumption estimates

Using region-speciﬁc energy density had only a minimal effect on lake trout
consumption estimates. Considering the numerous areas of uncertainty associated with
bioenergetics models, the mean 1.4% change over all lake regions represents a modest
increase in consumption and indicates that a fair approximation of consumption by lake
trout could be obtained using basin-wide mean energy density. We caution, however,
that this conclusion might reﬂect the speciﬁcs of the situation. For example, during
most of our time series, ﬁsh from the southern region dominated lake trout abundance.
A major change such that lake trout became much more abundant in the northern
region, could invalidate the use of the basin-wide mean as an appropriate
approximation.

Our evaluation of the effect of using lake-wide means for energy density or
values borrowed from the literature suggests that doing so would not be unreasonable.
For instance, although the mean published lake trout energy density was 20% higher
than the Lake Huron mean, estimated consumption by lake trout using the literature
mean was only 4% higher than our estimates. We suspect that in many cases using
published energy density data would cause moderate percentage changes, and would be
very unlikely to alter estimates of consumption by a factor of two. However, some
caution is needed when using published energy density data because this measure is
affected by many factors including changes in the ﬁsh community, food density, and
climatic conditions (Rand et a1. 1994). When possible, local evaluations of energy

density are preferable. Further, since bioenergetics models provide estimates of

72

individual consumption that are expanded to an entire population, better estimates of

predator abundance are also critical when estimating predator forage demand.

Comparison to Benchmarks

The observed patterns showing recent levels of consumption approaching
standing stock estimates of prey (Figure 3.8) and estimates of consumption by historical
lake trout populations (Figure 3.6) are of concern. Evidence of a corresponding decline
in lake trout and chinook salmon growth (McLeish et al., in preparation) suggest a
possible imbalance between predator and prey abundances. Therefore, caution should be
exercised with regard to either increasing stocking further or maintaining the current level
of stocking. If piscivores are exceeding forage ﬁsh capacity, stocking plans would need
to be changed to assure continued growth and survival of all piscivores.

In 1998, consumption apparently exceeded the prey biomass (Figure 3.8) because
consumption is estimated on an annual basis while prey biomass is measured at a point in
time. Annual production to standing stock biomass ratios can exceed unity, especially for
small ﬁsh, which means prey availability exceeds standing stock. Second, swept area
biomass estimates tend to underestimate the biomass. Estimating the proportion of ﬁsh
present in the area swept by the trawl gear that are actually retained by the gear is
difﬁcult and can effect biomass estimates by a factor of two or more . Finally, it is
important to recognize that the stock assessments and other calculations on which our
consumption estimates were based have substantial uncertainty, although we have no

reason in particular to suspect they overestimate the biomass consumed.

73

Rand and Stewart (1998) estimated prey ﬁsh exploitation rates by dividing
predator consumption by prey ﬁsh production. While they found only modest
exploitation rates (approximately 13%) during 1990-1991, they noted that most of the
forage demand was directed at adult alewife where the exploitation rate exceeded 100%.
In Lake Huron, exploitation rates have varied over time (Table 3.4) and were generally
much higher than the 13% observed in Lake Ontario, but much closer to the estimated
50% (Brandt et al. 1991) observed in Lake Michigan in 1987.

Although forage ﬁsh abundance declined (Figure 3.8) while predator abundance
(Figure 3.3) increased over time, we cannot say with certainty that the predators were the
major cause of the decline. Forage demand per gram of predator in Lake Huron has
varied only slightly over time and only apparently exceeded the grams of prey available
per predator in 1998 (Table 3.4). Forage ﬁsh population abundance estimates were
highly variable over time and do not reﬂect a simple connection to predator numbers.
However, given the change in the predator-prey ratio, especially in recent years (Figure
3.9), predators will likely exert a larger role in future years.

To put current consumption levels in perspective, we compared contemporary
predator abundance and prey consumption against estimates of these quantities for lake
trout (the dominant predator historically) in the main basin of Lake Huron during the
period of a stable ﬁshery, 1912 to 1940. Current consumption levels approached our
estimates of historic prey consumption (Figure 3.6). Our estimates of historic lake trout
abundance and consumption required many assumptions (Bence and Dobiesz 2000) and
the analysis did not include burbot, a second potentially important native predator.

Nevertheless, our analysis suggests that current and historical prey consumption was

74

reasonably similar, despite the dependence of the current predator community on

stocking.

Projections of Consumption

Projecting consumption under various management scenarios can improve our
understanding of how these actions may affect prey consumption by the key predators in
Lake Huron. We explored three alternative management scenarios: a baseline, a
reduction in sea lamprey-induced mortality, and a 50% reduction in chinook salmon
stocking (Figure 3.9). The Baseline is informative as a reference point of consumption
estimates if the status quo management plan is followed. Reduction in sea lamprey-
induced mortality on lake trout and burbot related to the treatment of the St Marys River
improves survival of these predators with a corresponding increase in consumption.
Similarly, reducing chinook salmon stocking lowers their abundance and future
consumption. While these management actions have an effect on predator forage
demand, neither result represents more than a 40% change in consumption levels.

Fishery managers need estimates of predator forage demand in order to deﬁne a
quantitative measure of how much consumption can be supported without high risk of
forage ﬁsh stock collapses. Our estimates of consumption by the key predators were
based on static diet information and energy density collected during one year. These and
other data are known to vary by season and lake region. Estimates of consumption could
be enhanced by the continued collection of data such as diet, predator weight-at-age, and
energy density, collected seasonally and spatially where unique predator populations

exist. Improved understanding of predator-prey interactions in Lake Huron should also

75

include stock assessment models to estimate key predator consumption in Georgian Bay
and the North Channel, and investigation of functional response models to improve our
understanding of the connection between forage ﬁsh dynamics and piscivore
consumption.

These projections could be improved by addressing areas of uncertainty in our
estimates of predator abundance, biomass, production, and consumption. We have the
highest conﬁdence in the estimates for lake trout, which were based on age-structured
stock assessment models ﬁt to ﬁshery and survey data. In contrast, we have the lowest
conﬁdence in the estimates for burbot, which were based on estimates of the relative
abundance of burbot using lake trout abundance (Bence and Dobiesz 2000). Some
uncertainty in these estimates could be minimized by the following actions: (1) a ﬁshery
independent assessment of trends in chinook salmon abundance; (2) improved
understanding of wild recruitment of chinook salmon; (3) assessments of burbot
abundance; and (4) a study of the mixed population of walleye in Saginaw Bay and the
main basin of Lake Huron including the sources of ﬁsh in different regions of the lake

and the seasonal patterns of movement.

76

Literature Cited

Adams S. M., R. B. Mclean, and J. A. Parrotta. 1982. Energy partitioning in largemouth
bass under conditions of seasonally ﬂuctuating prey availability. Transactions of
the American Fisheries Society 111: (5) 549-558.

Adams, S. M. and J. E. Breck. 1990. Bioenergetics. In Schreck, CB. and PB. Moyle
(eds.), Methods for ﬁsh biology. American Fisheries Society, Bethesda,
Maryland, p. 389-415.

Baldwin, N. S., R. W. Saalfeld, M. A. Ross, and H. J. Buettner. 1979. Commercial ﬁsh
production in the Great Lakes, 1867-1977. Great Lakes Fishery Commission
Technical Report 3: 187p.

Bence J .R. and NE. Dobiesz. 2000. Estimating forage ﬁsh consumption by predators in
Lake Huron. Great Lakes Fishery Commission Project Completion Report.
Available for download at http://www.glfc.org/research/cap.htm.

Bence, J. R. and M. P. Ebener (eds). 2002. Summary status of lake trout and lake
whiteﬁsh populations in 1836 treaty-ceded waters of Lakes Superior, Huron and
Michigan in 2000, with recommended yield and effort levels for 2001. Technical
Fisheries Committee, 1836 Treaty-Ceded Waters of Lakes Superior, Huron and
Michigan.

Bence, J. R. and K. D. Smith. 1999. An overview of the recreational ﬁsheries of the
Great Lakes. In Taylor, W. W. and P. Ferreri (eds.), Great Lakes Fisheries Policy
and Management: A Binational Perspective, Michigan State University Press.

Boisclair D. and W. C. Leggett. 1989. Among-population variability of ﬁsh growth: 1.
Inﬂuence of the quantity of food consumed. Canadian Journal of Fisheries and
Aquatic Sciences 46 (3): 457-467.

Boisclair D. and P. Sirois. 1993. Testing assumptions of ﬁsh bioenergetics models by
direct estimation of growth, consumption and activity rates. Transactions of the
American Fisheries Society 122: 784-796.

Borgmann, U. and D. M. Whittle. 1992. Bioenergetics and PCB, DDE, and mercury
dynamics in Lake Ontario lake trout (Salvelinus namaycush): a model based on
surveillance data. Canadian Journal of Fisheries and Aquatic Sciences 49: 1086-

1096.

Brandt S.B., D.M. Mason, E.V. Patrick, R.L. Argyle, L. Wells, P.A. Unger, and DJ.
Stewart. 1991. Acoustic measures of the abundance and size of pelagic

77

planktivores in Lake Michigan. Canadian Journal of Fisheries and Aquatic
Sciences 48: 894-908.

Brett J. R. and TDD. Groves. 1979. Physiological energetics, p 280-345. In W. S.
Hoar, D. J. Randall, and J. R. Brett [eds.] Fish Physiology. Academic Press: New
York , NY.

Christie, G.C., and Regier, H.A. 1988. Measures of optimal thermal habitat and their
relationship to yields for four commercial ﬁsh species. Can. J. Fish. Aquat. Sci.
45: 301-14.

Colby, P.J., Lewis, C.A., Eshenroder, R.L., Haas, RC, and Hushak, L]. 1994. Walleye-
rehabilitation guidelines for the Great Lakes area. Great Lakes Fish. Comm., Ann
Arbor, Mich. 112 p.

Ebener, M. P., J. E. Johnson, D. M. Reid, N. P.Payne, R. L. Argyle, G. M. Wright, K.
Kruger, J. P. Baker, T. Morse and J. Weise. 1995. Status and future of Lake
Huron ﬁsh communities. In Munawar, M., T. Edsall and J. Leach (eds.), The
Lake Huron ecosystem: Ecology, Fisheries and Management. Ecovision World
Monograph Series, SPB Academic Publishing, Amsterdam, The Netherlands,
pp. 125-170.

Eby, L. A., L. G. Rudstam and J. F. Kitchell. 1995. Predator responses to prey
population dynamics: An empirical analysis based on lake trout growth rates.
Canadian Journal of Fisheries and Aquatic Sciences 52(7): 1564-1571.

Edsall, T.A., A.M. Frank, D.V. Rottiers, and J .V. Adams. 1999. The effect of
temperature and ration size on the growth, body composition, and energy content
of juvenile coho salmon. Journal of Great Lakes Research 25(2): 355-362.

Elliott J .M. and MA. Hurley. 2000. Daily energy intake and growth of
piscivorous brown trout, Salmo trutta. Freshwater Biology 44: 237-
245.

Elliiott J. M. and M. A. Hurley. 2001. Modeling growth of brown trout, Salmo trutta, in
terms of weight and energy units. Freshwater Biology 46: 679-692.

Eshenroder, R. L. and M. K. Bumham-Curtis. 1999. Species succession and
sustainability of the Great Lakes ﬁsh community. In Taylor, W. W. and P. Ferreri
(eds.), Great Lakes Fisheries Policy and Management: A Binational Perspective,
Michigan State University Press.

Eshenroder, R. L., N. R. Payne, J. E. Johnson, C. Bowen H, and M. P. Ebener. 1995.

Lake trout rehabilitation in Lake Huron. Journal of Great Lakes Research 21
(Suppl. 1): 108-127.

78

Fielder D.G. 2002. Sources of walleye recruitment in Saginaw Bay, Lake Huron. North
American Journal of Fisheries Managment 22: 1032-1040.

Foy, R]. and A.J. Paul. 1999. Winter feeding and changes in somatic energy content of
age-0 Paciﬁc herring in Prince William Sound, Alaska. Transactions of the
American Fisheries Society 128: 1193-1200.

Hansen, M. J ., D. Boisclair, S. B. Brandt, S. W. Hewett, J. F. Kitchell, M.
C. Lucas and J. J. Ney. 1993. Applications of bioenergetics models
to ﬁsh ecology and management: Where do we go from here?
Transactions of the American Fisheries Society 122: 1019-1030.

Hartman K. J. and F. J. Margraf. 1992. Effects of prey and predator
abundances on prey consumption and growth of walleyes in

Western Lake Erie. Transactions of the American Fisheries Society
121: 245-260.

Hewett. S. W. and B. L. Johnson. 1995. Fish Bioenergetics Model 3. University of
Wisconsin Sea Grant Institute, WIS-SG-91-250.

Hill T. D. 1997. Life history and bioenergetics of chinook salmon in Lake Oahe, South
Dakota. Ph.D. dissertation, South Dakota State University, SD.

Holey, M. E., R. F. Elliott, 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: 202-210.

Hurley, D. A. 1986. Growth, diet, and food consumption of walleye: an application of
bioenergetics modeling to the Bay of Quinte, Lake Ontario, population. In C. K.
Minns, D. A. Hurley, and K. H. Nicholls [ed3.] Project Quinte: point-source
phosphorus control and ecosystem response in the Bay of Quinte, Lake Ontario.
Canadian Special Publication of Fisheries and Aquatic Sciences 86: 0706-6481.

Jones, M. L., J. F. Koonce and R. O’Gorman. 1993. Sustainability of hatchery-
dependent salmonine ﬁsheries in Lake Ontario: The conﬂict between predator

demand and prey supply. Transactions of the American Fisheries Society 122:
1002-1018.

Kelso R. 1972. Conversion, maintenance, and assimilation for walleye as affected by
size, diet and temperature. Journal of the Fisheries Research Board of Canada 29
: 1181-1192.

Kitchell, J. F., D. J. Stewart, and D. Weininger. 1977. Applications of a bioenergetics

model to yellow perch (Perca ﬂavescens) and walleye (Stizostedion vitreum
vitreum). Journal of the Fisheries Research Board of Canada 34: 1922-1935.

79

Kitchell, J. F. and L. B. Crowder. 1986. Predator-prey interactions in Lake Michigan:
model predictions and recent dynamics. Environmental Biology Of Fishes 16:
205-211.

Kitchell, J. F., L. A. Eby, X. He, D. E. Schindler and R. A. Wright. 1994. Predator-prey
dynamics in an ecosystem context. Journal of Fish Biology 45(Suppl. A): 209-
226.

Labar G. W. 1993. Use of bioenergetics models to predict the effect of increased lake
trout predation on rainbow smelt following sea lamprey control. Transactions of
the American Fisheries Society 122 (5): 942-950.

Lange, R. E., G. C. LeTendre, T. H. Eckert and C. P. Schneider. 1995. Enhancement of
sportﬁshing in New York waters of Lake Ontario with Hatchery-reared
salmonines. American Fisheries Society Symposium 15: 7-11.

Lantry B. F., and D. J. Stewart. 1993. Ecological energetics of rainbow smelt in the
Laurentian Great-Lakes - an interlake comparison. Transactions of the American
Fisheries Society 122: 951-976.

Madenjian C. P., D. V. O'Connor, D. A. Nortrup. 2000. A new approach toward
evaluation of ﬁsh bioenergetics models. Canadian Journal of Fisheries and
Aquatic Sciences 57 (5): 1025-1032.

McLeish, D. A and 12 other authors. In preparation. SCOL H Lake Huron Case
History. Great Lakes Fishery Commission Report.

Negus, M. T. 1995. Bioenergetics modeling as a salmonine management tool applied to
Minnesota waters of Lake Superior. Northern American Journal of Fisheries
Management 15:60-78.

Ney, J. J. 1990. Trophic economics in ﬁsheries: Assessment of demand-supply
relationships between predators and prey. Reviews in Aquatic Sciences 2: 55-81.

Ney, J. J. 1993. Bioenergetics modeling today: Growing pains on the cutting edge.
Transactions of the American Fisheries Society 122: 736-748.

Oglesby R. T. 1977. Relationships of ﬁsh yield to lake phytoplankton standing crop,
production, and morphoedaphic factors. Journal of the Fisheries Research Board
of Canada 34 (12): 2271-2279 1977

Pazzia I., M. Trude], M. Ridgway, and J. B. Rasmussen. 2002. Inﬂuence of food web

structure on the growth and bioenergetics of lake trout (Salvelinus namaycush).
Canadian Journal of Fisheries and Aquatic Sciences 59: 1593-1605.

80

Petersen J. H. and D. L. Ward. 1999. Development and corroboration of a bioenergetics
model for northern pikeminnow feeding on juvenile salmonids in the Columbia
River. Transactions of the American Fisheries Society 128: 784-801.

Rand P. S., B. F. Lantry, R. O’Gorman, R. W. Owens, D. J. Stewart. 1994. Energy
density and size of pelagic prey ﬁshes in Lake Ontario, 1978-1990: implications

for salmonine energetics. Transactions of the American Fisheries Society 123(4):
519-534.

Rand PS. and DJ. Stewart. 1998. Prey ﬁsh exploitation, salmonine production, and
pelagic food web efﬁciency in Lake Ontario. Can. J. Fish. Aquat. Sci. 55:318-
327.

Rice J. A. and RA. Cochran. 1984. Independent evaluation of a bioenergetics model for
largemouth bass. Ecology 65(3): 732-739.

Rudstam, L. G., P. E. Peppard, T. W. Fratt, R. E. Bruesewitz, D. W. Coble, F. A. Copes
and J. F. Kitchell. 1995. Prey consumption by the burbot (Lota lota) population
in Green Bay, Lake Michigan, based on a bioenergetics model. Canadian Journal
of Fisheries and Aquatic Sciences 52: 1074-1082.

Schaeffer J. S., R. C. Hass, J. S. Diana, and J .E. Breck. 1999. Field test of two energetic
models for yellow perch. Transactions of the American Fisheries Society 128:
414-435.

Schleen, Larry R, Christie, Gavin C., Heinrich, John W., Bergstedt, Roger A., Young,
Robert J ., Morse, Terry J ., Lavis, Dennis S., Bills, Terry D., Johnson, James E.,
Ebener, Mark P. in press. Development and implementation of an integrated
program for control of sea lampreys in the St. Marys River. Journal of Great
Lakes Research, SLIS 11 Special Issue.

Smith, S. H., H. J. Buettner, and R. Hile. 1961. Fishery Statistical Districts of the Great
Lakes. Great Lakes Fishery Commission Technical Report No. 2.

Smith, S. H. 1972. Factors of ecological succession in oligotrophic ﬁsh communities of
the Laurentian Great Lakes. Journal of the Fisheries Research Board of Canada
29: 717-730.

Stewart, D. J ., J. F. Kitchell and L. B. Crowder. 1981. Forage ﬁshes and their salmonid
predators in Lake Michigan. Transactions of the American Fisheries Society.
110: 751-763.

Stewart, D. J ., D. Weininger, D. V. Rottiers, and T. A. Edsall. 1983. An energetics

model for lake trout, Salvelinus namaycush: application to the Lake Michigan
population. Canadian J oumal of Fisheries and Aquatic Sciences 40: 681-698.

81

Stewart, D. J. and M. Ibarra. 1991. Predation and production by salmonine ﬁshes in
Lake Michigan, 1978-88. Canadian Journal of Fisheries and Aquatic Sciences 48:
909-922.

Swenson WA. and LL. Smith Jr. 1973. Gastric digestion, food consumption, feeding
periodicity, and food conversion efﬁciency in walleye (Stizostedion vitreum
vitreum). Journal of the Fisheries Research Board of Canada 30: 1327-1336.

Tody, W.H., and H.A. Tanner. 1966. Coho salmon for the Great Lakes. Fish Management
Report No. 1, Fisheries Division, Michigan Department of Natural Resources.

Lansing, MI. 38 p.

U. S. v. Michigan. 2000. U. 8. District Court, Western District of Michigan, Case No.
2:73 CV 26.

82

Table 3.1 — Mean energy density used for sensitivity analyses. Lake Huron energy

density was determined in Chapter 2. These represent the mean basin-wide values. ‘*’

represent species that used seasonally or regionally varying energy density in the

bioenergetics models. Published energy density values represent the mean value for all

noted references for the species.

 

Lake Huron energy Mean published

 

 

Species density energy density References
Jog" wet weight Jog’1 wet weight

Alewife* 4,187 6,232 Rottiers and Tucker (1982);
Flath and Diana (1985);
Rand et al. (1994)

Burbot 5,630 4,661 Rudstam et al. (1995);
Johnson et al. (1999)

Bloater 5,514 8,665 Rottiers and Tucker (1982);
Vondracek et al. (1996)

Chinook salmon 5,575 6,678 Cummins and Wuycheck (1971);
Rottiers and Tucker (1982)

Lake trout“ 8,190 10,294 Rottiers and Tucker (1982);
Johnson et al. (1999);
Madenjian and O'Connor (1999)

Rainbow smelt“ 5,151 5,269 Cummins and Wuycheck (1971);
Rottiers and Tucker (1982);
Lantry and Stewart (1983);
Rand et al. (1994);
Rudstam et al. (1995);
Vondracek et al. (1996)

Wallgye 6,435 4,605 Kelso (1972)

 

83

Table 3.2 -- Gross conversion efﬁciency estimated from bioenergetics models

 

 

Burbot Chinook salmon Lake trout Walleye

Age 1973-84 1985-99 North Central South Sag Bay South
0 0.321 0.316
1 0.078 0.247 0.254 0.215 0.171 0.218
2 0.066 0.185 0.171 0.195 0.192 0.175 0.168 0.185
3 0.083 0.083 0.079 0.148 0.156 0.139 0.174 0.189
4 0.082 0.059 0.066 0.118 0.130 0.116 0.154 0.173
5 0.077 0.105 0.114 0.105 0.151 0.170
6 0.072 0.108 0.111 0.110 0.142 0.160
7 0.068 0.092 0.095 0.094 0.129 0.147
8 0.069 0.081 0.084 0.085 0.118 0.135
9 0.066 0.072 0.076 0.077 0.107 0.123
10 0.064 0.066 0.069 0.070 0.092 0.106
11 0.062 0.060 0.064 0.065 0.079 0.091
12 0.060 0.056 0.060 0.061 0.079 0.092
13 0.058 0.053 0.056 0.057
14 0.057 0.051 0.053 0.072

15+ 0.054 0.042 0.055 0.063

 

Table 3.3 — Estimates of mean consumption in millions of kg for 1996-1998

 

 

Burbot Chinook Lake Trout Walleye Total
Alewife 1.1 10.7 3.4 1.9 17.1
Rainbow smelt 1.0 6.9 3.4 0.5 11.9
Other 3.3 1.9 0.3 1.4 6.9
Total 5.4 19.6 7.1 3.8 35.9
Alewife+Smelt 2.1 17.7 6.8 2.4 29.0

 

84

Table 3.4 — Summary of predator consumption and forage demand in the main basin of

Lake Huron, 1984-1998.

 

 

Year Consumption Alewife and Total Prey available Consumption
by all rainbow smelt predator per predator per predator
predators abundance biomass biomass biomass
(millions kgL (millions kg) (millions kg) (kg/kg) (kg/kg)
1984 26.47 59.78 9.18 6.51 2.88
1985 28.54 123.59 8.91 13.87 3.20
1986 33.30 68.99 9.93 6.95 3.35
1987 33.29 160.91 10.38 15.50 3.21
1988 32.15 79.65 11.02 7.23 2.92
1989 32.32 96.21 10.54 9.13 3.07
1990 32.47 44.70 10.02 4.46 3.24
1991 29.95 46.67 10.02 4.66 2.99
1992 28.89 47.58 10.08 4.72 2.87
1993 29.10 45.99 9.91 4.64 2.94
1994 31.27 104.50 9.89 10.56 3.16
1995 32.86 67.58 10.47 6.46 3.14
1996 32.64 52.80 10.86 4.86 3.01
1997 33.92 42.61 11.38 3.75 2.98
1998 41.04 34.90 11.61 3.01 3.53

 

85

 

Region

MH1 U
0H1
Northern
Region 0H2
Central /
ﬂ ”'3

Southern /

Region

Saginaw
Bay y

Figure 3.1 — Statistical districts in the US and Canadian waters of Lake Huron (Smith et
al. 1961) grouped into lake regions. Statistical districts, used in sampling to denote

location, are shown as MH- (Michigan waters) or OH- (Ontario waters)

86

6.0

Chinook
salmon

Gross production (millions of kg)

Burbot

 

1 984 1 986 1 988 1 990 1 992 1 994 1 996 1 998
Year

Figure 3.2 - Estimated gross production of key predators in the main basin of Lake

Huron from 1984—1998.

87

40.0
36.0
' 32.0

- 28.0
Chinook

24-0 salmon

20.0
16.0

12.0

Consumption (millions 0 kg)

9°
0

4.0

Burbot

 

0.0
1 984 1 986 1 988 1 990 1 992 1 994 1 996 1 998

Year

Figure 3.3 — Estimated total consumption by key predators in the main basin of Lake

Huron from 1984-1998

88

 

 

 

 

30:0 T —I—Consumption T 5'0
- -o - Biomass i

. 250—:
a ‘ ’61
x ‘ x
E 15.0 - E
.5 . a
a i g
g 10.0: .9
re d m
0 .
o .

5.0 j

0.0:vir+1lvtritivivlvlvlrtrivi+ivi0.0

1968 1972 1976 1980 1984 1988 1992 1996
Year

Figure 3.4 - Estimated chinook salmon consumption and population biomass in the main

basin of Lake Huron from 1968-1998.

89

12.0

d
a

. ‘ Chinook
salmon

Lake trout

Standing biomass (millions of kg)

 

Burbot

 

1 984 1986 1988 1 990 1 992 1 994 1 996 1 998
Year

Figure 3.5 — Estimated biomass of key predators in the main basin of Lake Huron from

1984-1998.

90

Lake Trout
9-00 —0— Northern

8.00 - 4- -— Central

 

+ Southern

Consumption (millions of kg)

 

 

1984 1986 1988 1990 1992 1994 1996 1998
Year

Walleye —o— Saginaw Bay

450 - mitt- Southern

Consumption (millions of kg)

 

 

 

O.m I I l I I I l

1984 1986 1988 1990 1992 1994 1996 1998
Year

Figure 3.6 — Estimated annual consumption by lake trout and walleye by lake region,

1984-1998

91

45.0 -

 

 

 

’61
x
‘8
(D
C
g 25.0 —
E,
g 20.0 .
'3
§ 15.0 -
U)
8
o 10.0 ~
—I— Current Key Predators
5'0 7 — — Historic Lake Trout
0,0 ' l I I ' l I I I l ' l ' I
1984 1986 1988 1990 1992 1994 1996 1998

Year

Figure 3.7 — Comparison of estimated current key predator consumption in the main

basin of Lake Huron to estimated consumption by pro-collapse lake trout.

92

 

 

 

 

 

 

 

 

 

10070 1
80% ~
E 3.3
3 60% - m
o
r:
.2
‘1:
8- 40°/
0 _
2
“ 111
10 7 1_9
20% '1
1.1
0% “E t . _._ —--
Burbot Chinook Lake Trout Walleye

 

|'_'l Alewife I Rainbow smelt [:1 Other

 

Figure 3.8 — Diet composition of the key predators in the main basin of Lake Huron.

Proportion of each prey type in the diet represents the mean by weight for 1989 — 1999.

93

 

 

180 I - -A- - Consumption by predators
160 _‘ —I—Alewife& smelt biomass
140 {
3120{
P :
w .
g; 100:
I 801
(D
I“ :
<E> 60 ‘
E
40 . A
’ .*"""A--A--A-.‘__‘__‘._‘.-A--‘--A
20
O F i l i l T l r I i 1 T T I 1
1984 1986 1988 1990 1992 1994 1996 1998

Year

Figure 3.9 — Comparison of estimated key predator consumption to estimated combined

alewife and rainbow smelt biomass in the main basin of Lake Huron from 1984-1998.

94

 
 
 

 

   

 

 

‘ A A A A A A A
2"" ~ ..
‘5 40'0 I ,3 wgﬁmae~a~a~aee~e
g
E, 30.0 —
C
.9
a -t
5, 20.0 —_
C
O
0 ‘ +Baseline
10:0 : +Chinook stocking reduction
- 1» - Sea lamprey reduction
0.0 '7I"'iriui"rrr'il"'i
1984 1996 2000 2004 2008 2012 2016 2020

Year

Figure 3.10 - Projected consumption by key predators in the main basin of Lake Huron

through 2020 under three management scenarios.

95

Chapter 4

Parameterization of a Functional Response Model For Chinook Salmon In
The Main Basin of Lake Huron.

Introduction

Stocking of chinook salmon Oncorhynchus tshawytscha in Lake Huron tributaries
began in 1968 and has increased from 265,000 to approximately 4 million ﬁsh annually
(Ebener et al. 1995). Natural reproduction was not detected before 1988 (Ebener et al.
1995) but current levels of wild recruitment are believed to be approximately 15% of
total recruitment, although the actual amount is uncertain and may be much greater.
Increases in the number of chinook salmon stocked along with improvements in survival
of stocked ﬁsh and possible increases in wild recruitment account for approximately 60%
increase in abundance and consumption from the mid 19803 to peak values in the late
19903 (Chapter 3).

Stocking also inﬂuences the abundance of other piscivores in Lake Huron,
especially lake trout Salvelinus namaycush, with hatchery-reared ﬁsh constituting the
majority of recruitment (Ebener et al. 1995). Additionally, recent attempts have been
made to reduce the abundance of sea lamprey Petromyzon marinus (Bergstedt et al.
1998), a parasite that causes signiﬁcant mortality to lake trout. Improving the survival of
lake trout should increase their abundance and consumptive demand on the forage base.
While all of the piscivores share the same forage base, the fast growing chinook salmon
and long-lived lake trout take the largest proportion of the available prey ﬁsh, primarily
consuming the exotic species alewife Alosa pseudoharengus and rainbow smelt Osmerus

mordax. (Chapter 3).

96

Increases in salmonine stocking, unknown quantity of chinook salmon wild
recruitment, and various management actions that may increase lake trout abundance
have led to concerns that piscivore abundance could exceed the forage ﬁsh availability.
In Lake Michigan, declines in alewife abundance during the early 19803 precipitated
numerous changes throughout the Lake Michigan food web (Kitchell and Crowder 1986)
and may have caused the collapse of chinook salmon in Lake Michigan (Holey et al.
1998; Benajmin and Bence In press (a); Benajmin and Bence In press (b)). Total
abundance of alewife and rainbow smelt, the main constituents in the diet of Lake Huron
chinook salmon, have also varied nearly fourfold between 1974 and 1998 (Figure 4.1).
Between 1974 and 1984 chinook salmon growth declined in Lake Huron and although
there have been subsequent years with improved growth, it has not recovered to the pre-
1984 levels (Figure 4.2). While changes in prey abundance are often associated with
changes in growth, this relationship is not clearly evident for chinook salmon in Lake
Huron (Figure 4.3). Lacking critical data on the relationship between growth and prey
density, and concerned that Lake Huron predators may be exceeding forage ﬁsh capacity,
management agencies decreased chinook salmon stocking by 20% in 1999 in an attempt
to avoid a possible collapse of the predator populations.

In Lake Huron, predator forage demand and the effects of changes in prey fish
abundance on predator growth are not well understood. The amount of prey eaten and
the composition of the diet depend upon prey availability in unknown or only partially
understood manners. Researchers studying Lakes Michigan (Stewart et al. 1981; Stewart
and Ibarra 1991; Eby at al. 1995), Ontario (Jones et al. 1993), and Superior (Mason et al.

1998) have used various approaches including bioenergetics models, foraging theory, and

97

functional response models to help clarify predator-prey dynamics in those lakes. We
developed bioenergetics models and coupled them with age-structured population models
of the key predators in Lake Huron (Chapter 3). Estimates from this effort showed that
chinook salmon predation accounts for 54% of the total annual consumption of open-
water prey ﬁshes between 1996 and 1998. However, this approach does not predict how
consumption changes with variations in prey densities or how changes in the forage base
impact predator growth. Linking changes in growth to changes in prey density may
provide an indicator of disruptions in the balance between predator numbers and prey
abundance, and where predator abundance is primarily supported through stocking, allow
ﬁshery managers to reduce stocking and avoid a possible collapse of the predator
population.

A functional response model (Holling 1959) is needed to link predator
consumption with prey density. We developed a functional response model that
estimates the number of prey ﬁsh consumed by chinook salmon in the main basin of Lake
Huron based on prey abundance. Growth was linked to consumption through the
conversion of food ingested to changes in body mass.

The ﬁnal model ﬁtting was done by varying the numbers of search rate
parameters to test four hypotheses. In Model 1, our hypothesis was that the search rate
was independent of predator age or prey type being consumed. Since differences in prey
behavior or other species-speciﬁc factors can affect a predator’s reaction to prey, in
Model 2 we tested the hypothesis that prey type affects consumption by associating a
separate search rate parameter with each prey species. In Model 3, we evaluated the

effect of predator age on the model. Age 1 chinook salmon possess several unique

98

attributes not found in older fish. For instance, age 1 ﬁsh grow at a much faster rate than
other age classes (Figure 4.2) and they selectively consume smelt while other age classes
select for alewife (Appendix B Table B.3). Therefore, in Model 3 search rate parameters
are dependent on predator age but not prey type, with age 1 ﬁsh and ages 2—4 forming
two age groups. To evaluate the combined effects of predator age and prey species,

Model 4 allows search rates to vary by prey type and predator age.

Methods

Our goal was to develop a model that predicts annual consumption of prey by an
individual chinook salmon based on the abundance of prey of each type (species and size
category) and the size of chinook salmon. Symbols used in equations throughout this
document are given in Table 4.1. Equations not given in the text are in Table 4.2. We

used a multi-species Type II functional response (Holling 1959 and Murdoch 1973)

S N.t

P _ j,y,a,b sz9b a
j,y,a,b _

(1)
1+Z(hj,y.a,b Si,y.a,b Nj.y.b)
j

which predicts consumption of prey (P) in year y by a chinook salmon of age a based on
prey abundance (N) of each type (1) and size category (b). The search rate (S) and
handling time (h) are related to chinook salmon size and its inﬂuence on a predator’s
ability to locate, catch, and digest its prey. The amount of time spent foraging in the lake
(t) adjusts for age 4 chinook salmon that spawn and die before the end of the year. While

chinook salmon consume other prey items, the vast majority of prey eaten consists of

99

alewife and rainbow smelt (Appendix B Table B.3); therefore, the functional response
model includes only these two prey species.

We assumed the search rate depended on predator length, the ratio of prey to
predator length, and dietary preference (Table 4.2) in a known fashion with these effects
operating in a multiplicative way, following Jones et al. (1993). First, search rate was
assumed to be directly proportional to predator length because swimming speed is
proportional to predator length. Second, the relative search rate was adjusted using a
dome-shaped “preference” function (Figure 4.4) determined by the ratio of prey to
predator length, which peaked at an optimal ratio of 0.25 (Jones et al. 1993). Finally,
based on recent dietary studies, age-1 chinook salmon were assumed to prefer rainbow
smelt over alewife, whereas older ages were assumed to prefer alewife to rainbow smelt.
These effects only set the relative search rates for different prey types. When the model

was ﬁt to observed data (see below), an unknown scalar (Ola, Table 4.1) that determined

absolute search rates was estimated. Additionally, when search rates for alewife and
rainbow smelt were allowed to differ (Models 2 and 4), predator diet composition is not
held constant by the dietary preference assumption but allowed to vary with prey
abundance.

Handling times depended upon predator and prey sizes following relationships
assumed to be known (Table 4.2, Figure 4.5). Based on results from bioenergetics
modeling (Chapter 4 Appendix), handling times decreased with predator size because the
maximum mass of prey that could be consumed in a year (Cmax) increased with chinook
salmon size (Figure 4.5). Conversely, handling times increased with prey size because

larger prey weigh more. For age 4 chinook salmon, Cmax was lower (Figure 4.5)

because we only predicted consumption for this age through the time of spawning (day
214).

Data available for parameterizing the model included chinook salmon annual
weight-at-age, and annual prey abundance by type and size category for a time-series
extending from 1974 through 1998. Chinook salmon weight-at-age was used in two
different ways in the model. First, weight-at-age was used to determine chinook salmon
length, an important component in the handling time and search rate. Second, annual
changes in weight-at-age provided estimates of chinook salmon growth. We needed
estimates of observed growth because we lacked direct estimates of consumption to
compare with model predictions. Instead, we used equation 1 to estimate consumption of
prey given prey abundances and then converted these estimates of individual
consumption into predictions of individual chinook salmon growth. We then compared
the predicted growth to observed growth, which was calculated from the annual change in
weight-at-age.

Chinook salmon weight-at-age information was based on a combination of data
from creel surveys and sampling spawning runs (Bence and Dobiesz 2000). Direct
observations of weight were not available for some year and age combinations or were
represented by very small sample sizes. A catch-at-age model for chinook salmon in
Lake Huron included a dynamic von Bertalanffy growth model (e.g., Szalai et al. 2003)
and produced a smoothed estimate of weight-at-age over time to account for large
measurement errors (Bence and Dobiesz 2000).

Prey abundance was obtained from annual fall bottom trawl surveys of US waters

in Lake Huron conducted by the USGS Great Lakes Science Center. When using this

101

survey method, numbers of ﬁsh in each trawl are expanded from the actual area trawled
to all US waters based on the area swept by the trawls in different regions of the lake at a
series of depth strata, accounting for the total bottom area in each region and depth
station. Mid-year values of abundance for each prey species and size category were used
in the functional response (Chapter 4 Appendix).

Equation 1 predicts the numbers of each prey type and size category consumed by
a predator of a given age. Total biomass consumed by a predator was determined by
multiplying the predicted numbers of each prey type and size category by the associated
prey weight and summing over all prey sizes and prey types (Table 4.2). We converted
predicted biomass consumed into chinook salmon growth (increment in weight) using an
estimate of gross conversion efﬁciency (GCE) obtained from bioenergetics models
(Chapter 4 Appendix).

The overall model ﬁt was measured by the concentrated negative log-likelihood:

_ 111“.) = 100

 

10g [ Z Z (Gobserved _ G predicted ) 2 ] (2)
y a

which was minimized using a quasi-Newton numerical approach to adjust the unknown
parameters using ADModel Builder (Otter Research 2000). Inferences based on this
objective function depend upon the assumption that deviations from expected growth
were normally distributed. Estimates obtained from this concentrated likelihood are
equivalent to those obtained from the full negative log-likelihood equation, but the
numerical search is simpliﬁed because the residual variance is obtained analytically

rather than as an additional parameter adjusted during the search. We note that the

102

resulting point estimates are simply least squares estimates and the use of the
concentrated likelihood only plays a role when making inferences.

The ﬁnal model ﬁtting was done by varying the numbers of parameters to
evaluate the following hypotheses that the search rate: (1) was the same for all chinook
salmon ages and both prey types; (2) varied by prey type; (3) varied between age 1 and
age 2-4 chinook salmon but was the same for each prey type; (4) was dependent upon
prey type and predator age. We computed the Akaike’s Information Criterion (AIC)

(Hilbom and Mangel 1997) for each conﬁguration to compare the models.

Sensitivity Analysis

In the process of estimating the unknown search rate parameter(s), several
quantities were treated as known including chinook salmon maximum consumption, the
average day of consumption, and the size preference function shape variables. The
effects of these values on the model estimates were evaluated by reﬁtting the model using
alternative values for each quantity in turn (Table 4.3). In addition to computing AICs,
the estimated minimum and maximum values of the proportion of maximum ration
(Pmax) will be used to compare the effects of these assumed quantities.

Age-speciﬁc maximum consumption (Cmax) by chinook salmon plays a key role

in determining the handling time. A scalar, ka, was used to proportionally increase or

decrease (Table 4.3) the value of Cmax obtained from the length-dependent function
(Chapter 4 Appendix) by i20%. In a third alternative case, Cmax was held constant for

each age at the 1974 level, a time of high predator growth (Table 4.3). This represents an

103

extreme case but should evaluate our assumption that Cmax changed over time as
chinook salmon growth declined.

The average day of consumption (Chapter 4 Appendix) was used to adjust prey
abundance and predator length to a mid-year value. Changes in consumption by each age
caused the mid-year value to be different for each age, although ages 2 and 3 were almost
identical (Chapter 4 Appendix). To evaluate the effect of this age-speciﬁc mid-year
adjustment, we reran the model using the calendar mid-year, day 182, for all ages.

The parameters of the size preference function (Tables 4.1 and 4.2) were
borrowed from a functional response model for Lake Michigan (Jones et al. 1993, E.
Szalai, Pers. Comm). It is based on the optimum prey to predator length ratio of 0.25,
with “preference” declining above and below that ratio. In the standard model, we
treated both prey types the same. We made two changes to the size preference function,
and evaluated how sensitive the model was to the joint effect of these changes. First,
because of differences in body dimensions, alewife of a given length tend to weigh more
than rainbow smelt of the same length. Using the length-weight relationship for each
prey species (Chapter 4 Appendix), we determined that at equal mass, an alewife would
be 84% of the length of a rainbow smelt. We applied this percentage to the optimal prey
to predator length ratio, setting it to 0.21 for alewife while keeping the 0.25 ratio for
rainbow smelt (Figure 4.4). Second, we noted a signiﬁcant lack of consumption of small
prey sizes during model ﬁtting. To increase the preference for the smallest prey sizes, we
adjusted the left-hand limb of the size preference curve (Figure 4.4) by changing v in the

size preference function (Table 4.2) from 0 (Table 4.3). A different value of w (Tables

104

4.2 and 4.3) was chosen for the two species to avoid unintended effects to the right-hand

side of the preference function (e. g., negative values).

Results

We compared the ﬁt of four functional response models (denoted as Models 1-
4), with different search rate parameterizations, to observed growth of chinook salmon.
We used likelihood ratio tests (Berry and Lindgren 1996) to compare models with
different numbers of parameters and the AIC (Hilbom and Man gel 1997) to determine
the ﬁnal model (Tables 4.4 and 4.5). Model 1, with a single estimated parameter, ﬁt
observed growth poorly for all predator ages (Table 4.4). Although Model 2 was an
improvement over Model 1, its predictions for all predator ages substantially exceeded
observed growth during the 19903 (Figure 4.6). Model 3 had a lower AIC than Models
1 and 2 (Table 4.5) and its predictions matched observed growth better during the
second half of the time series (Figure 4.7). Model 4 matched observed growth
somewhat better than either Models 2 or 3, and had a lower (better) AIC than Models 1
through 3. Increasing the number of estimated search rate parameters from one to two,
either to distinguish predator groups or prey species, signiﬁcantly improved the ﬁt of
the model (Table 4.4). Increasing the number of parameters to four, to allow a unique
search rate parameter for each combination of prey species and predator group,
provided a closer match to observed growth as compared with models 2 and 3 (Table
4.4).

Although Model 4 outperformed the other models, there were three speciﬁc

areas where the model predictions did not match observed values. First, growth for age

105

31'

'l‘?

in

2 was overestimated in all but three years. The substantially better match to observed
data for other ages obscured this outcome in the AIC. Since age 2 chinook salmon
share many of the same attributes with age 3 ﬁsh (i.e., diet and growth rate), the reasons
for the differences in how the model ﬁts growth for these ages are not obvious.

Second, Model 4, like each of the other models, missed a sudden increase in growth
between 1989 and 1991 occurring in each age (Figure 4.7). Third, Model 4 failed to
match the decline in growth of age 1 ﬁsh during 1987 and 1988. It appears that no
functional response model of the type we considered would predict the increase in
growth during 1989-1990, because prey abundance of both rainbow smelt and alewife

were decreasing at this time.

Consumption

Over the modeled time series, prey abundance has varied dramatically from
year to year (Figure 4.1). We expected to see a response in consumption to these
varying levels of prey abundance, especially since growth varied over time (Figure 4.2).
However, the functional response predictions of consumption of prey biomass change
much less than proportionately with total prey biomass (Figure 4.8). There are
substantial variations in predicted consumption, unrelated to total prey abundance,
which stem from the composition of prey types and changes in predator size-at-age.
However, the pattern in Figure 4.8 suggests conditions where predators may be feeding
near their maximum capacity.

To better illustrate how predictions of consumption respond to prey abundance,

the composition (percent of each type) was ﬁxed at the average proportions seen between

106

1985 and 1996, and prey abundance was set to ﬁxed values ranging between 3.3E+08 and
8.26E+09, which spanned the observed total prey abundance. Predator weight-at-age was
ﬁxed at either a high level (1974) or a low level (1984). The four estimated search rate
parameters from Model 4 (Table 4.5) were used to generate predictions of per capita
consumption (Figure 4.9). At the lowest observed prey abundance, the functional
response model is predicting that consumption is increasing much less than
proportionately to increases in prey abundance. Ages 2 and 3 being the fastest growing
ﬁsh in the model have the lowest handling time and therefore are not as close to their

saturation value.

Growth

Since our previous analyses suggested that variations in growth were only
weakly tied to prey abundance, the root cause for the substantial changes in size-at-age
over time remains unclear. To explore this we examined the relationship between the
consumption by a cohort and its initial size at age-1 (Figure 4.10). When age 1 ﬁsh
were smaller for any given cohort, subsequent ages within that cohort grew less and
consumed less prey biomass than cohorts that began age 1 at a larger size. The
regression model predicted a 28% decrease in estimated consumption between the

cohorts with the smallest age 1 ﬁsh (1984) and the largest (1974).

Compgrison to bioenergetics models
Using bioenergetics models with Lake Huron speciﬁc data (Appendix B) we

generated estimates of age-speciﬁc annual consumption for an average chinook salmon

107

(Chapter 3). The functional response model produced estimates of numbers of prey
consumed, which we converted into estimates of biomass. Comparing the estimates
from these two models shows that they are similar and track the downward trend in
consumption over time (Figure 4.11). The models tended to estimate very similar
consumption for ages 1 and 3. However, the functional response model tended to

estimate higher consumption for ages 2 and 4 than the bioenergetics model.

Sensitivity analysis

With Cmax values reduced by 20%, estimated search rate parameters were
larger than Model 4 parameters (Figure 4.12) and the AIC was the lowest of all
alternatives analyzed (Table 4.6). Lowering Cmax also lowers the minimum and
maximum values of Pmax obtained in the model (Figure 4.13). The effects of higher
Cmax values, produced by increasing the base by 20% or by using Cmax values ﬁxed at
1974 levels, was to lower the values of the estimated search rate parameters from those
in Model 4 (Figure 4.12) and increase the AICs (Table 4.6). The range of Pmax values
is more highly affected by ﬁxing the Cmax value than by increasing it by a ﬁxed
amount (Figure 4.13).

Changing the prey size preference function did not have a large impact on the
model parameter estimates (Table 4.6) or minimum and maximum estimated for Pmax
(Figure 4.12). There is only a slight increase in the maximum Pmax values for ages 2-
4, whose diet preference favors alewife.

Setting the adjustment day to the actual middle of the year produced estimates

of the search rate parameters that were higher than those estimated by Model 4 (Figure

108

 

4.13). Changing the mid-year adjustment day to day 182 slightly lowered the AIC
(Table 4.6) and had its biggest effect on the estimates of search rate parameters for age
1 (Figure 4.13).

Overall, our assumptions regarding the prey size preference and the mid-year
adjustment day had a much smaller effect on the model than changes to Cmax.
Handling time sets the upper limit on consumption and is inversely related to Cmax.
Additionally, each sensitivity analysis produced some changes in Pmax when compared
to Model 4 but direction of these changes were essentially the same across predator
ages (Figure 4.13), although ﬁxing Cmax at 1974 levels substantially reduced both the

minimum and maximum values of Pmax.

Discussion

Studies of Lakes Michigan and Ontario (Stewart et al. 1981; Jones et al. 1993)
have shown the potential for stocked salmonids to outreach the forage ﬁsh capacity. In
Lake Huron, chinook salmon growth declined between 1974 and 1998 (Figure 4.2)
leading to concerns that predator growth was being limited by forage ﬁsh availability.
Since chinook salmon are the dominant predator in Lake Huron (Chapter 3), we
parameterized a functional response model to evaluate how chinook salmon consumption
was affected by prey abundance. We converted these estimates of consumption to
estimates of chinook salmon growth using GCEs estimated from bioenergetics models.
While our analysis did not include all factors that inﬂuence chinook salmon growth, we

expected that if variations in prey abundance were a primary determinant of chinook

109

salmon growth, and prey availability were limiting during the time period evaluated, this
would be uncovered by ﬁtting a functional response model.

Growth is closely tied to consumption but varies with food availability, food
quality, water temperature, time of hatching, gonad production, age, and activity costs,
making it difﬁcult to ﬁnd a simple relationship between growth and consumption
(Boisclair and Leggett 1989a, b; Hewett et al. 1991; Hewett and Kraft 1993). Studies
have attempted to correlate changes in growth with changes in prey abundance with
varying success (e.g., Stewart and Ibarra 1991; Breck 1993; Eby et al. 1995). Our
functional response model attempted to uncover a more subtle relationship by taking into
account variations in prey species and size composition. However, our analysis suggests
that variations in total consumption (and hence growth) have been only weakly tied to
measured prey abundance (Figures 4.3 and 4.8). Density-dependent effects related to
chinook salmon abundance were not evident (Figure 4.14) suggesting that chinook
salmon could always ﬁnd enough prey to feed close to Cmax.

Our functional response model suggests that over a large range of prey abundance
age 1-4 chinook salmon were feeding above 60% of their maximum rate of consumption
(Pmax) and variations in prey abundance explained little of the variation in observed
growth (Figure 4.14). This was also true when assumed known constants were varied in
the sensitivity analyses, with the exception being when Cmax values were constant over
time and set at values based on size-at-age observed in 1974. One explanation for why
the model predicted that predators were feeding near saturation (i.e., high Pmax values)
could be that observed growth was not related to measured prey abundance in a

straightforward way. The functional response model can only make growth weakly

110

related to large variations in prey abundance if the predicted feeding level is near the
asymptotic feeding rate at the lowest observed prey abundance.

Between 1974 and 1998, chinook salmon size—at-age varied substantially, with an
overall downward trend. The model ﬁtting results suggest that this decline cannot be
explained by variations in prey abundance. Nevertheless, the model was able to predict
some of the observed declines in growth (Figure 4.7). Accepting the model ﬁt at face
value, observed declines in growth must be related to other factors. We noted that
signiﬁcant differences in the weight-at-age 1 followed the cohort through its life span.
Weight-at-age 1 has varied from 1.21 kg in 1974 to 0.712 kg in 1987. Fish that weighed
less at age 1 consistently weighed less throughout their life span than ﬁsh whose weight
at age 1 was higher. The functional response model predicts lower growth of cohorts that
begin age 1 at a smaller size because they have a lower Cmax and less capacity for
growth (Figure 4.10). The nearly constant instantaneous growth rate (Figure 4.2) we
observed suggests that ﬁsh that start out smaller cannot “catch up” to ﬁsh that start out
larger. With the majority of recruitment coming from stocking, age 0 ﬁsh should be
approximately the same size, therefore, factors that effect early growth have an important
impact on subsequent consumption, and these factors were not represented in our model
of growth from age 1 to age 4.

These results have implications for the current mix of stocked and naturally
reproducing chinook salmon in Lake Huron. Studies in other ecosystems have shown
that hatchery-reared chinook salmon are smaller than wild recruits (Roni and Quinn
1995; Unwin and Glova 1997). If this were also true in Lake Huron, wild ﬁsh might

have a signiﬁcant advantage over stocked ﬁsh. If they begin life in Lake Huron at a

111

larger size, they could eat larger prey, and salmonids have been shown to grow larger
when they eat larger prey (Kerr 1971; Mittelbach and Persson 1998; Pazzia et al. 2002).
The cause of annual differences in weight-at-age 1 are unclear but growth has been
shown to be heritable in chinook salmon Mithler et al. 1987) and slower growth in some
cohorts could be driven by prey abundance, but in ways we were unable to uncover.
Another possible explanation for our model results is that the assumed
relationships and constants we used were substantially in error, and there is actually a
stronger relationship between predator consumption and prey availability. Of particular
concern were the assumptions that age-speciﬁc GCEs were constant over time and
estimates of these GCEs were based on maximum chinook salmon growth during 1974
but energy density of predators and prey observed during 1996-1997. Values of Cmax
for a given size chinook salmon were based on this same relationship between maximum
growth and consumption. Our values for energy density (Chapter 2) tended to be lower
than those published in the literature for other lakes and earlier time periods. Lower
energy densities would tend to lower the GCEs. If GCEs declined over time, the amount
of consumption required to achieve the maximum amount of growth, which may be a
physiological limit, might have increased. Thus, Cmax might have increased over time if
energy density of prey ﬁsh declined. If this occurred as we speculated in Chapter 2,
chinook salmon growth may have been limited by available prey even when prey
abundance was not declining. Additionally, since chinook salmon size-at-age changed
over time, Cmax, and therefore handling time, may have shifted in a way that was not

captured by our model.

112

There were also substantial uncertainties associated with our measurements of
predator growth and prey abundance. We lacked annual weight-at-age for chinook
salmon and instead used a dynamic von Bertlanfy growth model to estimate a smoothed
weight-at-age over time, reducing large measurement errors. The type of assessment gear
used to estimate relative prey abundance changed in 1992 with some concern about
proper adjustments to estimates. Also, prey ﬁsh abundance as measured in the fall may
not accurately reﬂect availability of prey to chinook salmon (Eby et al. 1995), or spatial
and temporal changes in prey availability may effect predator consumption (Kerr 1971;
Goyke and Brandt 1993). However, these uncertainties do not seem large enough that
they would obscure a strong relationship between predator growth and prey abundance.

Our intention was to improve our understanding of the linkage between chinook
growth and prey abundance. While we used the best available data, these efforts would
beneﬁt from improved prey assessments that measured changes in seasonal and temporal
patterns of prey ﬁsh availability. Similarly, annual measurements of predator and prey
energy density as well as seasonal diet information could improve model estimates.
Additionally, we examined only predator dynamics but studies that link both predator and
prey dynamics (e. g., Jones et al. 1993) could further enhance our understanding of the

relationship between predator growth and prey abundance.

113

Appendix

This appendix contains details and equations used in the chinook salmon

functional response model.

Day of average consumption

The day of the year when the average consumption occurred was determined
using bioenergetics models that estimate daily consumption. This day was used to adjust
prey abundance and chinook salmon length to a mid-year value. The day of the year

when the average consumption occurs is given by

1211)

where d is the day of the year with January 1St being day 1;
T is the number of days the predator is resident in the lake with ages 1-3
resident for 365 days and age 4 resident for 214 days; and

Cd is the consumption on day d.

The average consumption occurs on day 234 for age 1, day 208 for ages 2-3 and day 150

for age 4.

Prey abundance and size categories

Estimates of prey abundance in US waters were extended to estimates for the

entire main basin using a constant multiplier of 1.767 (G. Curtis, USGS Great Lakes

114

Science Center, Pers. Comm). Prey abundance was divided into 5 mm size categories.
The smallest and largest size categories contained many missing values over the time
series. These were combined into two plus groups representing 10-40 mm and 215-250
mm. Each was treated as a single bin with prey sizes ﬁxed at 40 and 250mm,
representing the most common size [Note: all ﬁsh larger than 250mm were classiﬁed as
250mm]. Each prey size category was adjusted to the average day of consumption by
assuming prey abundance changed exponentially with a constant per capita instantaneous

rate DCIWCCII IWO prey assessments:

a =1n(M er,,,_,,,)/365

13M
N15,!) = M Mb exp(a)' (D0 + 77))

where a; is the instantaneous rate of change between the previous and current

prey assessments, assumed to occur on October 15’“,

M j, y, b is the estimate of prey abundance for prey j in year y and bin b
N13»), is estimated prey abundance on day Da for prey j in year y and bin b,
Do is the day of the year when the average consumption by chinook salmon

occurs for age a, with a constant (77) to adjust for the start date of

October 15th rather than January 15’.

Prey Weight

The functional response model produces numbers of each prey type eaten from

each length bin. The numbers eaten were converted to biomass eaten using a weight-

115

length relationship (J. Schaeffer, USGS Great Lakes Science Center, Pers. Comm) for

each prey species:

_l (4.223x10‘05 be-“Zmooo j = alewife
“’ I(6.935x10’06xb2'945)/1000 j=smelt

where W] b is the mean weight (kg) of prey type j in bin b, and

b is the mid-point of the prey length bin (mm).

Predator Weight and Length

A weight-length relationship was determined from data collected from weir
sampling on the AuSable River, Michigan during 1974-1981 and 1996-1999 (J. Johnson,

Michigan Department of Natural Resources, Pers. Comm.) The length-at-age is given by
LM = exp(6.122) x ( WM) 0325+ 0.0014

where Ly,a is the length-at-age a in year y adjusted to day Da, and

Wyﬂ is the weight-at-age a in year y adjusted to day Da.
Predator length was adjusted to the age-speciﬁc average day of consumption (Da) by ﬁrst
adjusting the weight-at-age to 0,, then applying a weight-length relationship. Chinook

salmon weight-at-annulus was assumed to change exponentially between the start and

end of the year:

a; = ln(Vy+La+1 /VM)/365
WM =VM exp(0 D)

where w is the exponential rate of growth between the start and end of the year,

116

Vy, a is the weight-at-age a in year y at annulus, and
Wy, a is the weight-at-age a in year y adjusted to day Da.
We assumed that age 4 ﬁsh (the last age group) mature and die on day 214 so we used the

weight-at—annulus in the beginning of the year and the fall weight in the same year to

estimate the weight on day Da. The input data also contained an extra year (1999) of

weight-at-age data to allow the weight in the last year to be adjusted to day Da.

Cmax

Handling times were based on estimates of the annual maximum amount of
consumption possible (Cmax) by a chinook salmon of a given size and the mean weight

of a prey ﬁsh in each size category
r0
a )

where ka age-speciﬁc scalar for Cmax used in sensitivity analysis, otherwise set to

C maxa = ka (qa Ly

,

l;
qa age-speciﬁc intercept of power function (Table 4.3) relating length of

predator to Cmax;

Ly, a is the length-at-age a in year y adjusted to day Da; and
ra Slope of power function (Table 4.3) relating length of predator age a to

Cmax

ll7

Cmax was estimated directly from age-speciﬁc bioenergetics models (Chapter 3) that
predicted consumption from observed growth during 1974, the assumed period of
maximum growth. An exponential function relating predator length to Cmax was
developed from these data. A separate relationship was needed for ages 1-3 and age 4
since the annual maximum consumption of age 4 ﬁsh is limited by their maturation in the
time step (Figure 4.5). However, the difference between ages 1-3 and age 4 was not
proportional to the number of days spent in the lake, especially for larger fish. The
weight of a prey ﬁsh in each size category, the other key element in estimating handling

time, was determined from a weight-length relationship (see above).

Gross Conversion Efﬁciency

Using a bioenergetics model (Appendix B), age-speciﬁc GCEs were estimated
from observed growth during 1974, and were representative of the mean over the time

series 1974-1998 (Figure 4-15). GCEs were 0.226 for age 1, 0.140 for age 2, 0.130 for

age 3, and 0.066 for age 4 chinook salmon.

118

Literature Cited

Benajmin, D.M. and J .R. Bence. In press (a). Spatial and temporal changes in the Lake
Michigan Chinook Salmon ﬁshery, 1985-1996. Michigan Department of Natural
Resources, Fisheries Division, Fisheries Research Report.

Benjamin, D.M. and LR. Bence. In press (b). Statistical catch-at-age assessment of
Chinook salmon in Lake Michigan, 1985-1996. Michigan Department of Natural
Resources, Fisheries Division, Fisheries Research Report.

Bence J. R. and N. E. Dobiesz. 2000. Estimating forage ﬁsh consumption by predators
in Lake Huron. Great Lakes Fishery Commission Project Completion Report.
Available for download at http://www.glfc.org/research/cap.htm.

Bergstedt, R., M. Jones, G. Christie, K. Mullett, J. Heinrich, J. Adams, R. Young,
M.Fodale, R. McDonald. 1998. St. Marys River Assessment Plan. Submitted for
external review to the GLFC Sea Lamprey Integration Committee, April 22, 1998.

Berry, D. A. and B.W. Lindgren. 1996. Statistics: Theory and Methods. Duxbury Press,
Belmont, Albany, Bonn, Boston, Washington. p 476-479.

Boisclair, D. and WC. Leggett. 1989a. Among-population variability of ﬁsh growth:
I Inﬂuence of the quantity of food consumed. Canadian Journal of Fisheries
and Aquatic Sciences 46: 457-467.

Boisclair, D. and WC. Leggett. 1989b. Among-population variability of ﬁsh growth:
H Inﬂuence of prey type. Canadian Journal of Fisheries and Aquatic Sciences
46: 468-482.

Breck, J.E. 1993. Foraging theory and piscivorous ﬁsh: are forage ﬁsh just big
zooplankton? Transactions of the American Fisheries Society 122: 902-911.

Ebener, M. P., J. E. Johnson, D. M. Reid, N. P. Payne, R. L. Argyle, G. M. Wright, K.
Kruger, J. P. Baker, T. Morse and J. Weise. 1995. Status and future of Lake
Huron ﬁsh communities. In Munawar, M., T. Edsall and J. Leach (eds.), The
Lake Huron ecosystem: Ecology, Fisheries and Management. Ecovision World
Monograph Series, SPB Academic Publishing, Amsterdam, The Netherlands,
pp.125-170.

Eby, L.A., L.G. Rudstram and J .F. Kitchell. 1995. Predator responses to prey population

dynamics: An empirical analysis based on lake trout growth rates. Canadian
Journal of Fisheries and Aquatic Sciences 52(7): 1564-1571.

119

Goyke A. P. and S. B. Brandt. 1993. Spatial models of salmonine growth rates in Lake
Ontario. Transactions of the American Fisheries Society 122: 807-883.

Hewett S. W., C. E. Kraft, and B. L Johnson. 1991. Consumption, growth, and
allometry: a comment on Boisclair and Leggett (1989a, 1989b, 1989c, 1989d).
Canadian Journal of Fisheries and Aquatic Sciences 48: 1334-1337.

Hewett, S. W. and C. E. Kraft. 1993. The relationship between growth and
consumption: Comparisons across ﬁsh populations. Transactions of the American
Fisheries Society 122: 814-821.

Hilbom, R. and M. Mangel. 1997. The Ecological Detective. Princeton University Press,
Princeton, New Jersey, p 159-160.

Holey, M. E., R. F. Elliott, 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: 202-210.

Holling, CS. 1959. The components of predation as revealed by a study of small-
mammal predation of the European pine sawﬂy. Can.Entomol. 91: 293-320.

Jones, M. L., J. F. Koonce and R. O’Gorman. 1993. Sustainability of hatchery-dependent
salmonine ﬁsheries in Lake Ontario: The conﬂict between predator demand and
prey supply. Transactions of the American Fisheries Society 122: 1002-1018.

Kerr, SR. 1971. Prediction of growth efﬁciency in nature. Canadian Journal of
Fisheries and Aquatic Sciences 28: 809-814.

Kitchell, J. F. and L. B. Crowder. 1986. Predator-prey interactions in Lake Michigan:
model predictions and recent dynamics. Environmental Biology Of Fishes 16: 205-
211.

Mason, D.M., T.B. Johnson, and J.F. Kitchell. 1998. Consequences of prey ﬁsh
community dynamics on lake trout (Salvelinus namaycush) foraging efﬁciency in

Lake Superior. Canadian Journal of Fisheries and Aquatic Sciences 55: 1273-1284.

Mittelbach, G.G. and L. Persson. 1998. The ontogeny of piscivory and its ecological
consequences. Canadian Journal of Fisheries and Aquatic Sciences 55: 1454-1465.

Murdoch, W.W. 1973. The functional response of predators. Journal of Applied
Ecology 10: 335-342.

Otter Research. 2000. An introduction to AD Model Builder Version 4 for use in
nonlinear modeling and statistics. Otter Research Ltd., Sidney, BC, Canada.

120

Pazzia I., M. Trude], M. Ridgway, and J. B. Rasmussen. 2002. Inﬂuence of food web
structure on the growth and bioenergetics of lake trout (Salvelinus namaycush).
Canadian Journal of Fisheries and Aquatic Sciences 59: 1593-1605.

Roni, P. and T.P. Quinn. 1995. Geographic variation in size and age of North American
chinook salmon. North American Journal of Fisheries Management 15: 325-345.

Stewart, D.J ., J .F. Kitchell and LB. Crowder. 1981. Forage ﬁshes and their salmonid
predators in Lake Michigan. Transactions of the American Fisheries Society 110:
751-763.

Stewart, DJ. and M. Ibarra. 1991. Predation and production by salmonine ﬁshes in Lake
Michigan, 1978-88. Canadian Journal of Fisheries and Aquatic Sciences 48: 909-
922.

Szalai, E.B., G.W. Fleischer, and J .R. Bence. 2003. Modeling time-varying growth
using a generalized von Bertalanffy model with application to bloater (Coregonus
hoyi) growth dynamics in Lake Michigan. Canadian Journal of Fisheries and
Aquatic Sciences 60(1): 55-66.

Unwin M.J. and GJ. Glova. 1997. Changes in life history parameters in a naturally
spawning population of chinook salmon (Oncorhynchus tshawytscha) associated
with releases of hatchery-reared ﬁsh. Canadian Journal of Fisheries and Aquatic
Sciences 54 (6): 1235-1245.

Withler, R. E. , W.C. Clarke, B. E. Riddell, and H. Kreiberg. 1987. Genetic variation in

freshwater survival and growth of chinook salmon (Oncorhynchus tshawytscha).
Aquaculture 64: 85-96.

121

Table 4.1 - Symbols used in the chinook salmon functional response model.

 

 

Symbol Variable description Units
Cmax Maximum consumption by a predator of given length kg
6 Annual growth in year yfor predator age a kg
GCE Gross conversion efficiency for predator age a
h Handling time for prey type j by predator age a in year y yr
L Predator length mm
N Prey abundance adjusted to mid-year value
P Per capita consumption by chinook salmon yr -1
H Ratio of prey length to predator length
8 Search rate of chinook salmon yr -1
W Mean weight of prey type j in each size category b kg
2 Size preference of an age a chinook salmon for prey type j with
length b
I“ Log-likelihood
Constants assumed as known
d Dietary preference for prey type jfor an age a predator
k Age-specific scalar for Cmax
q Intercept of power function relating length of predator age a to
Cmax
r Slope of power function relating length of predator age a to
Cmax
t Proportion of a year the predator age a is resident in the lake
u Optimum preyzpredator length ratio for prey type j
v Preference for smaller sizes of prey type j in size preference
dome curve
w Width of the size preference curve for prey type j
Estimated parameters
a Estimated search rate parameter for predator age a and prey
type!
Subscripts
a Chinook salmon ages 1 - 4
b Prey size category in 5mm increments
j Prey type (alewife or smelt)
y Year (1974-1 998)

 

122

Table 4.2 — Equations used in chinook functional response model. Descriptions of

variables are shown in Table 4.1.

 

Handling time

W3};

J

,1):
y“ Cmax

Maximum consumption (Cmax)

Cmaxa 2 ka (qa Ly 0r“)

9

Search rate

2 am LN Z.

J,y,a.b

S d j,“

j,y,a,b

Size Preference

_exp(—1.0Ry,a’b—uj)

j,y,a,b '—

 

+vj(uj—Ry,ab)

W1

Estimated growth

1
Gm = Z 2 PM “’1.le GCEa
j b

 

 

 

123

Table 4.3 — Values of assumed constants used in the functional response model and the
sensitivity analyses. “Base model” denotes the functional response model with four
estimated search rate parameters (Model 4, by prey type and predator age). Other
scenarios represent the conﬁgurations for sensitivity analyses. All assumed constants
used in the “Base model” are listed with subscript indicators and values. Sensitivity

analyses scenarios list only those constants that were changed in the scenario.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Scenario name Symbol Subscript Value(s) Value
Base model d alewife, predator age 1 0.3194
rainbow smelt, predator age 1 0.6806
alewife, predator age 2+ 0.7585
rainbow smelt, predator age 2+ 0.2415
k predator ages 1-4 1.00
t predator ages 1-3 365
predator age 4 214
q predator ages 1-3 4.93E-06
predator age 4 4.90E-06
r predator ages 1-3 2.340
predator age 4 2.244
u alewife and rainbow smelt 0.25
v alewife and rainbow smelt 0.0
w alewife and rainbow smelt 0.0183
Cmax at 120% k predator ages 1-4
1.20
Cmax at 80% k predator ages 1-4
0.80
Cmax fixed Cmax predator age 1 15.755
predator age 2 30.961
predator age 3 44.574
predator aﬁ4 27.686
Adjust to day 182 See Chapter 4 Appendix for
details
Alternative species- u alewife 0.21
specific size rainbow smelt 0.25
preference v alewife 0.25
rainbow smelt 0.6
w alewife 0.03
rainbow smelt 0.03

 

124

Table 4.4 — Likelihood ratio tests for all combinations of model conﬁgurations.

 

Model
Comparisons

df Chi-2 p-value

 

1vs.2
1vs.3
2vs.4
3vs.4

1 3.594 0.0001

43.924 <0.00001
62.332 <0.00001
32.002 <0.00001

NM—L—L

 

125

Table 4.5 — Model hypotheses and estimated search rate parameter(s) on the log scale
with asymptotic standard errors for parameter estimates shown in parentheses. The

search rate parameters (a'jﬂ) control the overall search rates for the predator on a prey

species after adjusting for predator and prey sizes. The ﬁrst subscript is the prey type-
speciﬁc scalar for alewife (i=1) and rainbow smelt (i=2). The second subscript is the
predator age grouped by age 1 (a=1) and ages 2-4 (a=2). Some models ignored one or

more of these subscripts and these are represented by dashes in place of a value for the

 

 

subscript.
Model / Hypothesis AIC Parameters
Model 1
One search rate parameter for all 297-09 01-,-= '19'04
chinook salmon ages and for both (0-072)
prey species
Model 2 _ _
(X1 -— -18.61 (12 -— -20.375
285.49 ’ '

Search rate scalar by prey type (0.101) (0.421)
“85°“? I b d 255 16 00,1: -19.925 002: -18.880
earc rate sca ar y pre ator age - (0.108) (0.063)

Model 4 _ _
(X — -20.46 (X — -19.798
Search rate scalar by predator age 227.16 1’1 0 844 2’1 0 175
and prey type ( ' ) ( ' )
0112: -20.293 (122: -17.884
(0.352) (0.101)

 

126

Table 4.6 — Results of the sensitivity analysis using alternate values for assumed

quantities (Table 3). Estimated parameter values are shown with asymptotic standard

error in parentheses. For the search rate parameters (093a), the ﬁrst subscript is the prey

type-speciﬁc scalar for alewife (i=1) and rainbow smelt (i=2) and the second subscript is

the predator age grouped by age 1 (a=l) and ages 2-4 (a=2).

 

 

a a a a
Model AIC 1,1 1,2 2,1 2,2
Cmax at 80% 153.84 -19.900 -1 9.1 59 -19.319 -17.142
(0.990) (0.201) (0.401) (0.139)
Cmax at 120% 268.95 -20.759 -20.083 -20.723 -18.203
(0.853) (0.171) (0.360) (0.094)
Cmax fixed 308.95 -21.186 -20.058 -21.172 -18.276
(1.330) (0.189) (0.487) (0.096)
Adjust to day 182 214.03 -20.171 -19.409 -20.245 -17.805
(1.238) (0.216) (0.370) (0.101)
Size preference by weight of prey 218.17 -20.382 -19.848 -20.418 -18.114
(0.788) (0.167) (0.320) (0.098)

 

127

6.E+09 {
5.E+09 {
4.E+09 5 - ' -

J I ‘ I
3.E+09 { - ;'

Abundance

2.E+09 g -

1.E+09 { .

 

 

O. E+m | r T r r r r r 1 r r F r r r r T r r I 1 r r F 1

9.E+10 -
8.E+10 -
7.E+10 -
6.E+10 -
5.E+10 -

biomass (kg)

01 4.E+10 -
3.E+10 -

 

Standin

2.E+10 '1 '

 

1.E+10 .

 

OnE+00 T 7 I T I I I r l r l r I l r 1 1 r 17 r I 1 r _r* l

1974 1978 1982 1986 1990 1994 1998
Year

Figure 4.1 —Combined alewife and rainbow smelt abundance and standing stock biomass

for the main basin of Lake Huron from 1974-1998.

128

Instantaneous Growth

Growth (kg)

into

 

 

    

 

 

1 _.
‘5 a I. ’7“, $t‘ 5 b
a“ ' ~'Y ‘ ‘~
0-5 A Y..‘_ i A I b A: b ~ ‘~&' ‘3
O n r Y r l m 1 Y 1 n
1.4 a

 

 

4.-— l .. a """""A r. —A-~~x—--k---‘- a. J: limit “"~-‘~""“"w“‘ ‘s»--—~--..~~-§--~«~--s-—-.“,----A
0 i 1 Y Y Y T Y Y Y j Y n Y f Y Y Y If Y Y 1
1974 1978 1982 1986 1990 1994 1998
Year

+ Age 1 x Age 2 + Age 3 —-m——- Age 4

Figure 4.2 - Annual age-speciﬁc chinook salmon growth (top panel) and instantaneous

growth (bottom panel) from weight-at-age data.

129

A A
35‘ A A A A A
A A
A A A
A 3. A
b A A
'5; A. A A at A A
4‘25- A I "
V 0 °6 AI I
E - . 2 :8
c: I I I
I o I
0 o I 0 0
$154 0 o 0 0 o
g 0
1..
O O. . . ‘I .
0.5‘ O. C... 0.. . O . .
O I I I I f I

 

 

2.E+09 3.E+09 4.E+09 5.E+09 6.E+09 7.E+09 8.E+09
Combined alewife and rainbow smelt abundance

<>Age1 IAgeZ AAge3 oAge4

Figure 4.3 — Relationship between observed chinook salmon growth determined from
weight-at-age data and the combined alewife and rainbow smelt abundance between 1974

and 1998.

130

Slze preference

0.9 8
0.8 ~
0.7 -
0.6 -

0.4 —
0.3 -
0.2 ~
0.1 a

 

 

 

 

// . '. \
l ".
0.1 0.2 0.3
Prey:predator ratio
Alewife adjusted

Rainbow smelt adjusted - - - - Default

Figure 4.4 - Predator size preference for the prey species. The values of the function

shape variables are given in Table 4.3. The Default curve was used for both prey species

in the model ﬁtting process. The adjusted curves were used to test the sensitivity of the

size preference function to differences in prey weight for a given size category.

131

80

1

1

70
60 a

50

l

1

40

Cmax (kg)

1

20

10

l

 

0 J T l I I 1 I I 1 1 l 1 T l l
0 200 400 600 800 1000 1200 1400

Chinook salmon length(mm)

 

—Ages 1-3 Age 4
r0 )
a

Figure 4.5 — Relationship between age-speciﬁc maximum consumption (Cmax) and

Cmaxa = ka (qa Ly

9

chinook salmon length (mm) used to determine handling time in the functional response

model.

132

J

Age1

Growth (kg)
. DJ
01 N 01 OD

—L

 

 

Growth (kg)

 

 

Growth (kg)

 

 

 

F3
.h
1

Growth (kg)

 

F’
c: to
1

1 974 1 978 1 982 1986 1 990

Year

 

o Obsen/ed — — Mode|2 Mode|4

Figure 4.6 — Observed and predicted growth with search rate parameter related only to

prey type (Model 2) or related to both prey type and predator age (Model 4)

133

3- Age1

  
  

Growth (kg)

 

 

w -L
coin-torn:
11 1‘1

1
'1
.1
.1

Growth (kg)

—L

 

 

Growth (kg)

 

 

 

  

 

 

1 - Age 4
O‘) ’— ‘ ~
:5- 0.6 - .
r
E, 0.4 ~ '
(D
0.2 -
O f j I l r I T l I I 7 l T I f 1 T T T 1’ Y I T l
1974 1978 1982 1986 1990 1994 1998
Year

Model 4

 

I Observed— — Model3

Figure 4.7 -- Observed and predicted growth with search rate parameter related only to

predator age (Model 3) or related to both prey type and predator age (Model 4).

134

 

 

30 —
A
A A
i A f A
25 ~
1 A A
A 4 A A
.9 AA A .
v ‘4 A -
,5 20 q A I At A A ‘ A
a . A Al I
E 4 I . I
3 - I I
m A
g 15 '1 . . I - . l '- I
o .
'3 - I
a ~ . . . C
33 d o h 3 ' o 9 .. 8 ° ° 8
m o 9 .o 0<> 0
l O 0
5 - 0
o . . . . , . . . . , . . . . , . T A A ,
0.E+00 2.E+09 4.E+09 6.E+09 8.E+09
PreyAbundance

<>Age1 lAgeZ AAge3 oAge4

Figure 4.8 — Relationship between the estimated age-speciﬁc consumption of prey

biomass (kg) and combined alewife and rainbow smelt abundance.

135

45 ' High growth
40 1
35 .1
30 l
25 —
20 a
15 a 5
10 a /
5 _
O.E+00 2.E+09 4.E+09 6.E+09 8.E+09

Consumption (kg)

D
---- --

   
   

 

«P----—-

 

T I T T I Y Y 1 I

35 “ Low Growth

 

 

I
U
I
I
I
n
r

 

1' # T I' 1' l V T Y I

0.E+OO 2.E+09 4.E+09 6.E+09 8.E+09
Prey Abundance

-o—Age1+AgeZ +Age3-o—Age4

Figure 4.9 — Estimates of consumption from Model 4 using incremental prey abundance
and two levels of ﬁxed predator size representing high (1974) and low (1984) growth
periods. The vertical dashed line represents the lowest observed prey abundance between

1974-1998.

136

80*

 

30 a

20*

Consumption (kg) by cohort

10*

 

 

0 I I I I I I I
0.6 0.7 0.8 0.9 1 1.1 1.2 1.3

Weight at age 1 (kg)

Figure 4.10 - Consumption by a cohort and weight at age 1. Consumption = 39.76 W +

21.49 with R2 = 0.7696, where W is weight-at-age 1 (kg). Consumption is shown for 21

full cohorts over 1974—1998 (1995 was last cohort).

137

 

 

 

 

 

 

 

 

 

14} Age1 251 A992
12‘
3‘0 a?
C C
03 o
a a
56 \ e
a 3 10-
O I C! q
2 g .
34 o :
5..
2 Z
oﬁ.ﬁ,.ﬁ.r.ﬁﬁ,....r.... 0 ﬁg.
1974 1 979 1 984 1 989 1 994 1 974 1 979 1984 1989 1 994
Year Year
351 Age3 121 A994
30
A -" . A
3:25 MW 3
c . c
.020: . , O
E. : \ '3
3153 S :
O -‘ 0
c 3 c 41
81°: 8 2
5i 2?
oTﬁ Oﬂ
1974 1979 1984 1989 1994

1974 1979 1984 1989 1 994

Year Year

 

. Bioenergetics model
+Functiona| response model

 

 

 

Figure 4.11— Comparison of age-speciﬁc consumption from the bioenergetics models

(Chapter 3) and the functional response model.

138

(x1 ,1 a] ,2 “2,1 0‘2,2

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

(alewife,age 1) (alewife,age 2+) (smelt,agei) (smelt,age 2+)
1.000 7 r
0.800 -
m "2""
3 — .
ID a ‘ ;
,§ 0.600 5 .
a v a
2 0.400 — ,—
2 :f
g 0.200 -<
is
G
V 0.000 . ,. m . .
'as
‘3
2 -0.200 1
E _ _ . __
e :
- -0.400 — _ —
C
.2
‘5
-- -0.600 -
B
a _
-0.800 -
-1.000 4

 

Figure 4.12— Sensitivity of estimated search rate parameters to ﬁxing Cmax (ﬁrst bar),
increasing Cmax by 20% (second bar), decreasing Cmax by 20% (third bar), modifying
size preference for weight of prey ﬁsh (fourth bar), and adjusting values to actual mid-

year, day 182 (ﬁfth bar). Each grouping represents one search rate parameter (aid) as

deﬁned in Table 4.5.

139

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.90 ~— 1.00 -—
0.80 ~~ I J 090 r 1 L
0.70 -. l «mo-—
060“ i 0"“ ' l r l l
g 0.50 P l x 0.60 l
E 0. 40 4_ r I E 0.50 i l
‘L ‘L 0.40 1’
0.30 ~— 0.30 i—
0-20 " 0.20 -I
0.10 - 0.10 --
0-00 ‘I i i i d 0.00 I I I I H
Cmax Cmax Cmax Size Day182 Cmax Cmax Cmax Size Day 182
80% 120% ﬁxed Pref. 300/. 120% ﬁxed Pref,
e 3 e 4
0-901' A9 0.60 -— A9
0.80 ~~
0.70 + l | L l 0.50 -~ I J ,
0.0. | I i | . l i |
z 0.50 I , — g r I
08. 0.40 T I 5030 T I
0-30 ‘" 0.20 —~
0.20 -~
0.10 -— 0'10 i
0.00 I I I I I 0.00 I I I I I
Cmax Cmax Cmax Size Day 182 Cmax Cmax Cmax Size Day 182
80% 120% ﬁxed Pref. 80% 120% fixed Pref.

Figure 4.13— Results of the sensitivity analysis showing the minimum and maximum
values of Pmax for each age. The lower level of each bar represents the minimum Pmax
while the upper point represents the maximum value for each sensitivity analysis. The
dashed lines represent the minimum and maximum values of Pmax estimated from Model

4 using the base values of all assumed constants (Table 4.3).

140

 

 

1.0“
‘55 . I ~
i3"""0.8— I I.
5.."L - I I
tg ﬁance..ng 6 I
amoea 0 M 6-
05 6 ‘
3.0 g a o o
3§0.4~ °' '40:. ' 0.
ﬁg . ’
§§021
me .

0.0 TIT'fr‘ITTTFTYfrII[II"Tﬁ

0.0E+0 5.0E+05 1.0E+06 1.5E+06 2.0E+06 2.5E+06
Chinook salmon abundance

 

 

E 1.0

3

.§

x I

a 0.8 ' ' I

E c .I -~ .3! 6 6 .‘

o I

l: '3 0.6 2? ﬁ% 6

-2 e e

t O o o.

§§ 0.4 o :0 "0 ”“0

a 3 l '

E 0.2—

{5

.§

3 0.0 v r r v r v r w v i r . r 1 r
0.E+00 3.E+09 5.E+09 8.E+09

Prey abundance
o Age1 I Age 2 A Age 3 0 Age 4

Figure 4.14 — Estimated proportion of maximum consumption (Pmax) related to chinook
salmon abundance (top panel) and combined alewife and rainbow smelt abundance

(bottom panel).

141

0.3

 

 

0.25 , o
E ‘ O O O .
0 ' o ' ' o "'—'*
E 02— ’
t .
0
C
.2 as H '55 173 II: at
9 0'15 ‘13”fi7‘" .X..-..I...II;I- “3&0 .Mfax
g )rkiLﬁirLLLLLEL&X_—§_r;2{__>£.Lx_.
o
0 -
m 0.1
m
2
(5

0.05~

0 i l i l T l l l i l l
1974 1978 1982 1986 1990 1994 1998
Year
0 Age1 III Age2 x Age3 Age4

Figure 4-15 — Annual estimated gross conversion efﬁciency (calculated from

bioenergetics models, Chapter 3 and Appendix B) for age 1-4 chinook salmon in the

main basin of Lake Huron.

142

APPENDICES

143

Appendix A

Descriptive data for ﬁsh samples used in energy density analysis

This appendix describes the ﬁsh samples collected for use in the energy density
analysis (Chapter 2) and presents descriptive statistics for these data. There were 707
ﬁsh collected in Lake Huron from June 11, 1996 to September 24, 1997. Various
agencies (Michigan Department of Natural Resources, Chippewa/Ottawa Treaty Fishery
Management Authority, Ontario Ministry of Natural Resources, and Biological Research
Division-USGS) collected the ﬁsh throughout the year. Fish were measured for total
length and weighed in the ﬁeld, if possible. Individual whole ﬁsh were placed in plastic
bags without water and frozen immediately or kept on ice until a freezer was available.
Identiﬁcation tags were placed with each ﬁsh to indicate the collector, site, time of day,
date, length, and weight. We targeted for ﬁve ﬁsh in each size interval for each statistical
district and month. Size intervals for predators were 100 mm (>100-200, >200-300, etc.)
and 20 mm (>10-30, >30-50, etc.) for forage ﬁsh. An alternative procedure was
sometimes used for forage ﬁsh since their small size did not always allow for accurate
measurement of weight. Groups of small forage ﬁsh of the same species and from the
same collection site were either sorted by size interval into separate bags or grouped
together if the collector did not have time to sort by size class. An identiﬁcation tag was
placed in the bag with the same information outlined above. When possible, water was
added to each bag so that ﬁsh were frozen in ice to minimize weight loss. Bags were

frozen or placed on ice until a freezer was available.

144

To analyze regional differences, statistical districts were grouped into lake regions
(Table A. 1). Abbreviated species names (Table A.2) are used in this document when

space on a table or ﬁgure was limited.

Samples of all key predator and prey species were obtained (see species list in
Table A.2). However, there was incomplete coverage of months and statistical districts,
resulting in missing months and regions for many species. Also, only a subsample of ﬁsh
was processed in the bomb calorimeter to determine energy content. A linear regression
of percent water on energy density was modeled from these samples and used to estimate
the energy density of the remaining samples (see Chapter 2). Sample characteristics by
month, statistical district, and gender are shown in Tables A.3 — A5 for all samples and
for those processed in the calorimeter. Table A6 contains descriptive statistics (mean

and standard deviation) for several key variables.

Mean energy density and mean percent water content of all samples by species
are shown in Figure A. 1. Only lake trout, chinook salmon, and burbot samples were
obtained in all three regions of Lake Huron. Lake trout are known to reside in localized
regional populations (Eshenroder et al. 1995) while burbot and chinook salmon may not.
Regional differences in energy density were only found in the lake trout populations

(Chapter 2).

Literature Cited

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

145

Table A.1 — Number of samples from different statistical districts and lake regions. For

the regional analysis of energy density, statistical districts were grouped to represent a

particular lake region. The grouping of these statistical districts coincides with the

regional lake trout populations.

 

 

By Statistical District By Region
Region Statistical Frequency Percent Frequency Percent
District

North MH1 145 20.51 145 20.51
Central MH2 329 46.53 329 46.53
South MH3 25 3.54 233 32.96
South MH4 62 8.77

South MH5 24 3.39

South 0H3 69 9.76

South 0H4 43 6.08

South 0H5 10 1.41

 

Table A.2 — Species name abbreviations. These 3-letter codes are used in some tables

and ﬁgures when the full species name did not ﬁt into a table or ﬁgure.

 

 

Species Type Abbreviation
alewife prey ALE
bloater prey BLO
burbot predator BUR
chinook salmon predator CHS
lake trout predator LAT
rainbow smelt prey SME
sculpin prey SCU
stickleback prey STB
walleye predator WAE

 

I46

Table A3 -- Fish sample characteristics by month. Number of samples, N, are given as

the total number of samples (top) and the number of samples processed in the bomb

calorimeter (bottom, in parentheses).

 

Chinook Lake Rainbow

Stickle

 

 

Month N Alewife Bloater Burbot salmon trout smelt Sculpin back Walleye
Jan 22 9 4 9
(13) (6) (1) (6)
Mar 15 3 10 2
(3) (3)
Apr 1 1
(1) (1)
May 1 1 1 22 23 18 39 9
(52) (16) (14) (14) (6) (2)
Jun 217 72 2 44 15 53 31
(45) (12) (11) (10) (9) (3)
Jul 139 73 10 27 25 1 3
(34) (10) (7) (5) (9) (3)
Aug 1 1 1 35 3 40 32 1
(17) (4) (8) (5)
Sep 39 1 9 29
(18) (7) (11)
Oct 52 6 4 27 15
(20) (3) (3) (14)
Total 707 181 36 86 96 153 106 1 3 45
(203) (26) (25) (25) (49) (25) (25) (3) (25)

 

147

Table A4 - Fish sample characteristics by statistical district. Number of samples, N, are
given as the total number of samples (top) and the number of samples processed in the

bomb calorimeter (bottom, in parentheses).

 

Statistical Chinook Lake Rainbow Stickle
D'stﬂct N Alewife Bloater Burbot salmon trout smelt Sculpin back Walleye

 

MH1 145 34 17 25 46 22 1
(64) (25) (22) (9) (8)
MH2 329 161 21 44 50 49 3 1
(62) (21) (1 1) (8) (9) (10) (3)
MH3 25 1 4 15 3 2
(7) (1) (1) (1) (2) (2)
MH4 62 8 3 9 42
(33) (4) (1) (5) (23)
MH5 24 5 1 5 13
(9) (4) (1) (3) (1)
0H3 69 14 2 26 1 3 23
(1) (1)
0H4 43 8 9 26
(20) (6) (9) (5)
0H5 10 5 5
(7) (3) (4)
Total 707 181 36 86 96 153 106 1 3 45
(26) (25) (25) (49) (25) (25) (3) (25)

 

Table A5 -- Fish sample characteristics by gender. Number of samples, N, are given as
the total number of samples (top) and the number of samples processed in the bomb

calorimeter (bottom, in parentheses).

 

Chinook Lake Rainbow Stickle
Gender N Alewife Bloater Burbot salmon trout smelt Sculpin back Walleye

 

F 133 0 0 35 21 56 0 0 0 21
(14) (17) (12) (14)

M 171 0 0 47 38 66 0 0 0 20
(11) (26) (8) (9)

U 403 181 36 4 37 31 106 1 3 4
(26) (25) (0) (6) (5) (25) (0) (3) (2)

Total 707 181 36 86 96 153 106 1 3 45

 

148

Table A6 — Descriptive statistics by lake region and month for all ﬁsh samples collected.

Means are shown with standard deviations in parentheses.

 

 

 

 

Energy
Weight Length Percent density
Species Region Month N (kg) (cm) water J/g wet wt.
ALE Central 6 57 0.017 12.215 82.5 3,755.4
(0.02) (4.03) (3.06) (843.1 )
7 73 0.013 11.127 81.8 3,935.2
(0.01) (3.58) (3.51) (966.4)
8 30 0.017 12.432 77.9 5,009.5
(0.01) (3.21) (4.27) (1,174.6)
9 1 0.010 10.600 78.8 4,764.3
South 6 15 0.032 15.939 77.7 5,055.1
(0.01) (2.37) (2.80) (770.9)
8 5 0.029 15.420 77.4 5,139.3
(0.01) (1.41) (1.61) (443.4)
BLO North 1 9 0.139 25.025 78.1 4,965.5
(0.06) (2.80) (5.70) (1 .5673)
3 3 0.127 22.367 77.8 5,043.4
(0.06) (4.76) (3.02) (830.7)
5 22 0.142 26.167 74.5 5,948.8
(0.06) (3.16) (4.61) (1,269.2)
South 6 2 0.325 25.650 81 .9 3,900.2
(0.04) (1.06) (0.41) (113.8)
BUR Central 6 19 1.518 54.653 75.7 5,842.3
(0.72) (8.82) (3.76) (1 .3023)
8 2 0.830 43.600 76.0 5,760.8
(0.95) (17.54) (0.07) (24.3)
North 3 10 1.521 51.790 75.8 5,823.9
(0.29) (5.98) (2.79) (965.2)
5 1 0.249 32.800 82.2 3,587.5
10 6 0.962 47.550 75.8 5,827.5
(0.44) (4.34) (1 .55) (538.5)
South 5 22 2.002 59.414 76.1 5,709.9
(1.22) (13.74) (3.82) (1,323.1)
6 25 0.926 44.540 77.1 5,374.3
(0.36) (9.60) (1 .87) (649.3)
8 1 1.300 51.000 78.5 4,884.7

 

149

Table A.6 continued

 

 

 

Energy
Weight Length Percent density
Species Region Month N (kg) (cm) water NM
CHS Central 6 7 1 .702 39.386 75.7 5,393.3
(2.21) (23.41) (7.45) (2,583.0)
7 5 1.454 33.560 77.5 4,748.0
(2.61) (27.83) (4.92) (1 ,704.3)
8 31 0.790 31.244 76.8 4,989.6
(1.63) (17.94) (1.81) (626.3)
10 1 1.700 54.600 75.1 5,592.4
North 5 1 0.130 24.100 79.2 4,153.8
6 2 4.830 58.100 72.3 6,543.1
(5.69) (33.80) (1 .97) (682.2)
7 5 3.112 56.060 73.1 6,292.4
(3.81) (29.21) (3.72) (1,287.9)
8 9 1.230 47.911 75.4 5,472.6
(0.60) (7.45) (1 .84) (637.9)
9 8 2.540 60.105 74.6 5,762.3
(1.44) (13.48) (2.66) (920.3)
South 5 17 2.051 53.324 71.6 6,795.8
(1.86) (17.27) (4.70) (1,627.3)
6 6 1 .046 29.483 78.0 4,589.7
(1.74) (28.03) (8.17) (2,830.2)
9 1 1.600 46.100 73.0 6,328.0
10 3 1.617 47.400 72.0 6,662.5
(1 .23) (21 .81) (3.89) (1 ,347.7)
LAT Central 6 50 1.414 44.462 68.7 8,282.1
(1.14) (13.99) (5.38) (1,862.6)
North 1 4 0.945 47.875 74.0 6,438.8
(0.25) (5.87) (1 .36) (472.7)
3 2 0.780 40.600 73.9 6,480.2
(0.45) (7.64) (5.20) (1 ,800.6)
4 1 3.500 70.100 74.0 6,450.4
7 27 0.842 39.983 70.7 7,581.2
(0.47) (7.83) (4.68) (1 ,621.5)
10 12 1.539 49.583 71.9 7,156.8
(1.11) (12.31) (5.07) (1,755.4)
South 5 39 2.098 56.667 68.0 8,525.6
(1.25) (14.18) (3.16) (1,094.3)
6 3 0.787 39.767 71 .8 7,204.5
(0.41) (5.13) (1.19) (412.3)
10 15 3.009 64.093 63.9 9,937.5
(0.95) (7.14) (3.23) (1,120.5)

 

150

Table A.6 continued

 

 

 

 

 

Energy
Weight Length Percent density
Species Region Month N (kg) (cm) water J/g wet wt.
SCU North 7 1 0.001 5.000 79.3 4,635.5
SME Central 6 5 0.013 14.330 74.8 5,870.5
(0.01) (1.67) (4.19) (1,151.7)
7 22 0.005 8.186 80.5 4,285.8
(0.01) (3.27) (6.95) (1,911.4)
8 22 0.004 8.340 77.7 5,055.2
(<0.001) (2.51) (2.66) (730.6)
North 1 9 0.022 13.961 79.3 4,626.0
(0.01) (1.56) (1.89) (520.3)
5 9 0.012 12.356 75.0 5,811.3
(0.01) (1.85) (2.58) (709.1)
7 3 0.008 10.533 82.1 3,840.2
(0.01) (4.69) (4.44) (1,221.9)
8 1 0.032 17.600 78.6 4,820.4
South 6 26 0.097 20.898 74.8 5,855.2
(0.05) (5.52) (5.58) (1 ,533.4)
8 9 0.008 1 1.763 76.5 5,404.3
(<0.001) (1 .80) (2.89) (794.0)
STB Central 7 3 0.002 6.867 77.2 5,194.2
(<0.001) (0.65) (5.47) (1 ,505.8)
WAE Central 8 1 2.000 79.000 76.4 4,820.7
South 9 29 1.441 49.903 71.4 6,571.6
(0.83) (8.64) (2.21 ) (766.7)
10 15 1.333 49.239 72.2 6,279.4
(0.54) (6.56) (1 .92) (666.1)

 

151

 

11000-
10000:
9000.
8000:
7000:
6000:

SOOOj

meenergydensityUg:

4ooo~

3°°°i

 

A. Mean energy density
I

 

 

 

 

 

 

 

 

 

 

 

l 1 --   l l

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2000

f r

ALE BLO BUR CHS LAT SCU SME STB WAE

 

 

 

85f

 

80- J-
75‘

70~

thpaoa'tvaa

65..

60..

55~

 

 

 

90 ~ 8. Mean percent water

 

 

 

 

 

I-.-—
l'—'-—i
I-———-l

 

——1
F—1——l
,1.—

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

50

 

ALE

I I I T I TI

BLO BUR CHS LAT SCU SME S'IB WAE

 

 

Species %CV
ALE 25.665
BLO 25.207
BUR 1 8.803
CHS 27.429
LAT 20.706
SCU 0.000
SME 27.499
STB 28.991
WAE 1 1.987

Species %CV
ALE 4.829
BLO 6.641
BUR 4.003
CHS 5.874
LAT 7.099
SCU 0.000
SME 6.653
STB 7.087
WAE 3.103

 

 

 

Figure A.1 — Mean energy density (A) and mean percent water (B) of each species

collected with error bars representing one standard deviation. The percent coefﬁcient of

variation for each graphed variable is shown in the table on the right. Note, graphs use

non-zero origin. Abbreviated species names are shown in Table A.2.

152

Appendix B

Parameter Values Used in Bioenergetics Models

This appendix describes the bioenergetics parameters 1 used in my
implementation of the Wisconsin model (Hewett and Johnson 1995 Ver 3.0b) for each
predator population in Lake Huron. These models included one for burbot in the main
basin; one for chinook salmon in the main basin; three for lake trout corresponding to the
northern, central, and southern regions of the main basin; and two for walleye,
corresponding to Saginaw Bay and the region of the main basin south of Saginaw Bay.
Some model values were obtained from the energy density analysis of Lake Huron
species outlined in Chapter 2. Other values, such as water temperature and weight-at-
age, were derived from published data. Physiological parameters supplied with the
distributed version of the Wisconsin model were changed as needed to accommodate

individual predator populations (Table B.7).

Simulation length

All bioenergetics models were run for 365 days. The ﬁrst simulation day for
burbot and lake trout was January 1St and July 1St respectively. Chinook salmon and
walleye were each modeled in two time periods (see Growth section). For chinook
salmon, the ﬁrst day of simulation was January ISI for the pre-harvest period followed by
a post-maturation period commencing on day 214. Age 4 chinook salmon were assumed

to spawn and die on day 214 (August 2‘“). Simulation of the walleye growth period

153

began on May 1St proceeded by the maintenance period beginning on day 153 (October

18!).

Actual and preferred water temperature

The seven bioenergetics models cover different portions of the main basin
including: the entire main basin; the northern, central, and southern regions of the main
basin; and Saginaw Bay. Water temperature information for each of these areas (Table
B.1) was obtained from NOAA/GLERL reports (Grumblatt 1976; McCormick 1996;
Nalepa et a1. 1996; J ohengen et a1. 2000). In Saginaw Bay, inner bay data from 1994-
1996 was used except for missing months J anuary-March and November-December,
which were estimated from 1993 Bay City data.

For bioenergetics models, the water temperature experienced by a predator was
the actual water temperature unless it exceeded the preferred temperature (Table B.2). It
was assumed that predators would reside in their preferred water temperature or in lower
temperatures when the preferred temperature was not available. The preferred
temperature of age 0 chinook salmon was 180C, while age 1+ chinook salmon and lake
trout preferred 110C water (Stewart and Ibarra 1991). Burbot ages 1-3 preferred 120C
water while ages 4+ preferred 100C water (Rudstam et al. 1995). Preferred temperature

for all walleye age classes was set to 220C (Kitchell et al. 1977).

Diet composition
Diet composition for each predator population was estimated from data provided

by the Biological Research Division -- US Geological Survey; Chippewa/Ottawa Treaty

154

Fishery Management Authority; Michigan Department of Natural Resources; and Ontario
Ministry of Natural Resources. Predator ages were grouped into age classes. Mean prey
weights for each age class were estimated by summing all prey weights and dividing by
the total number of prey samples weighed. Where data were absent, mean prey weights
were set equal to adjacent age classes. Prey counts in each predator age class were
multiplied by the mean prey weight resulting in an estimate of prey biomass consumed.
The proportion of each prey item in the diet was determined by dividing prey biomass by
total biomass of each predator age-class (Table B.3). When sufﬁcient data were
available, the proportion of each prey species consumed by weight in each year (1991-
1999) was estimated. The mean across years became the proportion of each prey in the
diet. In some instances, prey item counts and weights were pooled over the data time
periods to provide a large enough sample size. With only three signiﬁcant digits used to
deﬁne the diet composition in the bioenergetics models, some rounding corrections were

needed to adjust the values to sum to 1.0.

Prey Energy Density

In the Wisconsin model (Hewett and Johnson 1995 Ver 3.0b), energy density
must be provided for each prey item identiﬁed in the diet composition. Prey energy
density may be constant or vary by day. In Chapter 2, the energy density was estimated
for the majority of prey items found in the diets of the key predators. Two diet items,
invertebrates and “other ﬁsh”, were not estimated with bomb calorimetry. For
invertebrates, the mean energy density used in previous studies of Lake Michigan

(Stewart et al. 1983; Rudstam et al. 1995) was used (Table B4). In the diet composition,

155

“other ﬁsh” represent species not normally found in the open waters of Lake Huron or
immature individuals of predator species. Here, the mean energy density for “other ﬁsh”
used in previous studies (Cummins and Weychuck 1971 as used in Stewart et al. 1983;
Stewart and Binkowski 1986) was used (Table B.4).

For the predominant prey species, regional and seasonal energy densities were
determined from Lake Huron samples (Chapter 2). Some adjustments were made to
these values to accommodate each predator population model. For bloater, two samples
came from the southern region of the lake while all others came from the north. Data
were insufﬁcient to identify regional or seasonal differences in energy density, so the
overall mean was used in bioenergetics models (Table B.4). For stickleback, the average
of the three processed samples was used (Table B.4). Only one sculpin was analyzed
from Lake Huron with an estimated energy density of 4636 J og-l wet weight. Because of
this limited sample, the value used in the bioenergetics models (Table B4) was an
average of this value and those published by Cummins and Weychuck (1971, as used in
LeBar 1993), and Rottiers and Tucker (1982).

Alewife energy density was found to vary by region and by month (Chapter 2).
However, no samples were available from the northern region and there were insufﬁcient
samples from the central and southern regions to determine seasonal trends for each
region. Differences in energy density between these two regions were minimal (Figure
B. l-A). Therefore, regional differences in energy density are ignored in our
bioenergetics models.

While alewife energy density was found to vary by month, samples were

collected only during the months of June through September. To determine energy

156

density for the missing months, energy density values from Hurley (1986), energy density
averaged over size—classes from Rand et al. (1994), and energy density averaged over
gender from Flath and Diana (1985) were used. First, monthly mean energy density for
both the Lake Huron estimates and the published values were obtained. The Lake Huron
means were consistently lower than the means of the published values (Figure B.l-B).
The ratio of the Lake Huron energy density to the published values for the months of
June, July, and August was found to be 0.791. This ratio was then applied to the mean
monthly energy density of the published data to obtain estimates of alewife energy
density for the missing months (Table B4).

The energy density of rainbow smelt was found to vary seasonally but samples
were available only from January and May through August (Figure B.2). Mean energy
density in the month of July was lower than all other months sampled, but variability was
high. Other studies have found that rainbow smelt energy density increases from May
through October (Foltz and Norden 1977; Rand et al. 1994; Vondracek et al. 1996).
Because energy densities of the Lake Huron samples were unusually low in July, these
samples were removed from the following analysis of seasonal energy density patterns.
The seasonal pattern of energy density in rainbow smelt was estimated in a similar
fashion as described for alewife. Published energy density from Vondracek et al. (1996)
and values averaged over size-classes from Rand et al. (1994) were averaged to obtain
monthly values. In some cases, a gap of one month in these data was estimated by
interpolation. Lake Huron rainbow smelt energy density was consistently higher than the

published values, with a mean proportional difference of 1.148 (Figure B.2-B). This

157

proportion was applied to the literature means to get an adjusted estimate of rainbow

smelt energy density for the missing months (Table B.4).

Predator Growth

Fish growth is represented by the change in weight from one time period to
another. For lake trout and burbot, weight-at-age data provided starting and ending
weights. The starting weight was approximated by the weight-at-age while the ending
weight was set to the weight-at-age for the next older age. Wei ght-at-age estimates for
burbot were obtained by ﬁtting mean weights for ages 3 through 17 (McLeish et al., In
preparation) to a von Bertalanffy curve. For all lake trout models, the mean weight-at-
age was estimated from spring gill net surveys conducted by the Michigan Department of
Natural Resources.

Walleye growth was divided into two periods - a growth period occurring
between May and October and a maintenance period from November to April. During
the maintenance period, weight was maintained at the October level except for gonadal
deve10pment (Hurley 1986). Weight gain during this time was 12% of body mass, which
was then lost during spawning. For the growth period, weight-at-age was estimated from
1985-1995 Lake Huron creel data and was used to identify starting and ending weights.
The ending weight for the last age in the growth period was estimated as the same
proportional increase experienced in the prior age. The ending weight for the growth
period became the starting weight for the gonadal development period. The ending
weight for this period was estimated as the starting weight plus the weight lost to

spawning (Table B5).

158

Chinook salmon growth was partitioned into pre-harvest and post-maturation time
intervals. Weir return weight-at-age was available for two time periods: 1973-1981 and
1985-1999 (McLeish et al., In preparation) and harvest weight-at-age from 1985-1998.
For the post-maturation interval, the mean weight-at—age was computed for each time
period from the data (Bence and Dobiesz 2000). Back—calculating these weights to
annulus formation produced weight-at-age estimates for the pre-harvest interval (Table
B.5). For age 0 chinook salmon weight-at-annulus formation was assumed to be 4.54 g

(Stewart and Ibarra 1991).

Predator spawning losses

Reproductive tissues are generated during the normal growth period and lost
during spawning. In the Wisconsin bioenergetics model, a proportion of ﬁsh biomass is
lost on a pre-deﬁned spawning day. The models do not differentiate between male and
female predators; therefore as recommended (Hewett and Johnson 1995), the gonadal
tissue lost by males and females is averaged to produce the percent biomass lost during
spawning. In all three lake trout models, an individual matured at age 6, losing 6.8% of
their biomass on simulation day 118 (Stewart et al. 1983). Burbot began spawning at age
3, losing 11% of their biomass on simulation day 32 (Rudstam et al. 1995). Walleye
matured at age 3, with an average loss of 12% of their body mass (Hurley 1986),
occurring on day 365 between simulated periods of growth and gonadal development.
Chinook salmon are semelparous and die after spawning. Adults matured at age 4,

spawning on simulation day 214, when they were dropped from the model.

159

Predator Energy Density

The Wisconsin Model (Hewett and Johnson 1995 Ver 3.0b) uses a linear
relationship to track changes in energy density as a ﬁsh grows. Two different
relationships can be applied, one above and one below a speciﬁed weight threshold. To
identify these relationships, predator weight was plotted against energy density to identify
mass cutoff values. A single linear relationship was tested as the simplest model. The
extra sums of squares test (Neter et al. 1996) was used to evaluate this reduced model
against a model that included separate intercepts and slopes above and below a weight
threshold, speciﬁc to each predator population. For burbot and walleye, no relationship
between energy density and weight was evident; therefore, the overall mean energy
density was used (Table B.6). For lake trout and chinook salmon, the relationship
between energy density and weight was better estimated by two linear relationships.
Each population had a unique mass cutoff, deﬁned by the intersection of the two lines
(Figure 83). Values of the intercepts, slopes, and mass cutoffs were used as parameters
(a1, bl, 32, b2, and mass cutoff) in the predator energy density equation in the Wisconsin
model (Table B.7). Based on the results of Chapter 2, predator energy density was

treated as not varying seasonally.

160

Literature Cited

Bence J. R. and N. E. Dobiesz. 2000. Estimating forage ﬁsh consumption by predators
in Lake Huron. Great Lakes Fishery Commission Project Completion Report.
Available for download at http://www. g1fc.org/research/cap.htm .

Cummins, K. W. and J. C. Wuycheck. 1971. Caloric equivalents for investigations in
ecological energetics. Mitt. Int. Ver. Limnol. No. 18.

Flath, L. E. and J. S. Diana. 1985. Seasonal energy dynamics of the alewife in
southeastern Lake Michigan. Transactions of the American Fisheries Society
114: 328-337.

Foltz, J. W. and C. R. Norden. 1977. Seasonal changes in food consumption and energy
content of smelt in Lake Michigan. Transactions of the American Fisheries
Society 106(3): 230-234

Hewett. S. W., and B. L. Johnson. 1995. Fish Bioenergetics Model 3. University of
Wisconsin Sea Grant Institute, WIS-SG-9l-250.

Hurley, D. A. 1986. Growth, diet, and food consumption of walleye: an application of
bioenergetics modeling to the Bay of Quinte, Lake Ontario, population. In C. K.
Minns, D. A. Hurley, and K. H. N icholls [eds.] Project Quinte: point-source
phosphorus control and ecosystem response in the Bay of Quinte, Lake Ontario.
Canadian Special Publication of Fisheries and Aquatic Sciences 86: 0706-6481.

Grumblatt, J. L. 1976. Great Lakes water temperatures, 1966-1975. NOAA Technical
Memorandum ERL—GLERL- 1 1-1 .

Johengen, T. H., T. F. N alepa, G. A. land, D. L. Fanslow, H. A. Vanderploeg, and M. A.
Agy. 2000. Physical and chemical variables of Saginaw Bay, Lake Huron in
1994-1996. NOAA Technical Memorandum GLERL-l 15.

Labar G. W. 1993. Use of bioenergetics models to predict the effect of increased lake

trout predation on rainbow smelt following sea lamprey control. Transactions of
the American Fisheries Society 122 (5): 942-950.

Kitchell, J. F., D. J. Stewart, and D. Weininger. 1977. Applications of a bioenergetics
model to yellow perch (Percaﬂavescens) and walleye (Stizostedion vitreum

vitreum). Journal of the Fisheries Research Board of Canada 34: 1922—1935.

McCormick, M. J. 1996. Lake Huron water temperature data Bay City, Michigan 1946-
1993. NOAA Technical Memorandum GLERL-93.

161

McLeish, D. A et al. In preparation. SCOL H Lake Huron Case History. Great Lakes
Fishery Commission Report.

Nalepa T. F., G. L. Fahnenstiel, M. J. McCormick, T. H. Johengen, G. A. Lang, J. F.
Cavaletto, G. Goudy. 1996. Physical and chemical variables of Saginaw Bay,
Lake Huron in 1991-1993. NOAA Technical Memorandum GLERL-91.

Neter, J ., M. H. Kutner, C. J. Nachtsheim, and W. Wasserman. 1996. Applied Linear
Regression Models, 3rd edition. Irwin: Chicago, p 260-267.

Rand P. S., B. F. Lantry, R. O’Gorman, R. W. Owens, D. J. Stewart. 1994. Energy
density and size of pelagic prey ﬁshes in Lake Ontario, 1978-1990: implications

for salmonine energetics. Transactions of the American Fisheries Society 123(4):
519-534.

Rottiers, D. V. and R. M. Tucker. 1982. Proximate composition and caloric content of
eight Lake Michigan ﬁshes. US. Fish and Wildlife Service Technical Paper 108.

Rudstram, L. G., P. E. Peppard, T. W. Fratt, R. E. Bruesewitz, D. W. Coble, F. A. Copes
and J. F. Kitchell. 1995. Prey consumption by the burbot (Lota Iota) population

in Green Bay, Lake Michigan, based on a bioenergetics model. Canadian Journal
of Fisheries and Aquatic Sciences. 52: 1074-1082.

Stewart D. J ., and F. P. Binkowski. 1986. Dynamics of consumption and food
conversion by lake-michigan alewives - an energetics-modeling synthesis.
Transactions of the American Fisheries Society 115 (5): 643-661.

Stewart, D. J ., D. Weininger, D. V. Rottiers, and T. A. Edsall. 1983. An energetics
model for lake trout, Salvelinus namaycush: application to the Lake Michigan
population. Canadian Journal of Fisheries and Aquatic Sciences. 40: 681-698.

Stewart, D. J. and M. Ibarra. 1991. Predation and production by salmonine ﬁshes in
Lake Michigan, 1978-88. Canadian Journal of Fisheries and Aquatic Sciences.
48: 909-922.

Vondracek, B., B. D. Giese, and M. G. Henry. 1996. Energy density of three ﬁshes from

Minnesota waters of Lake Superior. Journal of Great Lakes Research. 22:757-
764.

162

Table B.1 — Estimated Lake Huron water temperatures on the frist day of each month,
based on NOAA/GLERL reports (Grumblatt 1976; McCormick 1996; Nalepa et al. 1996;

Johengen et a1. 2000).

 

Month Lakewide North Central South Saginaw Bay

 

Jan 1 1 1 1 3
Feb 1 O 0 2 3
Mar 1 0 1 3 4
Apr 4 1 3 6 7
May 8 7 8 9 1 1
Jun 11 12 11 11 19
Jul 19 19 19 20 22
Aug 20 19 20 22 23
Sep 15 14 15 16 19
Oct 12 1O 11 14 12
Nov 8 8 8 8 6

Dec 3 3 2 2 4

 

163

Table B.2 — Water temperatures on the ﬁrst day of each month as experienced by
predators in Lake Huron during bioenergetics modeling. Estimated water temperatures
are used (Table B. 1) except when the preferred water temperature is exceeded. It was
assumed that predators would reside in their preferred water temperature or in lower
temperatures when the preferred temperature is not available. Shaded cells represent

preferred water temperatures.

 

 

Lake Lake Lake Walleye
Trout Trout Trout Walleye (Saginaw
Burbot Chinook salmon North Central South South Ba
Date Age 1-3 Age 4+ Age 0 Age 1+ Age 1+ Age 1+ Age 1+ Age 2+ Age 2+
Jan 1 1 1 1 1 1 1 1 3
Feb 1 1 1 1 1 0 2 2 3
Mar 1 1 1 1 1 1 3 3 4
Apr 4 4 4 4 1 3 6 6 7
May 8 8 8 8 7 8 9 9 1 1
Jun 11 1O 11 11 ' * 11 . (~11 11 11 19
Jul 12 10 18 11 , 11 . 311 , 11 2O 22
Aug 12 1O 18 11 11 11 11 22 22
Sep 12 10 15 11 11 11 >11 16 19
Oct 12 10 12 11 10 11 ’ 11 14 12
Nov 8 8 8 8 8 8 8 8 6
Dec 3 3 3 3 3 2 2 2 4

 

Table B.3 — Diet composition of Lake Huron predators by age class. Values represent the

proportion by weight of each prey item in the diet.

 

 

Prey Species
Age Rainbow Other
Class Alewife Bloater Invertebrat Sculpin Smelt Sticklebac fish
e k

Bu rbot 1-3 0.156 0.000 0.474 0.227 0.136 0.003 0.004

4-7 0.262 0.027 0.330 0.158 0.214 0.004 0.004

8+ 0.264 0.029 0.115 0.087 0.476 0.003 0.026
Chinook 0 0.1 1 1 0.000 0.000 0.002 0.479 0.001 0.408
Salmon 1 0.298 0.000 0.000 0.004 0.634 0.010 0.053

2+ 0.732 0.000 0.000 0.000 0.233 0.029 0.006
Lake trout 1-3 0.273 0.001 0.000 0.148 0.544 0.030 0.005
(North) 46 0.160 0.003 0.000 0.049 0.757 0.013 0.019

7+ 0.381 0.034 0.000 0.046 0.486 0.000 0.053
Lake trout 1-3 0.51 1 0.000 0.000 0.010 0.473 0.004 0.002
(Central) 4-6 0.556 0.000 0.000 0.001 0.439 0.003 0.001

7+ 0.768 0.005 0.000 0.001 0.222 0.000 0.004
Lake trout 1-3 0.512 0.000 0.000 0.006 0.478 0.001 0.003
(South) 46 0.555 0.000 0.000 0.000 0.443 0.001 0.001

7+ 0.836 0.000 0.000 0.001 0.129 0.000 0.034
Walleye 2-3 0.805 0.000 0.000 0.002 0.190 0.002 0.002
(South) 4+ 0.598 0.000 0.000 0.003 0.378 0.000 0.021
Walleye 2-3 0.429 0.000 0.000 0.003 0.135 0.003 0.431
(Saginaw 4+ 0.436 0.000 0.000 0.004 0.078 0.000 0.482

Bay)

 

165

Table B.4 — Energy density of Lake Huron prey species used in this implementation of
the Wisconsin model. Data were derived from samples collected in Lake Huron (see
Chapter 2) except for invertebrates and “other ﬁsh”, which were not sampled. Mean
energy density for invertebrates (Cummins and Weychuck 1971; Stewart et al. 1983;
Stewart and Binkowski 1986) and for “other ﬁsh” (Stewart et al. 1983; Rudstam et al.

1995) was derived from published values.

 

 

 

Energy density estimated from Energy density
Lake Huron samples estimated from
literature
Month Alewife Bloater Sculpin Rainbow Stickleback Invertebrates Other
smelt fish

Jan 5695 5514 4997 4626 51 94 4248 51 53
Feb 4944 5514 4997 4970 51 94 4248 51 53
Mar 4257 5514 4997 5315 5194 4248 51 53
Apr 5936 551 4 4997 5563 51 94 4248 51 53
May 4549 551 4 4997 581 1 51 94 4248 51 53
Jun 4026 5514 4997 5858 51 94 4248 51 53
Jul 3935 551 4 4997 5540 51 94 4248 51 53
Aug 5028 551 4 4997 51 46 51 94 4248 51 53
Sep 4566 551 4 4997 6061 51 94 4248 51 53
Oct 6297 551 4 4997 7065 51 94 4248 51 53
Nov 6142 5514 4997 5817 5194 4248 5153
Dec 6486 5514 4997 5221 5194 4248 5153

 

 

 

166

Table B.5 — Predator starting weights (grams) as used in the bioenergetics models. The
ending weights were the starting weights for the next age class. For age 2 walleye, no
value is given for the maintenance period since these represent immature individuals that
are not spawning. Therefore, the end weight for age 2 walleye was the starting weight for

age 3. For chinook salmon, starting weight for age 0 ﬁsh was used as in Stewart and

 

 

 

 

 

 

 

 

Ibarra (1991).
Burbot Lake trout Walleye
Age North Central South Growth Maintenance
1 391 45 45 56
2 535 206 147 318 437 n/a
3 685 568 462 790 713 1040
4 835 1,028 957 1,399 1,040 1357
5 980 1,509 1,575 2,064 1,357 1721
6 1,120 1,961 2,232 2,724 1,721 2085
7 1,251 2,359 2,861 3,340 2,085 2411
8 1,373 2,694 3,443 3,891 2,411 2687
9 1,485 2,968 3,951 4,370 2,687 2908
10 1,587 3,188 4,390 4,777 2,908 3033
1 1 1,680 3,362 4,765 5,118 3,033 3080
12 1,764 3,499 5,085 5,400 3,080 3128
13 1,839 3,604 5,357 5,631
14 1,906 3,686 5,578 5,819
15+ 2,018 3,749 5,757 6,208
Chinook salmon
1973- 1981 1982- 1998
Age Annulus Fall Annulus Fall
formation spawning formation spawning_
0 4.54 238 4.54 196
1 572 1 ,739 458 1 ,242
2 3,073 4,791 2,160 3,401
3 7,128 8,823 4,865 5,956
4 9,361 10,378 6,324 7,136

 

167

Table B.6 — Regression results for the ﬁnal model used for each predator. Predators with
no mass cutoff showed no evidence of a relationship between weight and energy density.
Predators with a mass cutoff value were best deﬁned with one model below the cutoff

and another above the cutoff (see Figure B4).

 

 

Mass Line
R2 F df p-value cutoff intersection

Burbot 0.01 11 0.94 1, 84 0.3347 n/a n/a
Chinook salmon 0.3861 19.29 3, 92 <0.0001 4.0 2.98
Lake trout 0.3414 7.26 3, 42 0.0005 1.5 1.51
(North)
Lake trout 0.6173 24.74 3, 46 <0.0001 1.5 1.33
(Central)
Lake trout 0.3502 9.52 3, 53 <0.0001 2.0 1.85
(South)
Walleye 0.0720 3.34 1, 43 0.0746 n/a n/a

 

168

Table 3.7 — Physiological parameters used in the Wisconsin bioenergetics models for

Lake Huron predators. The equations (Eq) and parameters (e.g., CA, FA, etc.) refer to

bioenergetics models as presented by Hewett and Johnson (1995).

 

 

 

 

Consumption Respiration Egestion/ Predator energy Spawning loss
Excretion density
Burbot Eq 2 Eq 2 Eq 1 Eq 1
CA 0.099 RA 0.01 FA 0.17 Joule den 5630 "/0 lost 0.11
CB 0195 RB -0.17 FB 0 Loss day 32
CO 2.41 RQ 1.88 FG 0
CTO 13.7 RTO 21 UA 0.09
CTM 21 RTM 24 UB 0
CTL 0 RTL 0 UG 0
CK1 0 RK1 0
CK4 0 RK4 0
Chinook Eq 3 Eq 1 Eq 3 Eq 2
salmon CA 0.803 RA 0 FA 0.212 PA1 4566
CB -0.275 RB 022 PB 0222 PB1 0.877
CQ 5 R0 0.07 FG 0.631 Mass 2982
cutoff
CTO 15 RTO 0.02 UA 0.0314 PA2 7182
CTM 18 RTM 0 UB 0.58 PB2 0
CTL 24 RTL 25 U6 -0.299
CKA 0.36 RK1 1
CKB 0.01 RK4 0.13
ACT 9.7
BACT 0.04
SDA 0.17
Lake Eq 1 Eq 1 Eq 3 Eq 2
trout CA 0.059 RA 0 PA1 5302
(north) CB 0307 RB -0.3 FA 0.212 PB1 2.285 %lost 0.06
8
00 0.123 RQ 0.06 FB 0222 Mass 1509 Loss day 118
cutoff
RTO 0.02 FG 0.631 PA2 8752
RTM 0 UA 0.0314 PB2 0
RTL 11 UB 0.58
RK1 1 UG -0.299
RK4 0.05
ACT 11.7
BACT 0.04
SDA 0.17

 

 

 

 

 

 

169

Table B.7 continued.

 

 

 

 

 

 

 

 

 

Consumption Respiration Egestion/ Predator energy Spawning loss
Excretion density
Lake Eq 1 Eq 1Eq 3 Eq 2
trout CA 0.059 RA 0 FA 0.212 PA1 5787 % lost 0.06
8
(central) CB 0307 RB -0.3 FB -0.222 PB1 2.431 Loss day 118
CC 0.123 RQ 0.06 FG 0.631 Mass 1325
cutoff
RTO 0.02 UA 0.0314 PA2 8196
RTM OUB 0.58 PB2 .614
RTL 11 UG -0.299
RK1 1
RK4 0.05
ACT 11.7
BACT 0.04
SDA 0.17
Lake Eq 1 Eq lEq 3 Eq 2
trout CA 0.059 RA 0FA 0.212 PA1 6429 °/o lost 0.06
8
(south) CB -0.307 RB 03 FB -0.222 PB1 1.784 Loss day 118
CC 0.123 RQ 0.06 FG 0.631 Mass 1849
cutoff
RTO 0.02 UA 0.0314 PA2 9427
RTM 0 UB 0.58 PB2 0
RTL 11 U6 -0.299
RK1 1
RK4 0.05
ACT 11.7
BACT 0.04
SDA 0.17
Walleye Eq 2 Eq 25a 2 Eq 1
CA 0.25 RA 0.01 FA 0.158 Joule den 6435
(Southern CB 027 RB -0.2 FB -0.222
region CO 2.3 R0 2.1FG 0.631
and CTO 22 RTO 27 UA 0.0253
Saginaw CTM 28 RTM 32 UB 0.58
Bay) CTL 0 RTL 0 UG -0.299
CK1 0 RK1 0
CK4 0 RK4 0

 

 

A. Alewife regional energy density

 

 

 

 

 

Month Central Cl South

B. Alewife estimated seasonal energy density

 

 

8000 -

7000 -

Energy Density (J/g)
:s (n a:
o o o
o o o
o o o

3000 -

 

2000

 

 

 

 

 

 

 

 

Figure 8.1 — Alewife seasonal energy density in 10g1 wet weight. Samples were
available from June through September but only from the central and southern regions
(A). To approximate a seasonal energy density pattern, missing months were estimated
as the proportional difference from published values of alewife energy density from other

Great Lakes (B).

171

 

A. Rainbow smelt seasonal energy from Lake Huron

 

8000 -
7000 —
6000 a
5000 -
4000 -
3000 -
2000 -
1000 -

Energy denslty In Jlg wet weight

 

0 ' r r I

Jan May Jun
Month

 

 

 

 

 

B. Estimated rainbow smelt seasonal energy density

 

8000 -
7000 4

3 6000 -

4000 4
3000 -

 

 

Energy dens

2000 ‘ Literature values
1000 _ + Lake Huron data
- - - Estimated Lake Huron values
0 T I I I . T l ' I V T l I ‘I
J F M A M J J A S O N D
Month

 

 

 

 

 

 

 

 

 

 

 

 

Figure B.2 — Rainbow smelt seasonal energy density in Jog"1 wet weight. Samples were
available for J anurary and from May through August; samples from all lake regions were
pooled (A) as results from Chapter 2 showed no signiﬁcant differences between regions.
To approximate a seasonal energy density pattern, missing months were estimated as the
proportional difference from published values of rainbow smelt energy density from

other Great Lakes (B).

172

Figure B.3 — Linear relationships between predator weight and energy density used in this
implementation of the Wisconsin model. Where two different relationships were

employed, the mass cutoff separating the two lines is indicated below the title.

173

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Burbot Chinook Salmon
<= 4ng J/g = 4566 + 877 Weight
J/g = 5630 > 4kg: J/g = 7182
12000 .
10000 i
e , . g»
i 80004 °; 3 . 3 . . A ,, .
e- .1 .. .- V
E60001,,°‘ovx°3 O . é 0. ° . e
> i O “t . o. I >-
0! Q Q 0 D
a 400°: .° . z
5 1 ° .5
2000 1
o . . . . s - . . . . r . . . ,
0123455 012345678910
Weight (Kg) Weight (Kg)
Lake trout (North) Lake trout (Central)
<= 1.5Kg : J/g = 5302 + 2285 * Weight <= 1.5Kg : J/g = 5787 + 2431 Weight
> 1.5kg : J/g = 8752 > 1.5kg : 8196 + 614 Weight
12000 1 12000 ~ W.»-
. . .
10000 . . I
3 6000« ' 2 .
a , 3 .
B 3‘.
6 400°“ 3 ‘
r. - - r. .
2000 J 2000 ~ ’
0 . T . , . o s T . . f .
0 1 2 3 4 s 6 o 1 2 3 4 5 6
Welght (Kg) Weight (Kg)
Lake trout (South) Walleye
<= 2Kg : J/g = 6148 + 1784 Weight
> 2kg : 9427 J/g = 6435
12000 . 12000 ~
. 0. 1
10000 - e . .OQA : .e. .. 10000 ~
8 1 z.{".' " V a .
3. s 0 ‘o 0 0 ° ’ 3 .
E 8000 W ° e E 8000 , e :. e:
a 0 ,. "1§W;‘~~¢“‘
g 6000 d e E 6000 L . 9 e
,g ‘ >. ‘ O . 0
g 4000 . 9", 4000 .
r. . a .
2000 . 2000 .
o T r r 1 o ‘17 T I I 1 1
0 1 2 3 4 5 6 0 1 2 3 4 s 6
Welght (K9) Weight (Kg)

 

 

 

174

 

 

 

 

Appendix C

Data and Assumptions Used for Projections of Consumption

To estimate future consumption by the key predators, several assumptions were
made regarding mortality rates, weight-at-age, diet composition, and GCE, during the
projection period, 1999 — 2020. Models of the key predators included one for burbot in
the main basin; one for chinook salmon in the main basin; three for lake trout
corresponding to the northern, central, and southern regions of the main basin; and two
for walleye, corresponding to Saginaw Bay and the region of the main basin south of
Saginaw Bay. This appendix describes the assumptions and default values used to
project consumption. A summary of the assumptions is given in Table C] while a more

detailed description is given below.

Mortality rates

Natural mortality rates, excluding sea lamprey—induced mortality, for the
projection period were constant (Table C2) and set to the value used in the last year of
the assessment models (Bence and Dobiesz 2000).

Several types of ﬁshing mortality were applied during the projection period
depending on the predator species. Southern walleye and burbot used a single source of
ﬁshing mortality that was set to the value of the last year of the assessment models; for
Saginaw Bay walleye the average of the last three years was used (Table C.3). All three
lake trout models and the chinook salmon model (Bence and Dobiesz 2000) contained

commercial and recreational ﬁshing mortality calculated as the product of selectivity and

175

ﬁshing intensity (Table C.3). Fishing mortality for projections in the southern lake trout
model used constant selectivity from the last year of data and set ﬁshing intensity to the
average of the last three years. For northern and central lake trout, selectivity and ﬁshing
intensity were allowed to vary over time during pre-projection years. Both variables
were set to the average of the last three years for estimation of ﬁshing mortality in
projected years.

The chinook salmon population model (Bence and Dobiesz 2000) operates with
two time periods within a year consisting of the ﬁrst seven months, then a “pulse” harvest
and maturation process, followed by the remainder of the year. The harvest and
maturation proportions (Table C.4) were set to the estimates for the last year in the
assessment model (Bence and Dobiesz 2000).

Sea lamprey induced mortality was applied to the burbot model and all three lake
trout models (Table C.5). For the projection period, this mortality source was adjusted by
a scaling factor (Schleen et a1. 2002) intended to reﬂect the reduction of sea lamprey

abundance resulting from treatment of the St Marys River (Table C.6).

Weight-at-age

For northern and southern lake trout, burbot, and Saginaw Bay and southern
region walleye, weight-at-age did not change over time in the assessment models (Bence
and Dobiesz 2000). These constant values were used for the projection period (Table
C.7). However, weight-at-age varied over time for chinook salmon and central lake trout
during the pre-projection period. The value for the projection period was the average of

the last three years used in the assessment model (Table C7).

176

Diet and gross conversion efﬁciency

Diet composition (Table CS) and gross conversion efﬁciency (Table C.9) were
assumed constant for estimates of recent and projected consumption. Diet composition
was estimated from agency-collected data (Appendix C). Gross conversion efﬁciency

was estimated from bioenergetics models of each predator population (Appendix C).

Recruitment

During the projection period, recruitment in each year was attributed to natural
reproduction and/or stocking, varying by predator species. Burbot and southern walleye
recruitment was due exclusively to natural reproduction and was held constant during the
projection period (Table C. 10). Neither of these populations was stocked during the
projection period.

A constant number of wild recruits was used for walleye in Saginaw Bay (Table
C.10). The number of walleye stocked into Saginaw Bay varied during 1999 and 2001 of
the projection period and was constant after 2002 to the end of the projection (Table
C .10).

Constant wild recruitment and number of stocked ﬁsh were used for chinook
salmon recruitment during the projection period (Table C. 10). The number of stocked
ﬁsh represents a 20% reduction in chinook salmon stocking which began in 1999.

Lake trout natural recruitment was set to zero for projections. Recruitment from
stocking lake trout in each lake region was obtained using a movement matrix (Table

C11) and a stocking table (Table C. 12). The movement matrix deﬁnes the proportion of

177

ﬁsh stocked at each stocking location that recruit to each lake region. The stocking table
lists annual numbers of ﬁsh stocked in each stocking location. The matrix product of the
stocking table and the movement matrix is a matrix containing the annual number of
recruits in each lake region. After the number of recruits per region was estimated, a
post—stocking survival rate of 0.7399 was applied to the recruits in the southern region

only.

Size regulations

Size regulations in the recreational ﬁshery (Table C.13) were used in the northern
and central lake trout models during the projection period. Recreational mortality rates
(Table C.3) were multiplied by a factor (Table C. 16) to adjust for hooking mortality
experienced by ﬁsh smaller than the minimum size limit in a given year. The adjustment

factor was estimated by

xa,y = PM + (1 — PM ) X h
where xa’ y is the age- and year-speciﬁc adjustment factor that will be applied to

recreational ﬁshing mortality rates; Pa, y is the age- and year-speciﬁc proportion of ﬁsh

that are larger than the minimum size limit; and h is a constant hooking mortality of O. 15.

The proportion of ﬁsh above a speciﬁc size limit

pm =1—Z(sy,aa,oa)

was determined using a cumulative normal distribution with an age-speciﬁc mean (11a)

and standard deviation (Ga) derived from a von Bertalanffy growth model, where sy is a

178

 

year-speciﬁc size limit. Mean length (11a) was estimated using von Bertalanffy growth

parameters (Table C. 14) for northern and central lake trout models. The standard

deviation (Ga) was estimated by multiplying the age-speciﬁc mean length by a constant

coefﬁcient of variation of 0.15 (Table C.15).

Literature Cited

Bence J. R. and N. E. Dobiesz. 2000. Estimating forage ﬁsh consumption by predators
in Lake Huron. Great Lakes Fishery Commission Project Completion Report.
Available for download at http://www.glfc.org/research/cap.htm .

Schleen, Larry R, Christie, Gavin C., Heinrich, John W., Bergstedt, Roger A., Young,
Robert J., Morse, Terry J ., Lavis, Dennis S., Bills, Terry D., Johnson, James E.,
Ebener, Mark P. in press. Development and implementation of an integrated
program for control of sea lampreys in the St. Marys River. Journal of Great
Lakes Research, SLIS 11 Special Issue.

179

Table C.l -— Assumptions used during the projection period. These are default

assumptions in the Consumption Projection Model software but the user may change

them.

 

Item

Natural mortality rates

Assumptions / Settings during projection

Constant during projection period. set to the value used in
the last year of the assessment models. Excludes sea
lamprey-induced mortality.

 

Fishing mortality

Southern and Saginaw Bay
walleye 81 burbot

Constant during projection period. Single source of fishing
mortality is set to the value of the last year of data.

 

Chinook salmon and northern,
central, and southern lake trout

Commercial and recreational fishing mortality included.
Constant during projection period. Value set to last year of
assessment data

 

Sea lamprey-induced mortality
Burbot and northern, central,
and southern lake trout

Used in projection period only. Sea lamprey-induced
mortality from assessment model is adjusted by a scaling
factor to reflect reduction of sea lamprey abundance from
treatment of the St Marys River.

 

Maturation proportion for
chinook salmon

Set to the estimates for the last year in the assessment
model

 

Weight-at-age

Constant during projection period.

 

Diet composition and GCE

Constant during projection period.

 

Stocking
Lake trout

One stocking table Cy lake region used for all lake trout
populations. When used in conjunction with movement
matrix, recruitment data will be changed.

 

All other species

Constant during projection period.

 

Movement matrix for lake trout

Constant during projection period. One movement matrix
used for all lake trout populations. Works with lake trout
stocking table.

 

Size regulations for lake trout

 

Lake trout only

Recreational fishery: 20” in 2001, 22” in 2003, 24” in 2005 W

 

 

Natural recruitment

 

Constant during projection period.

 

180

 

Table C.2 — Natural mortality rates used in projections of consumption.

 

 

Bu rbot Chinook Lake trout Lake trout Lake trout Walleye Walleye
_A_ge salmon North Central South Bay South
0 1 .3048
1 0.6663 0.3000 0.4983 0.5631 0.4168
2 0.3184 0.1000 0.2282 0.2087 0.1911 0.3190 0.2900
3 0.1716 0.1000 0.2282 0.2087 0.1911 0.3190 0.2900
4 0.1235 0.1000 0.2282 0.2087 0.191 1 0.3190 0.2900
5 0.1077 0.1000 0.2282 0.2087 0.1911 0.3190 0.2900
6 0.1025 0.2282 0.2087 0.191 1 0.3190 0.2900
7 0.1008 0.2282 0.2087 0.1911 0.3190 0.2900
8 0.1003 0.2282 0.2087 0.191 1 0.3190 0.2900
9 0.1001 0.2282 0.2087 0.191 1 0.3190 0.2900
10 0.1000 0.2282 0.2087 0.191 1 0.3190 0.2900
1 1 0.1000 0.2282 0.2087 0.191 1 0.3190 0.2900
12 0.1000 0.2282 0.2087 0.191 1 0.3190 0.2900
13 0.1000 0.2282 0.2087 0.191 1
14 0.1000 0.2282 0.2087 0.191 1
15+ 0.1000 0.2282 0.2087 0.191 1

 

181

Table C.3 — Fishing mortality used in projections of consumption.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Fishing mortality Commercial mortality Recreational mortality
Burbot Walleye Lake trout Lake trout
Age Bay South North Central South North Central South
0
1 0.01 10 0.0009 0.0002 0.0007 0.0001 0.0001 0.0005
2 0.0192 0.0000 0.2061 0.0015 0.0006 0.0024 0.0002 0.0002 0.0012
3 0.0306 0.0965 0.4724 0.0102 0.0027 0.0119 0.0014 0.0016 0.0100
4 0.0320 0.0965 0.4061 0.1389 0.01 19 0.0509 0.0117 0.0219 0.0992
5 0.0292 0.0965 0.2549 0.6842 0.0298 0.1061 0.0477 0.0764 0.3220
6 0.0350 0.0965 0.2977 0.8685 0.0335 0.0956 0.0681 0.0798 0.3735
7 0.0387 0.0965 0.2623 0.8360 0.0229 0.0614 0.0710 0.0635 0.3594
8 0.0400 0.0965 0.3052 0.7321 0.0124 0.0360 0.0713 0.0413 0.3350
9 0.0401 0.0965 0.3484 0.5496 0.0061 0.0204 0.0713 0.0218 0.3054
10 0.0414 0.0965 0.5133 0.3310 0.0028 0.0114 0.0713 0.0099 0.2716
1 1 0.0436 0.0965 0.5739 0.1613 0.0013 0.0064 0.0713 0.0042 0.2349
12 0.0481 0.0965 0.4785 0.0688 0.0006 0.0037 0.0713 0.0017 0.1974
13 0.0500 0.0277 0.0003 0.0022 0.0713 0.0007 0.1612
14 0.0500 0.0112 0.0002 0.0014 0.0713 0.0003 0.1282
15+ 0.0500 0.0048 0.0001 0.0009 0.0713 0.0002 0.0995

 

 

 

 

 

Table C.4 — Fishing and maturation proportions for chinook salmon used in projections of

consumption. The chinook salmon model operates with two time periods within a year

consisting of the ﬁrst seven months (prior to a “pulse” harvest and maturation process)

followed by the remainder of the year.

 

Chinook salmon proportion

 

Age Harvest Maturation
0 0 0
1 0.0328 0.041 7
2 0.0929 0.0947
3 0.3320 0.3975
4 0.3320 0.7071
5 0.3320 1 .0000

 

182

Table C.5 — Sea lamprey-induced mortality for lake trout and burbot used in projections

of consumption, before applying the scaling factor (Table C.6).

 

 

Burbot Lake trout

Age North Central South
1 0.0057 0.0000 0.0000 0.0000
2 0.0057 0.0164 0.0121 0.0862
3 0.0190 0.1293 0.0853 0.2632
4 0.0981 0.2615 0.1918 0.3551
5 0.1 203 0.3459 0.2742 0.3805
6 0.0809 0.3864 0.3216 0.3842
7 0.1795 0.4032 0.3472 0.3832
8 0.2937 0.4095 0.3614 0.3816
9 0.3476 0.41 17 0.3697 0.3802
10 0.3520 0.4123 0.3748 0.3793
1 1 0.4134 0.4122 0.3782 0.3786
12 0.5326 0.4119 0.3804 0.3781
13 0.8489 0.41 16 0.3820 0.3777
14 1.0000 0.4114 0.3831 0.3775

15+ 1.0000 0.4114 0.3831 0.3775

 

183

 

Table C.6 - Sea lamprey-induced mortality scaling factor for projection periods. After

2015 the last value of O. 1601 was used for all other years.

 

 

Year Scaling

Factor
1 998 1 .0000
1 999 1 .0142
2000 0.8146
2001 0.4461
2002 0.5090
2003 0.431 7
2004 0.3439
2005 0.3068
2006 0.2623
2007 0.2289
2008 0.2065
2009 0.1 937
2010 0.1789
201 1 0.1702
2012 0.1 639
2013 0.1 61 O
2014 0.1 602
2015 0.1 601

 

184

 

Table C.7 — Predator weight-at-age (kg) used in projections of consumption. Burbot
weight-at-age was obtained from 3 von Bertalanffy growth model ﬁtted to Michigan
Department of Natural Resources (MDNR) data. Lake trout weight-at-age was obtained
from MDNR spring gill new surveys. Walleye weight-at-age was estimated from 1985-
1995 Lake Huron creel data. Both the Saginaw Bay and southern region walleye

populations used the same weight-at-age values.

 

 

Bu rbot Chinook Lake trout Walleye
Age salmon North Central South
0 0.23
1 0.39 1.02 0.05 0.05 0.06
2 0.54 2.68 0.21 0.17 0.32 0.44
3 0.68 5.00 0.57 0.66 0.79 0.71
4 0.83 7.03 1.03 0.90 1.40 1.04
5 0.98 8.60 1.51 1.31 2.06 1.36
6 1.12 1.96 2.03 2.72 1.72
7 1.25 2.36 2.74 3.34 2.08
8 1.37 2.69 3.45 3.89 2.41
9 1.48 2.97 4.02 4.37 2.69
10 1.59 3.19 4.47 4.78 2.91
11 1.68 3.36 4.78 5.12 3.03
12 1.76 3.50 5.00 5.40 3.08
13 1.84 3.06 5.33 5.63
14 1.91 3.69 5.57 5.82
15+ 2.02 3.75 5.72 6.07

 

185

 

Table C.8 — Diet composition for the projection period.

 

 

Prey Species
Age Rainbow Other
Class Alewife Bloater Invertebrate Sculpin Smelt Stickleback fish

Burbot 1-3 0.156 0.000 0.474 0.227 0.136 0.003 0.004
4-7 0.262 0.027 0.330 0.158 0.214 0.004 0.004
8+ 0.264 0.029 0.115 0.087 0.476 0.003 0.026
Chinook 0 0.1 1 1 0.000 0.000 0.002 0.479 0.001 0.408
Salmon 1 0.298 0.000 0.000 0.004 0.634 0.010 0.053
2+ 0.732 0.000 0.000 0.000 0.233 0.029 0.006
Lake trout 1-3 0.273 0.001 0.000 0.148 0.544 0.030 0.005
(North) 4-6 0.160 0.003 0.000 0.049 0.757 0.013 0.019
7+ 0.381 0.034 0.000 0.046 0.486 0.000 0.053
Lake trout 1-3 0.51 1 0.000 0.000 0.010 0.473 0.004 0.002
(Central) 4-6 0.556 0.000 0.000 0.001 0.439 0.003 0.001
7+ 0.768 0.005 0.000 0.001 0.222 0.000 0.004
Lake trout 1-3 0.512 0.000 0.000 0.006 0.478 0.001 0.003
(South) 4-6 0.555 0.000 0.000 0.000 0.443 0.001 0.001
7+ 0.836 0.000 0.000 0.001 0.129 0.000 0.034
Walleye 23 0.805 0.000 0.000 0.002 0.190 0.002 0.002
(South) 4+ 0.598 0.000 0.000 0.003 0.378 0.000 0.021
Walleye 2-3 0.429 0.000 0.000 0.003 0.135 0.003 0.431
(Saginaw 4+ 0.436 0.000 0.000 0.004 0.078 0.000 0.482

BM

186

 

 

Table C.9 — Age-speciﬁc gross conversion efﬁciencies used during the projection period.

 

 

Bu rbot Chinook Lake trout Walleye
salmon
Age North Central South Sag Bay South
0 0.316
1 0.078 0.254 0.215 0.171 0.218
2 0.066 0.171 0.195 0.192 0.175 0.168 0.185
3 0.083 0.079 0.148 0.156 0.139 0.174 0.189
4 0.082 0.066 0.118 0.130 0.116 0.154 0.173
5 0.077 0.105 0.114 0.105 0.151 0.170
6 0.072 0.108 0.111 0.110 0.142 0.160
7 0.068 0.092 0.095 0.094 0.129 0.147
8 0.069 0.081 0.084 0.085 0.118 0.135
9 0.066 0.072 0.076 0.077 0.107 0.123
10 0.064 0.066 0.069 0.070 0.092 0.106
1 1 0.062 0.060 0.064 0.065 0.079 0.091
12 0.060 0.056 0.060 0.061 0.079 0.092
13 0.058 0.053 0.056 0.057
14 0.057 0.051 0.053 0.072
1 5+ 0.054 0.042 0.055 0.063

 

187

Table C. 10 - Number of recruits assumed for projection period.

 

 

 

 

 

 

Natural

Predator Population Recruitment Stocking
Burbot 1 ,137,604 0
Chinook 953,791 2,976,465

US waters 2,578,305

Canadian waters 398,160
All lake trout populations 0 Determined by stocking matrix

(Tables 0.11 and C12)

Southern walleye 366,421 0
Saginaw Bay walleye 389,434 1,006,377 in 1999

 

 

1,106,000 in 2000

645,951 in 2001

1,000,000 fish from 2002 to the
end of the projection.

 

188

Table C.11 — Lake trout movement matrix used during the projection period. This matrix

deﬁnes the percent of ﬁsh stocked in each stocking location that become resident in each

lake region.

 

 

 

 

Stocking Lake region

location North 1 Central South

or 0.973 0.013 ; 0.014

MH1 0.720 " 0.229 0.051

MH2 0.349 H 0.548 ’ 0.103 ’
MH3 0.097 0.355 80.548

MH4 0.000 0.132 ’ 0.868 ‘
MH5 10.000 0.000 ; "1.000”
MH6 10.000 1 0.000 1 1.000

0H3 i 0.349 a 0.548 ; 0.103

0H4 ; 0.000 1 0.132 | 0.868

SFBYR i 0.048 0.091 ”0.861“

 

 

 

 

189

Table C. 12 — Lake trout stocking matrix used during the projection period. This matrix
identiﬁes the number of fish stocked at each location by year. Values after 2001 are

estimates of the numbers to be stocked. No stocking was reported in MH6, 0H3, or 0H4

during the projection period.

 

Year

 

Stocking locations

 

 

DI

 

MH1

 

MH2

 

MH3

 

MH4

 

MH5

 

SFBYR

 

1999'

2000
'2001
2002
2003
2004

2005.
2006
2007;
2008-
2009.
2010’
20113
2012‘

W2013
* 2014

2015.

2016

2017E
2018i
'2019‘

2020

130,000
130,000"
130,000
130,000:
130,000
130,000
130,000;
130,000
130,000?
130,000
130,000
130,000
130,000
130,000
130,000?
130,000
130,000
130,000:
130,000
130,000
130,000

130,000

141,055
147,371

279,000

279,000
279,000
279,000

279,000
' 279,000

279,000
279,000
279,000

279,000 '. '

279,000
279,000

279,000
279,000

279,000

‘ 279,000
279,000

279,000
279,000
279,000

216,900;

226,612-
183,000
338,000
338,000 ”

338,000

338,000
338,000
338,000»

338,000

338,000 ~

338,000

338,000 7 '
338,000

338,000

338,000 ‘

338,000

338,000

338,000
338,000

68,210

134,334 ‘
134,334 '

134,334
134,334

134,334
134,334
134,334

134,334
134,334

134.334"

134,334
134,334

134,334

134,334

134,334

134,334
134,334

195358,
204,106

51 .000:
3 134,333
134,333
"134,333
‘ 134,333

134,333

134,333?
' 134,333-
134,333:
134,333
134,333
134,333 ’
134,333 '
134,333 ‘
134,333 ‘
134,333.
”134,333
134,333
134,333.

1 34,333

18,600;
0:
' 48,000
"134,333
”1 34,333 *
134,333;
134,333]
134,333
134,333
134,333
134,333
‘ 134,333 ‘
134,333 ‘
134.333
134.333 '
134,333
134,333
134,333

134,333 '
134,333
134,333

134,333

360,000
360,000

‘ 360,000

ooooooooooooooo'oooo

 

190

 

 

Table C. 13 — Lake trout recreational ﬁshery minimum size limits during the projection

period.

 

 

Size limit in

Year Inches mm

2001 20 508.0
2002 20 508.0
2003 22 558.8
2004 22 558.8
2005 24 609.6
2006 24 609.6

 

 

Table C. 14 - The von Bertalanffy growth model parameters used to estimate length-at-

age (mm) for northern and central lake trout during the projection period.

 

L... K to cv

”Orthem'aketrout 767.1 0.209 0.00608 0.15
Cent'a"aket'°m 892.8 0.175 -0.1026 0.15

 

191

Table C. 15 — Actual mean length (mm) and standard deviation used to estimate the

adjustment factor on recreational ﬁshing mortality for northern and central lake trout

during the projection period.

 

Northern lake trout

Central lake trout

 

 

Standard Standard
Age Mean length deviation Mean lejgth deviation
1 144.20 21.630 157.16 23.573
2 261 .93 39.290 275.63 41 .344
3 357.41 53.612 375.02 56.253
4 434.85 65.227 458.41 68.761
5 497.65 74.647 528.37 79.255
6 548.58 82.287 587.06 88.058
7 589.88 88.482 636.30 95.444
8 623.38 93.506 677.60 101 .641
9 650.54 97.581 712.26 106.839
10 672.57 100.886 741.34 111.200
1 1 690.44 103.566 765.73 114.859
12 704.93 105.739 786.19 1 17.929
13 716.68 107.502 803.36 120.504
14 726.21 108.931 817.77 122.665
15 733.94 110.091 829.85 124.477

 

 

 

192

Table C. 16 — Size limit adjustment factor on recreational ﬁshing mortality of northern
and central region lake trout. The recreational ﬁshing mortality (Table C3) is multiplied
by the adjustment factor to simulate the effect of hooking mortality related to the
enforcement of the minimum size limit regulations during the projection period.

Northern lake trout

 

 

 

 

 

_Age 2001 2002 2003 2004 2005 2006
1 0.15 0.15 0.15 0.15 0.15 0.15
2 0.15 0.15 0.15 0.15 0.15 0.15
3 0.1521 0.1521 0.1501 0.1501 0.15 0.15
4 0.2614 0.2614 0.1744 0.1744 0.1531 0.1531
5 0.5281 0.5281 0.3254 0.3254 0.2068 0.2068
6 0.7357 0.7357 0.533 0.533 0.3448 0.3448
7 0.8492 0.8492 0.6917 0.6917 0.5 0.5
8 0.9077 0.9077 0.7918 0.7918 0.6248 0.6248
9 0.9388 0.9388 0.8525 0.8525 0.7132 0.7132
1 0 0.9563 0.9563 0.8897 0.8897 0.7737 0.7737
11 0.9668 0.9668 0.9134 0.9134 0.8151 0.8151
12 0.9734 0.9734 0.929 0.929 0.8439 0.8439
13 0.9778 0.9778 0.9397 0.9397 0.8643 0.8643
14 0.9808 0.9808 0.9472 0.9472 0.8791 0.8791
15 0.9829 0.9829 0.9526 0.9526 0.89 0.89
Central lake trout

_A_ge 2001 2002 2003 2004 2005 2006
1 0.15 0.15 0.15 0.15 0.15 0.15
2 0.15 0.15 0.15 0.15 0.15 0.15
3 0.1577 0.1577 0.1505 0.1505 0.15 0.15
4 0.3501 0.3501 0.2113 0.2113 0.1619 0.1619
5 0.6612 0.6612 0.4479 0.4479 0.2798 0.2798
6 0.843 0.843 0.682 0.682 0.4891 0.4891
7 0.924 0.924 0.8229 0.8229 0.6686 0.6686
8 0.9595 0.9595 0.897 0.897 0.786 0.786
9 0.9762 0.9762 0.9359 0.9359 0.8569 0.8569
1 0 0.9848 0.9848 0.9572 0.9572 0.8996 0.8996
1 1 0.9894 0.9894 0.9696 0.9696 0.926 0.926
12 0.9922 0.9922 0.9771 0.9771 0.9429 0.9429
13 0.9939 0.9939 0.982 0.982 0.9542 0.9542
14 0.9951 0.9951 0.9852 0.9852 0.9619 0.9619
1 5 0.9959 0.9959 0.9875 0.9875 0.9673 0.9673

 

193

 

Appendix D

Survey instrument and descriptive analysis of results

Introduction

While an increasing number of computer programs have become available for
modeling ﬁsheries (e.g., CAGEAN, Wisconsin Bioenergetics model of Hewett and
Johnson 1995, and Breck 1998), natural resource management has generally lagged
behind private corporations in implementing user-friendly computer interfaces. Such was
the case with the “No Name” (J. Bence, unpublished data) model, which projected
consumption by key predators in Lake Huron using multiple linked spreadsheets.
However, projecting consumption under multiple management scenarios was
cumbersome and prone to errors common to spreadsheet structure (i.e., copying cells).
To simplify the process of projecting consumption and improve the model interface for
ﬁshery managers, the spreadsheet version of the consumption model was recreated as a
user-friendly computer program . The resulting Consumption Projection Model (CPM)
provides an easy-to-use interface that allows the creation of multiple management

scenarios and comparisons between them.

Objectives
The CPM was intended to improve upon the function and design of the previous

spreadsheet model. To evaluate the effectiveness of the CPM, I conducted a half-hour

194

training session, then asked participants to complete a survey (Figure D-1) designed to
determine the usefulness and ease-of—use of the CPM. Satisfaction with new features,
such as the Windows interface, error messages, and help facilities, was also examined.
The test subjects were stakeholders concerned with piscivore stocking and ﬁshery
management in Lake Huron, including managers in state, tribal, and provincial ﬁsheries
management agencies. Differences in management styles and objectives of these
agencies need to be reﬂected in the CPM computer program to accommodate individual
agency needs. Several questions requiring a written response were used (Figure D-l) to

elicit these differences.

CPM Training Session

A training session was conducted during the July 2002 Lake Huron Technical
Committee (LHTC) meeting in Gore Bay, Ontario. CPM was loaded onto laptop
computers brought by each participant. A 10-minute presentation that reviewed the
consumption model and how the program works preceded the training session. This was
followed by 20 minutes of hands-on demonstration and training on the use of CPM. A
baseline and a modiﬁed scenario were demonstrated while users followed along on their
laptops. Due to time limitations, other program capabilities such as plot ranges, scenario
information, and integrated help were not demonstrated.

At the end of the demonstration and a short question-and-answer period, the
participants were given the survey (Figure D-1) and asked to complete it before the end
of the meeting the next day. Since participants had limited training time to use CPM,

they were asked to use the program on their own during the remainder of the meeting and

195

 

encouraged to ask questions regarding its operation. Participants returned completed
surveys to a third party who placed them in an envelope. Survey responses to the ﬁrst
section were tabulated (Table D. 1) and written responses to the second part were

reviewed and summarized.

Survey Instrument

The survey (Figure D-1) contained two types of questions. The ﬁrst page
contained statements about the overall utility of the new computer program. It contained
three subsections: usefulness, ease-of-use, and general issues. Participants used a 5-point
response scale to indicate disagreement (value=1) through agreement (value=5) with each
statement. The purpose of the response-scale questions was to determine the level of
satisfaction participants gained from using the program. To measure the usefulness of
CPM, respondents were asked if the new program would enhance job performance and be
useful in their daily jobs. The ease-of-use section evaluated the CPM operation and user
interface. The general section evaluated the CPM program vs. the “N 0 Name” model as
well as several other different aspects of the program (e.g., look-and-feel, help facility,
error messages, etc.). [Note, the CPM program was originally named Consume and that
name is used in the survey instruments (Figure 0.1 and Table D.1 ). It refers to the same
computer program]

The second page contained questions that prompted respondents to identify the
parts that worked and those that did not. The written responses were important for
gathering auxiliary information about user satisfaction with the functionality built into the

new computer program. They also served as a method of identifying important functions

196

that would be considered as additions to the CPM before the ﬁnal version of the
computer program was distributed. Detailed input from the users of the new program
was important since only cursory surveying of potential users was done prior to creating
the program. In addition, this feedback helped determine whether user expectations of

the new program were met or not, and identify areas needing improvement.

Survey responses

There were approximately 12 participants in the training session. Since all of the
LHTC meeting attendees were not required to participate, those with less interest in this
topic may not have taken an active role in the hands-on training. Each participant
received a survey; eight surveys were returned. Surveys were later tallied and
summarized (Table D. 1). One participant did not answer any items in the Usefulness

section.

Usefulness

All answers, with the exception of one, to statements in this section received
marks of 3 or higher indicating that the participants believe the CPM program will be
useful in their job (Table 1D.1). The majority of answers in this section (22 of 28) were 46
or 53 indicating that the CPM was perceived as useful for ﬁshery managers.
The last statement in this section, “Consume [CPM] has all the functions and capabilities
I need”, received some of the lowest scores and had the highest variation between scores.
Scores ranged between 2 and 5, with a mean of 3.57. In the written section, many

respondents noted functions they would like to see added to the CPM. Users generally

197

 

agreed that the program provides an important service in their jobs, but some additional

functions would be helpful.

Ease Of Use

In this section, 38 of 40 answers rated the ease-of—use as 4 or higher (Table D.1).
Overall, the CPM interface and methodology was easy to understand and users acquired
sufﬁcient information to operate the basic functions of the program in one short training
session. This attests to the difﬁculty of setting up multiple scenarios and retrieving
graphical output from the “N 0 Name” model. The fourth statement, “Organization on the
screens is clear”, elicited the highest scores, with all participants giving it the highest
mark of 5. All other statements in this section sought to determine whether the program
was easy to use even with minimal training; 30 of 32 answers to these statements were
scored as 4 or 5. It seems that the interface is clear and users ﬁnd it understandable and
easy to use but the process of using CPM and creating a scenario may be somewhat
cumbersome or not well documented. The lack of sufﬁcient training may have

inﬂuenced these results.

Genergl Usage

These statements sought to evaluate many different aspects of the program and,
unlike the previous two sections, each question response will be discussed separately. In
the ﬁrst statement, respondents found the interface pleasing with all answers scoring 4 or
higher (Table D. l). The CPM was rated as a big improvement over the “No Name”

model (7 of 8 respondents scored it as 5) in the second statement. Most respondents did

198

 

not have sufﬁcient time to try out other functions of CPM such as the help facility.
Therefore, 5 out of 8 respondents rated the third statement concerning the use of the Help
facility as “not applicable.” The limited time to use the program before returning the
survey most likely played a role in the number of “not applicable” answers (3 of 8) to the
statement concerning the clarity of error messages. Two of the 5 scored answers were
lower than 4, which may indicate a problem with how the CPM identiﬁes errors it
encounters. The CPM installation program was not available for all Windows® versions,
so I manually installed the CPM on each participant’s laptop prior to the training session.
Since the users could not perform the installation themselves, half of the respondents (4
of 8) scored the statement about ease of installation as “not applicable”. All respondents
scored the last question related to overall satisfaction with the CPM as 4 or 5, indicating

the CPM was generally perceived as easier to use than the “No Name” model.

Written Responses
The purpose of the open-ended questions in this section was to determine which
CPM features the users liked and did not like, and to obtain feedback on improvements to
the CPM that users would like to see. A summary of responses to the four questions are
given below:
1. Are there things that need to be changed or that did not work as predicted?
There were seven responses to this question. Four respondents indicated they could
not respond to this question due to lack of time with the program and three
respondents indicated no change was needed. One respondent indicated a “few

minor bugs” were detected but did not list them.

199

2. Are there things about Consume that you did not like?
Six participants responded to this question. Four respondents indicated that there
was nothing they did not like about the program. One respondent indicated that
more time was needed to evaluate the program while another gave suggestions
about improving the interface and updating parameters.

3. Are there things in Consume that worked well or better than the spreadsheet

model?
Six of the seven respondents thought that CPM was an improvement over the
previous spreadsheet implementation of the model. The seventh respondent
indicated insufﬁcient time to use the program to evaluate the CPM. Four
respondents commented that the graphics and visualization were the important
improvements. Others noted that the automation was the major advance over the
spreadsheet version.

4. Are there additional features that would make this program more useful to you or

yourjob?
Six participants responded to this question; two indicated more time was needed to
evaluate the program. Suggestions for additional features by other respondents
included (1) estimation and projection of sea lamprey induced mortality by species;
(2) documentation describing the source of data used in the model; (3) providing
standard pre-run scenarios depicting commonly used management actions; and (4)

manual and documentation.

200

 

 

Conclusion

Clearly, survey respondents felt that the CPM computer program was an
improvement over the “No Name” spreadsheet version and provided a better user
interface. Most respondents agreed that the CPM was easy to use but several respondents
noted some additional features that would make the program more useful. Many of these
features will be added to the ﬁnal version of the CPM. In particular, a hardcopy manual
will be provided and some standard pre-deﬁned scenarios will be created. These
additions should enhance the usefulness and ease-of-use of the CPM.
Similarly, there were problems with the CPM installation process that prohibited users
from installing the program themselves or scoring the statement concerning ease of
installation. A dependable installation process is necessary to insure that distributed
copies of the CPM can be installed on any Windows computer. Further, a working
installation process for computer programs is considered a norm. The software package
used to create this installation process could not accommodate all versions of Windows
operating systems. It will be abandoned in lieu of a more complete software package that

supports all versions of Windows.

 

While generally quite satisﬁed with the CPM, users needed more time to work
with the program before responding. In scoring the statements on the ﬁrst page of the
survey, a number of answers were marked “not applicable”. Responses to the open-
ended questions on page two often showed that users had insufﬁcient time using the CPM
to evaluate it. These scores and comments point to the need for users to spend more time
using the CPM and to become acclimated to it. Also, some of these responses may be

attributed to the brevity of the training session. A longer, in depth training session, which

201

includes more hands-on examples and exploration of other features available in the CPM,

might have addressed these concerns.

Literature Cited

Breck, J .E. 1998. Development of a warmwater ﬁsh community model. Michigan
Department of Natural Resources, Fisheries Division. Fisheries Research Report
2033.

Dobiesz NE. 2003. Computer Projection Model (installable software and documentation
included). Available for download at the anonymous FTP site at glpd.fw.msu.edu

in the directory CPM V1.0.

Hewett. S. W. and B. L. Johnson. 1995. Fish Bioenergetics Model 3. University of
Wisconsin Sea Grant Institute, WIS-SG-91-250.

202

Table D.1. Answers to survey questions. The number of answers is shown in bold.

 

Statement to be evaluated by respondent

Disagree >>>> Agree
1 2 3 4 5 N/A

Total

Answers

 

USEFULNESS

Using Consume would enhance my effectiveness on the job
Using Consume would make it easier to do my job

Consume would provide an important service I need in my job

Consume has all the functions and capabilities I need

.3
thIUI
NN-‘N

\IVVN

 

EASE OF USE

Learning to use Consume was easy

Consume is simple to use

I find it easy to get Consume to do what I want it to do
Organization of information on the screens is clear

It was easy to define a scenario and run a projection

0000000000

 

GENERAL

The interface is pleasant

The program is an improvement over the spreadsheet models
It was easy to find the information I needed in help files

Error messages clearly identified how to fix problems
Installing Consume on my computer was easy

Overall, I am satisfied with this program

 

 

 

 

 

 

 

000000000000

 

203

 

 

Figure D.1. Questionnaire for
f evaluating the program Consume
? Instructions:

 

Please rate your use of the Consume program. Respond to
each item by filling in the circle that best describes your
experience using the program. For items that are not applicable, use N/A.

Confidentiality

Your responses to this survey are completely confidential. Your privacy will be protected to
the maximum extent allowable by law. By completing and returning this form, you indicate
your voluntary agreement to participate in this survey. This research is supported by

Federal Aid grant F-80-R-2.

If you have questions or concerns regarding your rights as a study participant, or are

dissatisfied at any time with any aspect of this study, you many contact—anonymously, if

you wish—Ashir Kumar, Chair of the University Committee on Research Involving Human

Subjects (UCRIHS) by phone: (517) 355-2180, fax: (517) 353-2976, email: ucrihs @msu.edu

or regular mail: 246 Administration BIdg., East Lansing, MI 48824.

 

 

Disagree >>>> Agree

USEFULNESS 1
Using Consume would enhance my effectiveness on the job
Using Consume would make it easier to do my job

Consume would provide an important service I need in my job

0000
0000

Consume has all the functions and capabilities I need

EASE OF USE

Learning to use Consume was easy

Consume is simple to use

I find it easy to get Consume to do what I want it to do

Organization of information on the screens is clear

000004

It was easy to define a scenario and run a projection

GENERAL

The interface is pleasant

The program is an improvement over the spreadsheet models
It was easy to find the information I needed in help files

Error messages clearly identified how to fix problems

Installing Consume on my computer was easy

0000004
00000010
00000000
0000004?-
0000000'

Overall, I am satisfied with this program

2

OOOOOm

3

0000

000000

4

0000

0000043

5

0000

OOOOOm

N/A

0000

N/A

2
3+, 00000

000000

 

Please continue on the next page. 0'

204

 

 

 

OTHER COMMENTS

Are there things that need to be changed or that did not work as predicted? (Please
specify)

Are there things about Consume that you did not like? (Please specify)

Are there things in Consume that worked well or better than the spreadsheet
model? (Please specify)

Are there additional features that would make this program more useful to you or
your job? (Please specify)

Before the end of this meeting, please return
this questionnaire to Jim Bence.

 

205