,mmxuﬂﬂmuhﬁ
U .n. . [Edward .
. V ,4 . a
, . . ‘31...
, «f2
.‘ﬁﬁw

.r an
mxﬂidmwmm.‘ \-
. u”.

. ‘ .
. _ .
A :1:
.a 1).” .
z» ..
. i to “Eat:
«a. n .23.

1:4:
:5?
.10.”. 3.... ELY. ...l.:
v auxgamxuuvc..ul~:t:tlﬂ:i. .
v 1.3.. 1%.){3 .
V f“ .

u; .6.
95:... ‘

3..

.x ... .3

o.

 

 

Quasar

123.25,. 1 . . ‘ . ,

m

l
.2 .At

 

 

in9 ' UBRARY
Michigan State
University

 

This is to certify that the
dissertation entitled

EXOGENOUS CORRELATES OF MIGRATION,
SPAWNING, EGG DEPOSITION AND EGG MORTALITY
IN THE LAKE STURGEON (A CIPENSER F UL VESCENS)

presented by

Patrick Stephen Forsythe

has been accepted towards fulﬁllment
of the requirements for the

Ph.D. degree in Zoology
Ecology, Evolutionary Biology
and Behavior

 

 

//..; Spa

Major Professor’s Signature
4/ - 6- lo
Date

 

 

MS U is an Afﬁrmative Action/Equal Opportunity Employer

 

-.-o-.-.-«—.-.—‘— —

.o--.-.-o--—n-:-c—n-.—

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

Lam 1.2 291.2
I h I 4‘ fl

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5l08 KzlProjIAocaPresIClRCIDateDueindd

EXOGENOUS CORRELATES OF MIGRATION, SPAWNING, EGG DEPOSITION
AND EGG MORTALITY IN THE LAKE STURGEON (ACIPENSER F UL VESCENS)

By
Patrick Stephen Forsythe

A DISSERTATION

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

DOCTOR OF PHILOSHOPHY

Zoology
Ecology, Evolutionary Biology and Behavior

2010

ABSTRACT

EXOGENOUS CORRELATES OF MIGRATION, SPAWNING, EGG DEPOSITION
AND EGG MORTALITY IN THE LAKE STURGEON (ACIPENSER F UL VESCENS)

By
Patrick Stephen Forsythe

Understanding patterns of migration, the timing of reproduction, breeding site
selection and the sources and magnitude of early life mortality provides a unique
perspective into exogenous regimes that inﬂuence patterns in recruitment as well as those
leading to the evolution of important and adaptive life history traits. Unfortunately, such
knowledge of how migratory patterns and reproductive behaviors, the consistency of
reproductive events and the time/location of reproduction relates to exogenous variables
is largely lacking for many species, especially long-lived iteroparous vertebrates. Further,
associations between habitat variables and patterns in the occurrence of early life stages,
quantitative estimates of mortality, and the sources of loss are relatively unknown. In
order to address these uncertainties, I used tagging data from a long-term lake sturgeon
monitoring program (2000-2008) established in the Black River, Nﬂ as well as
experimental manipulations and laboratory studies to: 1) quantify relationships between
migration patterns and spawning behaviors with exogenous variables, 2) estimate the
consistency (i.e., repeatability) for individual spawning events across years, 3) assess
whether stream covariates were predictive of spawning site selection, egg deposition and
survival during incubation in nann'al settings and 4) quantify the magnitude, sources and
the abiotic and biotic conditions (e. g., substrate size, predator density and size) that
inﬂuence egg mortality. Migratory activity (i.e., the daily number of adults entering the

river) and spawning activity (i.e., the daily number of adults collected at upstream

spawning locations) were found to be signiﬁcantly associated with exogenous variables
including daily photoperiod, water temperature, river discharge and the lunar cycle.
However, the relative importance of variables interacted in different ways at different
sequences in migration continuum (i.e., river entry, upstream movement, entry to
spawning locations). Individual male and female lake sturgeon were also found to be
highly repeatable in their choices of spawning time and spawning location (i.e., upstream
vs. downstream preference). Egg deposition in natural stream settings was spatially
explicit (i.e., patchy) and patterns were non-randomly and signiﬁcantly associated with
stream covariates including substrate size and water velocity across all sampled locations.
Although egg mortality was consistent between sites over three days (70% mortality
rate), environmental covariates, particularly substrate size, were predictive of mortality
rates. Finally, egg mortality was high (91%) and quite variable (75-97%) across treatment
groups in experimental settings and the relative magnitude of mortality sources varied
signiﬁcantly across simulated conditions, further suggesting strong relationships with the
ecological features eggs experience naturally during incubation and survival.
Developmental mortality (i.e., mortality due to abnormal egg development) was the
major source of loss (84%) followed by predation (16%). However, egg mortality due to
predation was shown to be high in laboratory experiments (80% aﬁer 24 hrs) with
approximately 17-104 eggs consumed on average per predator. The highest levels of
mortality were observed in treatments with high predator density and little habitat
complexity. Information provided has been largely unknown and will be valuable in the

ongoing restoration efforts for this unique and imperiled migratory ﬁsh.

Copyﬁght by
Patrick Stephen Forsythe

2010

ACKNOWLEDGMENTS

I ﬁrst of all would like to thank my wife, Nicole, for supporting me over the past
9 years and providing unwavering love, encouragement and support at times when this
process was nearly unbearable. I would also like to thank my parents, vae and Patricia,
as well as my mother in law, Bonnie, for sacriﬁcing their own time for helping to make
sure that this chapter of my life was completed to the best of my abilities. Finally, I
would like to dedicate the ﬁnal stages of this dissertation to my twin sons, Conner and
Landen, who I hope will eventually go conﬁdently in the direction of their own dreams
and live the life they imagine (W.B. Yeats). Also, remember that God is great, beer is
good and yes, people are deﬁnitely crazy (B. Currington), especially your father!

Several people also made a valuable contribution to my academic growth during
my time at Michigan State University that I wish to thank. First, I thank my graduate
advisor Dr. Kim Scriber for giving a small town boy from Central Illinois a chance to
work on an amazing project. I would also like to thank Dr. Scribner for his valuable and
numerous comments on the many drafts of this dissertation. Second, I would like to thank
my fellow molecular ecology lab members, a multitude of ﬁeld technicians hired over the
years, ﬁsheries biologists from the Michigan Department of Natural Resources Gaylord
Ofﬁce and members of the Black Lake Chapter of Sturgeon for Tomorrow including
Jamie Crossman, Christin Davis, Patrick Vandaele, Tim Cwalinski, Niel Godby, Dave
Borgeson, Jason Woiderski, Dick and Sylvia Sanford and Gil and Brenda Archambo for
their overwhelming assistance in the ﬁeld and helping with the experimental setup and
design. Third, I owe a special thanks to my “statistical advisors” Andrew Finley, Nora
Bello, and Ashok Ragavendran for providing unmatched statistical consultation and for

their enormous efforts to assure that this data set was properly analyzed. Fourth, I thank

V

my committee members Dr. Gary Mittelbach, Dr. Daniel Hayes, Dr. Mary Bremigan and
Dr. Ed Baker for reviewing and providing comments on this material and. for their
mentorship over the years both in the classroom and out.

Finally, my thanks goes out to generous support in the form of numerous grants,
scholarships and fellowships from the Michigan Department of Natural Resources, Great
Lakes Fishery Trust, Michigan Experimental Station, Sturgeon For Tomorrow, the
National Fish and Wildlife Foundation Budweiser Conservation Scholarship Program
sponsored by Anheuser-Busch, the Department of Zoology, the Ecology, Evolutionary
Biology and Behavior Program and Presque Isle Electric and Gas Co-Op. The research
was conducted under an animal care permits issued by the Animal Use and Care
Program, Michigan State University, and collection and research permits issued by the

Michigan Department of Natural Resources.

vi

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................. ix
LIST OF FIGURES .......................................................................................................... xii
INTRODUCTION ............................................................................................................... 1
CHAPTER 1 ........................................................................................................................ 5
EXOGENOUS CUES ASSOCIATED WITH MIGRATION AND SPAWNING 1N
LAKE STURGEON
INTRODUCTION ................................................................................................... 5
METHODS .............................................................................................................. 8
RESULTS .............................................................................................................. 16
DISCUSSION ........................................................................................................ 23
CHAPTER 2 ..................... '. ................................................................................................ 34
REPEABILTY FOR SPAWNING TIME AND SPAWNING SITE SELECTION IN
THE LAKE STURGEON
INTRODUCTION ................................................................................................. 34
METHODS ............................................................................................................ 37
RESULTS .............................................................................................................. 43
DISCUSSION ........................................................................................................ 46
CHAPTER 3 ...................................................................................................................... 56

SPATIALLY EXPLICIT ANALYSIS OF LAKE STURGEON EGG DEPOSITION
AND MORTALITY IN NATURAL STREAM SETTINGS

INTRODUCTION ................................................................................................. 56
METHODS ............................................................................................................ 60
RESULTS .............................................................................................................. 68
DISCUSSION ........................................................................................................ 73
CHAPTER 4 ...................................................................................................................... 81

SOURCE-BASED ASSESSMENT OF LAKE STURGEON EGG MORTALITY IN
NATURAL AND EXPERIMENTAL STREAM SETTINGS
INTRODUCTION ................................................................................................. 81

METHODS ............................................................................................................ 85

vii

RESULTS .............................................................................................................. 92

DISCUSSION ........................................................................................................ 96
SYNTHESIS AND CONCLUSIONS ............................................................................. 103
APPENDIX A .................................................................................................................. l 12

TABLES AND FIGURES
LITERATURE CITED .................................................................................................... 150

viii

LIST OF TABLES

TABLE 1.1 .......................................................................................... 113

Generalized chi-square (xz/degrees of freedom) and R-square (R2) ﬁt statistics for models
explaining daily numbers of individuals entering the river mouth and arriving to upstream
spawning areas in the Upper Black River, Michigan and those that account for temporal
autocorrelation in counts (AC), environmental (ENV) and lunar (LUNAR) parameters or
a combination of all three.

TABLE 1.2 .......................................................................................... 114
Parameter estimates, parameter conﬁdence intervals (lower and upper 95% conﬁdence
intervals), and test statistics for environmental (Dlag = 24 hr lagged discharge, Tlag = 24
hr lagged temperature, Tchange = 24 hr change in temperature and Dchange = 24 hr
change in discharge) and lunar effect (9 = lunar angle) models describing daily
observations of individuals entering the mouth of the Upper Black River, Michigan,
2000-2002.

TABLE 1.3 .......................................................................................... 115
Parameter estimates, parameter conﬁdence intervals (lower and upper 95% conﬁdence
intervals), and test statistics for environmental (Dlag = 24 hr lagged discharge, Tlag = 24
hr lagged temperature, Tchange = 24 hr change in temperature and Dchange = 24 hr
change in discharge) and lunar effect (0 = lunar angle) models describing daily numbers
of individuals arriving to spawning locations in the Upper Black River, Michigan 2001-
2008.

TABLE 1.4 .......................................................................................... 116
Parameter estimates, parameter conﬁdence intervals (lower and upper 95% conﬁdence
intervals), for combined environmental (Dlag = 24 hr lagged discharge, Tlag = 24 hr
lagged temperature, Tchange = 24 hr change in temperature and Dchange = 24 hr change
in discharge) and lunar effect (6 = lunar angle) models describing daily numbers of
individuals entering the mouth of the Upper Black River, Michigan and arriving to
spawning locations, 2001-2008.

TABLE 2.1 .......................................................................................... 125
Estimates of repeatability by sex for three size classes of lake sturgeon in relation to
standardized spawning day, river zone and environmental variables of the Upper Black
River, MI. Variables are rank ordered by repeatability. Asterisks indicate variables where
statistical evidence supported a model with heterogeneous variances and thus size/age-
based differences in repeatability.

TABLE 3.1 .......................................................................................... 133
Summary of egg deposition non-spatial and spatial models’ parameter estimates for Site 1

2 . . . . .
(o = variance of the random spatial effect; (p = spatlal range orientation, Ag and 1 =
effective range parameter). Parameter posterior credible intervals i.e., 50 (2.5, 97.5)
percentiles are provided. Those B parameters for environmental covariates [i.e., Depth =

ix

water depth (m); Velocity = water velocity (m3/s); Angub = average substrate size
(mm); MaxSub = maximum substrate size (m)] that are signiﬁcant at the 0.05 level are
bolded. Model ﬁt in the form of Deviation Information Critera (DIC) and model’s
effective number of parameters (pD) are also provided for comparison.

TABLE 3 .2 .......................................................................................... 134
Summary of egg deposition non-spatial and spatial models’ parameter estimates for Site 2

(o = variance of the random spatial effect; (1) = spatial range orientation, A0 and 1 =
effective range parameter). Parameter posterior credible intervals i.e., 50 (2.5, 97.5)
percentiles are provided. Those B parameters for environmental covariates [i.e., Depth =

water depth (m); Velocity = water velocity (m /s); Angub = average substrate size
(mm); MaxSub = maximum substrate size (m)] that are signiﬁcant at the 0.05 level are

bolded. Model ﬁt in the form of Deviation Information Critera (DIC) and model’s
effective number of parameters (pD) are also provided for comparison.

TABLE 3.3 .......................................................................................... 135
Summary of egg deposition non-spatial and spatial models’ parameter estimates for Site 3

(o = variance of the random spatial effect; (p = spatial range orientation, A0 and 1 =
effective range parameter). Parameter posterior credible intervals i.e., 50 (2.5, 97.5)
percentiles are provided. Those B parameters for environmental covariates [i.e., Depth =
water depth (m); Velocity = water velocity (m /s); Angub = average substrate size
(mm); MaxSub = maximum substrate size (m)] that are signiﬁcant at the 0.05 level are

bolded. Model ﬁt in the form of Deviation Information Critera (DIC) and model’s
effective number of parameters (pD) are also provided for comparison.

TABLE 3.4 .......................................................................................... 136
Summary of egg deposition non-spatial and spatial models’ parameter estimates for Site 4
(o = variance of the random spatial effect; (p = spatial range orientation, 7‘0 and 1 =
effective range parameter). Parameter posterior credible intervals i.e., 50 (2.5, 97 .5)
percentiles are provided. Those B parameters for environmental covariates [i.e., Depth =
water depth (m); Velocity = water velocity (m /s); Angub = average substrate size
(mm); MaxSub = maximum substrate size (m)] that are signiﬁcant at the 0.05 level are
bolded. Model ﬁt in the form of Deviation Information Critera (DIC) and model’s
effective number of parameters (pD) are also provided for comparison.

TABLE 3.5 .......................................................................................... 137
S of egg count change non-spatial and spatial models’ parameter estimates for

Site 3 (t = measurement error; 0 = variance of the random spatial effect; (p = spatial

range orientation, 710 and 1 = effective range parameters). Parameter posterior credible
intervals i.e., 50 (2.5, 97 .5) percentiles are provided. Those B parameters for
environmental covariates [i.e., Depth = water depth (m); Velocity = water velocity

(m3/s); Angub = average substrate size (mm); MaxSub = maximum substrate size
(m)] that are signiﬁcant at the 0.05 level are bolded Model ﬁt in the form of Deviation

Information Critera (DIC) and model’s effective number of parameters (pD) are also
provided for comparison.

TABLE 3.6 .......................................................................................... 138
S of egg count change non-spatial and spatial models’ parameter estimates for

Site 4 (r = measurement error; 0 = variance of the random spatial effect; (p = spatial

range orientation, A0 and 1 = effective range parameters). Parameter posterior credible
intervals i.e., 50 (2.5, 97 .5) percentiles are provided. Those B parameters for
environmental covariates [i.e., Depth = water depth (m); Velocity = water velocity

(m3/s); Angub = average substrate size (mm); MaxSub = maximum substrate size
(m)] that are signiﬁcant at the 0.05 level are boldecl Model ﬁt in the form of Deviation
Information Critera (DIC) and model’s effective number of parameters (pD) are also
provided for comparison.

TABLE 3.7 .......................................................................................... 139
S of cross-validation multivariate spatial models’ parameter estimates for Sites 3
and 4 (o = variance of space-varying matrix elements 1, 2 and 3; (p = spatial range

parameter for matrix elements (a) and the space-varying correlation (v); 1' = residual
measurement error for time I and 2 covariance matrix. Parameter posterior credible .
intervals i.e., 50 (2.5, 97.5) percentiles are provided. Those Bparameters for
environmental covariates [i.e., Depth = water depth (m); Velocity = water velocity

(m3/s); Angub = average substrate size (mm); MaxSub = maximum substrate size
(mm)] collected during each sampling period (t =1, t = 2) that are signiﬁcant at the 0.05
level are bolded.

xi

LIST OF FIGURES

FIGURE 1.1 ......................................................................................... 117

Measurements of daily water temperature (DC) of the Upper Black River, Michigan
collected from April 22 through June 5 for 2000-2002 (upper panel), for 2003-2005
(middle panel) and 2006-2008 (lower panel).

FIGURE 1.2 ......................................................................................... 118

Measurements of daily river discharge (m3/s) of the Upper Black River, Michigan
collected from April 22 through June 5 for 2000-2002 (upper panel), for 2003-2005
(middle panel) and 2006-2008 (lower panel).

FIGURE 1.3 ......................................................................................... 119
Number of adult lake sturgeon observed entering the Upper Black River, Michigan 2000-
2002 as function of the hour of the day.

FIGURE 1.4 ......................................................................................... 120
Observed, predicted and 95% conﬁdence intervals for the daily number of lake sturgeon
entering the Upper Black River, MI from 2000-2002.

FIGURE 1.5 ......................................................................................... 121
Relationship between the daily number of lake sturgeon observed entering the mouth of
the Upper Black River, Michigan with 1 day lagged river discharge and 1 day lagged
water temperature.

FIGURE 1.6 ......................................................................................... 122
The number of lake sturgeon observed entering the Upper Black River or captured at
upstream spawning location as a ﬁmction of the date in 2001 and 2002.

FIGURE 1.7 ......................................................................................... 123
Observed, predicted and 95% conﬁdence intervals for the daily number of lake sturgeon
captured at spawning locations the Upper Black River, M] from 2001-2008 and
descriptions of the proportion of lunar illumination for each day.

FIGURE 1.8 ......................................................................................... 124
Relationship between daily numbers of lake sturgeon captured at spawning locations on
the Upper Black River, Michigan and lunar angle with A) 24 hour lagged river discharge,
B) 24 hour lagged water temperature and C) the magnitude and change in water

temperature over the previous 48 hours.

FIGURE 2.1 ......................................................................................... 126
Map of the Upper Black River, MI and spawning areas (zones) used by adult lake
sturgeon

FIGURE 2.2 ......................................................................................... 127
Number of male and female lake sturgeon captured during each day of the season in each
of eight consecutive years (2001-2008).

xii

FIGURE 2.3 ......................................................................................... 128
Distribution of total number of capture and recapture events (N =1260) for individual
male and female lake sturgeon (N=678) collected over eight consecutive years (2001-
2008) in the Upper Black River, MI.

FIGURE 2.4 ......................................................................................... 129
Frequency distribution of observed inter-annual spawning intervals in years for female
(top) and male (bottom) lake sturgeon spawning in the Upper Black River, Michigan. The
data are presented as the probability of returning to spawn in year t + x given that
spawning occurs in year t. For example, there is an 18% chance that a male spawning in
year t will spawn the next year. Asterisk indicates upwardly biased probability estimates
due to individuals likely missed between captures.

FIGURE 2.5 ......................................................................................... 130
Predicted standardized spawning day (:h 1 SE) for recaptured male (N = 232) and female
(N = 70) lake sturgeon captured on 3 2 occasions during 8 consecutive years (2001-
2008). Individuals are presented in rank order by predicted standardized spawning day.

FIGURE 2.6 ......................................................................................... 131
Odds of lake sturgeon spawning one of 5 different river zones in the Upper Black River,
MI relative to the most upstream zone (Zone 1) in relation to standardized spawning day.

FIGURE 2.7 ......................................................................................... 132
Distribution of male and female lake sturgeon body size as a function of standardized
spawning day for the Upper Black River, MI population. Data represents captures of all
individuals collected from 2001-2008, including recaptures.

FIGURE 3.1 ......................................................................................... 140
Map of the Upper Black River and the location of four Sites used for spawning activity

(Right).

FIGURE 3.2 ......................................................................................... 141
Frequency histograms for average substrate size, maximum substrate size, water depth
and water velocity for Sites 1-4.

FIGURE 3.3 ......................................................................................... 142

Interpolated surfaces of water depth (m), velocity (m3/s), average substrate and

maximum substrate (mm) environmental covariates and observed egg count (top to
bottom) for Site 3.

FIGURE 3.4 ......................................................................................... 143
Interpolated surfaces of the non-spatial model residuals, non-spatial ﬁtted values, spatial
random effects and ﬁtted values of the spatial model (top to bottom) for Site 3.

FIGURE 3.5 ......................................................................................... 144
Interpolated surfaces of live eggs (time 1 and 2), random spatial effects (w(s) time 1 and
2), and residual spatial correlation (p t=1, t=2 (5)) from model (3) for Site 3 (top to
bottom). Points represent locations where there was at least one live egg in time 1.

xiii

FIGURE 3.6 ......................................................................................... 145
Interpolated surfaces of live eggs (time 1 and 2), random spatial effects (w(s) time 1 and
2), and residual spatial correlation (p t=l, t=2 (5)) from model (3) for Site 4 (top to
bottom). Points represent locations where there was at least one live egg in time 1.

FIGURE 4.1 ......................................................................................... 146
Experimental design used to estimate egg mortality and partition components of loss
(Top). Four treatments of in situ exclosures and hypotheses regarding the source,
direction and magnitude of exogenous removal (i.e., predation and water currents) and
the magnitude of mortality due to developmental mortality (Bottom).

FIGURE 4.2 ......................................................................................... 147
Least square mean differences in the A) percent of eggs remaining, B) percent of eggs
dead due to developmental failure and C) percent of eggs dead due sources leading to
exogenous removal among Open, Large mesh, Small mesh and Fine mesh exclosure
treatments during 120 hours of incubation in natural stream conditions. Treatment means

with overlapping conﬁdence intervals generally indicate signiﬁcant pair-wise differences
(a=0.05).

FIGURE 4.3 ......................................................................................... 148
Least square mean proportion of eggs surviving across three substrate sizes (Open,
Gravel and Large Rock and four different simulated densities of crayﬁsh predators.

FIGURE 4.4 ......................................................................................... 149

Proportion of eggs consumed as a function of mean predator size in each of four density
treatments.

xiv

INTRODUCTION

Migratory animals have captivated scientists for decades as I have been for the
past seven years. Migratory ﬁshes in particular have received considerable attention
(reviewed in 1977 by Leggett) because these animals often move over great distances,
they have a strong tendency to return to speciﬁc locations of streams often of natal origin
and they arrive to reproduce at extremely predictable times of the year (Stewart et a1.
2002). Migratory ﬁshes have evolved this behavior in part due to the ﬁtness beneﬁts of
placing their offspring in environments, often streams and rivers, best suited for survival
and growth (Jorgensen et a1. 2008). However, the road to achieving reproductive success
is not easy. For example, individuals must ﬁrst enter and move upstream at times that
reduce energetic and physiological demand (Cooke et a1. 2004) or minimize the risk of
mortality occurring en route to spawning areas (Quinn et a1. 2001). They must also
coordinate the time of arrival to spawning sites with gonad maturation, release offspring
(i.e., embryos) at times best matched with developmental requirements during incubation
(e.g., temperature) and select locations that provide the best possible chance of survival
(Einum and Fleming 2000; Malcolm et a1. 2003).

The direct exposure of migratory ﬁsh embryos to the environment not
surprisingly leads to disproportionally high levels of mortality, relative to older juvenile
stages, that often plays a critically important role in both population dynamics and
regulation (Houde 1987). In fact, variation in a single source of mortality such as
predation has been shown to generate large and predicable annual ﬂuctuations in
recruitment or the addition of new individuals to the population (Bailey and Houde

1989). However, because of these survival and ﬁtness consequences to both offspring and

adults, natural selection has also played a vital role in the evolution of a host of genetic
(i.e., heritably for spawning time) and physiological (i.e., visual and olfactory
mechanisms) traits that tend to enhance the ability of individuals to use exogenous cues
for selecting the best times to migrate and reproduce (Quinn and Adams 1996). Likewise,
migratory ﬁshes have also evolved an array of reproductive behaviors (e.g., nest
construction, egg fanning), physiological defenses (e. g., induced hatching), anti-predator
mechanisms and abilities to use exogenous information for selecting breeding locations
that compensate for excessive loss in natural settings (Knouft et a1. 2003). Interestingly,
the times and habitats chosen for reproduction can also impose selection on phenotypic,
behavioral, and morphological traits of offspring, most of which are also tied in some
way to survival during subsequent early life stages that have long since dispersed from
breeding grounds (Hendry et a1. 2005).

While the ecology of migration and aspects the early life history of many ﬁshes is
becoming better understood, the responses of migratory ﬁshes to exogenous cues during
migration, while spawning or to habitats used for reproduction are not always concordant
(e. g., Dahl et a1. 2004). Further, information regarding the relative response of individuals
to exogenous variables for migration and reproduction and the role environmental
covariates in egg mortality is based largely on a limited number of species that have very
similar ecologies. Take sturgeons for example, a highly migratory lineage of ﬁshes with
extremely unique life history characteristics (i.e., longevity and iteroparity) and mating
behaviors (i.e., broadcast spawning) where much tighter associations between responses
that have ﬁtness consequences and exogenous cues that elicit them may be expected.

However, we know little about the exogenous cues that regulate the reproductive cycle

and those that synchronize annual migration and reproductive events. Quantitative
information regarding inter-annual variation in natural recruitment and of the
environmental or biotic agents responsible for inter-annual variation during early life
stages, particular during the embryonic stage, is lacking. In fact, the entire life history,
reproductive ecology and processes dictating levels of annual recruitment of sturgeon
species has been described as a jigsaw puzzle more pieces missing than not (Jager et a1.
2002).

My dissertation research empirically addresses this uncertainty using the lake
sturgeon as a focal species. In my ﬁrst chapter, I decompose annual reproduction into
several stages including the daily numbers of individuals initiating migration at the river
mouth, upriver movement and daily number of adults entering areas upstream to spawn. I
then investigate whether observed patterns are related to exogenous cues including
photoperiodism, lunar cycle, photoperiod, water temperature and river discharge. In the
second chapter, I use data from individuals captured over multiple years to estimate
spawning periodicity (i.e., inter-annual spawning intervals) and to quantify the relative
degree of consistency (i.e., repeatability) in spawning time within a breeding season with
respect to either time-related cues or environmental covariates. In chapter 3, I use
hierarchal mixed spatial models to determine whether stream variables are predictive of
variation in egg deposition and survival and test whether rates of embryo mortality are
constant across locations after environmental covariates are accounted for. Finally, in
chapter 4, I estimate mortality during the egg stage as a function of experimentally
manipulated incubation conditions and partition mortality into its dominant sources. I

compare and contrast the signiﬁcance of my results in light of what is known about other

migratory ﬁshes and from ecological ﬁ'om evolutionary and conservation perspectives in
the synthesis and conclusions section.

I also hope that my research will aid considerably in the rehabilitation of this
species which where once reported to be one of the ﬁve most abundant commercial ﬁshes
in the Great Lakes during the late 1800’s but are now believed to be at less than 1% of
historic levels (Hay-Chmielewski and Whelan 1994). The main causes of decline are
overexploitation, destruction of spawning habitat, barriers to migration and the pollution
of streams and rivers (Auer 1999). Currently, lake sturgeon are classiﬁed as endangered,
threatened, or a species of special concern across its geographic range (Auer 1999). In
recent years, conservation and rehabilitation of lake sturgeon has become a high priority
throughout the Great Lakes region and worldwide for other sturgeon species. The
management activities include reducing natural and ﬁshing mortality of adults,
enhancement of spawning sites through the addition of substrate, the implementation of
ﬁsh passage solutions, stocking, regulation of rivers during migration and spawning and
pollution control. Despite these actions and high expectations for recovery, most extant
populations continue to show traces of natural reproduction. Since our knowledge of
fundamental aspects of the species’ reproductive ecology is lacking, I believe that
restoration programs such as spawning site enhancement or river regulation tactics
currently based on “best professional judgment” rather than quantitative data will

continue to be relatively ineffective.

CHAPTER 1

EXOGENOUS CUES ASSOCIATED WITH MIGRATION AND SPAWNING IN
LAKE STURGEON

INTRODUCTION

There is strong selection for animals to breed at times and in locations that
enhance the survival of offspring. Migration between areas used for foraging and areas
used for reproduction is a behavior that has evolved in many organisms in part because of
ﬁtness advantages associated with placing progeny in habitats that enhance probabilities
of survival during early life stages (Alerstam and Enckell 1979; Alerstarn et a1. 2003).
The ability of migratory animals to coordinate arrival times at breeding locales and
during periods of optimal exogenous (i.e., all external inﬂuences) conditions has
important ﬁtness consequences, especially when recruitment variation is attributed to
high rates of mortality during early life history stages at or near the sites of reproduction
(Smith and Moore 2005; Forchhammer et a1. 2002; Secor and Houde 1995).

Migration can be viewed as a continuum whereby individuals respond to
exogenous cues, processing different sources of information en route that then culminates
in reproduction (Duriez et al. 2009). However, the relative importance of exogenous cues
may vary at different sequences during migration (i.e., onset of migration vs. onset of
reproduction after arrival), and cues may interact to enlist different responses in different
species, populations or individuals. For example, the probability that sea turtles will
depart foraging areas and initiate migration to nesting areas is associated with rising
chlorophyll-a concentration and rising sea-surface temperature (Sherrill-Mix et a1. 2008).
However, geomagnetic cues are used by adults to orient in open water once migration is

initiated (Luschi et al. 2007), chemical cues released from natal beaches are used for ﬁne-

scale homing (Lohmann et al. 1999), and local environmental cues such as rainfall,
temperature and tides are correlated with the probability individuals will emerge ﬁom the
ocean and initiate nesting behavior (Pike 2008).

Initiation of movement into tributaries and the timing of reproduction represent
important behavioral decisions that are elicited by reliable cues (e.g., day length), and
cues related to features of the physical environment, and which are associated with
adaptive outcomes in ﬁshes (e.g., successful reproduction) (Williams and Nichols 1984).
Migrating at favorable times can reduce mortality en route and may conserve energy
needed to compete for mates and quality breeding locations (Cooke et al. 2004; Crossin
et a1 2003; Hinch and Rand 1998). Adults must also time reproduction so that offspring
are placed in conditions suited for growth and survival (Einum and Fleming 2000,
Hendry et al. 1998). During and following reproduction, eggs and larvae have
developmental requirements, and growth and survival are inﬂuenced by many physical
biotic factors including water temperature, river discharge, and predator abundance
(Bailey and Houde 1989; Chotkowski and Marsden 1999; Mion et al. 1998; Smith and
Marsden 2009). The strength of associations between exogenous cues and expression of
behavioral responses will be dictated by the timing and location of selection (i.e., adults
or gametes/offspring) during the migration continuum.

Nearly 300 ﬁsh species exhibit migratory behavior culminating in spawning in
stream habitats (Northecote 1978; Frank and Leggett 1985). Migration for purposes of
reproduction in ﬁshes is mediated by factors associated with risks of mortality during
migration to and from spawning areas and beneﬁts of placing fertilized gametes and

offspring in high-quality incubation and nursery areas (Alexander 1998). The chronology

of migration is generally deﬁned by stages deﬁned as river entry, upstream movement,
and the initiation of reproduction (Beall and de Gaudemar 1999; Bruch and Binkowski
2002; McMicheal et a1 2005). Examples of exogenous variables found to be associated
with stages in this continuum in ﬁshes include tidal height (Karppinen et al 2004),
photoperiod (Quinn and Adams 1996), ambient light (Bizzotto et a1 2009; Parsley et al
2008), river discharge (Keefer et al. 2009; Overton and Rulifson 2009), turbidity
(Rakowitz et al. 2008), river water temperature (Binder and McDonald 2008; Salinger
and Anderson 2006; Workman et al 2002), the lunar cycle (Kuparinen et al 2009;
Bizzotto et al 2009) and social/demographic inﬂuences associated with spawner
abundance and sex ratio (Kiﬂawi and Mazeroll 2006; Couzin et al. 2005).

Responses of migratory ﬁshes during different stages of migration to exogenous
cues are not always concordant (e.g., Dahl et al. 2004; Quinn and Adams 1996). Sex-
speciﬁc responses to exogenous cues may occur due to contrasting reproductive
strategies utilized by males and females to increase reproductive success (Bateman 1948)
but are rarely documented (Ball and Ketterson 2008). As a consequence of this
uncertainty, our ability to generalize the magnitude of ecological/evolutionary responses
of migratory ﬁsh populations to exogenous cues and to predict the impacts of
environmental or anthropogenic induced change is limited

Information regarding the importance of different exogenous cues to the initiation
of behaviors at sequential stages of migration is lacking for many species. This is
especially true for long-lived iteroparous migratory ﬁshes and for broadcast spawning
species that do not exhibit post-ovulatory parental care where tight linkages between

responses that have ﬁtness consequences and environmental cues that elicit them may be

expected. Inference may also be confounded by temporally correlated observations, both
of dependent and independent variables (Thorstad et al. 2008).

We used data from a long-term research program collected over a range of
exogenous conditions and integrated novel methods of statistical inference to quantify the
relative role of environmental and lunar covariates (collectively referred to as exogenous
cues or variables) during the stages of migration in lake sturgeon (Aczpenserﬁxlvescens),
a long-lived highly migratory iteroparous ﬁsh species. The speciﬁc objectives of this
study were to 1) describe inter-annual, intra-annual and die] variation in the onset of
migration at the river mouth and quantify associations between movement with
exogenous variables, 2) estimate variation in pre-spawn river residency and determine
whether inter-annual variation in the interval between ﬁrst entry and ﬁrst arrival to
spawning areas can be tied to river conditions, 3) describe inter- and infra-annual
variation in the onset of arrival to spawning locations, quantify associations between
arrival with exogenous conditions, and test for evidence of sex-speciﬁc responses.

METHODS
Study site

Migratory and reproductive activities of a population of lake sturgeon in the
northern lower peninsula of Michigan were monitored over eight consecutive years
(2001-2008). This population is of moderate size (approximately 1,100 sexually mature
adults, upper 95% conﬁdence limit) and is isolated ﬁ'om other water bodies by dams to a
large inland lake (Black Lake ~3500 ha). Migration and reproduction occurs in the spring
(late April through May) and is restricted to the Upper Black River (UBR) (Smith and

Baker 2005; Baker and Borgeson 1999), a shallow fourth order tributary of Black Lake

(Smith and King 2005). Approximately 11 km of the UBR located downstream from
Kleber Darn (ca. 1949) to the river mouth is composed of unrestricted habitat used for
staging and spawning each year. Fish spawn within a shallow and wadable 1.5 km section
of the UBR 9 km upstream from the river mouth (Smith and Baker 2005). These
characteristics allow unrestricted access to adults. Based on genetic determination of
parentage, approximately 90% of spawning adults are sampled entering the spawning
areas in many years (Duong et al. in review). Large sample sizes permit quantiﬁcation of
inter and intra-annual variation in rrrigratory and spawning behavior and quantiﬁcation of
relationships between behaviors observed during different stages of migration and
exogenous variables.
Observations of river entry

Movements of lake sturgeon into the mouth of UBR were quantiﬁed over three
consecutive years using a weir which restricted passage to a narrow channel of the stream
and past a video-surveillance system. The speciﬁc dates for weir set-up and thus the
beginning of sampling were 4/22, 4/15 and 4/16 from 2001-2002, respectively. Gill nets
were also set at the river mouth in 2000 beginning on 4/17. The weir was constructed of
aluminum fencing and metal poles arranged in a V-shaped conﬁguration that guided
migrants through an opening approximately 3 m wide. A sheet of white painted
aluminum was placed on the stream bottom to enhance ﬁsh visibility as they passed
through the weir. A video recorder was placed overhead at the weir opening to record
images of passing individuals. Halogen lights also illuminated the weir opening at night
allowing for continuous 24 hour observation The video-surveillance system included a

Sony Digital Handycam (Model DCR-TRV720) attached to a Panasonic real motion

cassette recorder (Model RT600A). The time-lapse option was used to condense a 24-
hour sampling period onto a standard l60-minute VHS tape. Numbers of migrants were
summed by hour for each day within a season.
Upstream movement and observations of arrival to spawning areas

Adult lake sturgeon were captured as they arrived or shortly thereafter while
spawning at upstream locations over eight consecutive years (2001-2008). Trapezoid-
shaped handled landing nets (78.7 cm base, 30.5 cm top, and 70.5 cm length) were used
for sampling spawning adults from early April until mid-June by walking the entire
spawning area between 1000 — 1700 h each day. Sex was determined at the time of
collection by cloacal morphology or by physically inducing gamete emission (Bruch et
al. 2001). Characteristics of spawning UBR lake sturgeon including size, age and sex
ratios as well as other aspects of spawning and migratory behaviors have been described
previously (Smith and Baker 2005). Simultaneous observations of movement at stages 1
(i.e., river entry) and 3 (i.e., spawning site arrival) were provided in 2001 and 2002. We
use this data to estimate the duration of pre-spawning river residency for early migrants
(Stage 2) and determine whether inter-annual variation in the interval (i.e., lag time)
between ﬁrst entry and ﬁrst arrival to spawning areas is a function of river conditions.

Explanatory variables

Exogenous variables explored in this study were selected based on commonalities
found in numerous ecological studies of ﬁsh migration including sturgeons and based on
the likelihood of their inﬂuence on the timing of migratory and spawning behavior of
lake sturgeon in the UBR system Those selected encompassed three general categories

including 1) time-related variables (i.e., seasonal and diel photoperiodism; see, e.g.,

10

Doroshov et el. 1997; Webb et al. 2001; Smith and Smith 1997; Binder and McDonald
2008; Baril and Magnan 2002), environmental stream variables such as water
temperature and river discharge (see, e.g., J onsson 1999; Salinger and Anderson 2006;
Cooke et al. 2004; Kemp et a1. 2006) and the lunar phase (i.e., lunar illumination; see,

e.g., Kuparinen et a1. 2009).
River water temperature was monitored hourly using HOBO® data loggers (Onset

Computer Corp.) placed at the river mouth. A United States Geological Survey (USGS)
gauging station was removed from the UBR in 2002. However, estimated UBR discharge
(2001-2008) was based on comparisons between long-term (i.e., 50 years) data from
UBR prior to gauge removal and from the Pigeon River which still has a gauge, a nearby
tributary of Mullet Lake, MI (Lat 45° 09' N, Long 84° 28' W). Discharge data for
comparisons were obtained online from the USGS National Streamﬂow Information
Program (http://water.usgs.gov/nsip). Data on lunar phase expressed as the fraction of the
moon illuminated for Eastern Standard Time was obtained online from the Astronomical
Applications Department of the US. Naval Observatory
(http://aa.usno.navy.mil/data/docs/IVIoonFraction). Actual light levels, moonrise or set
times or whether the moon was visible on a given night were not taken
STATISTICAL ANALYSES

Diurnal patterns of movement into the river mouth were analyzed using a linear
mixed model developed within the MIXED procedure of SAS (version 9.1, SAS Institute,
Cary, NC). Hourly movement was modeled with the ﬁxed effects of hour and year. A
multivariate ﬁrst-order auto-regressive covariance matrix was used to account for

correlations of observations of individuals across hour of the day. Modeling error as an

11

auto-regressive process was appropriate because observations of hourly migratory
activity which are more proximate in time were believed to be more highly correlated
than more temporally separated observations. Post-hoc pair-wise corrrparisons of least
square means were performed for each hour and nominal alpha levels were adjusted for
multiple comparisons using the Tukey-Kramer method.

We used a hierarchal generalized linear mixed modeling (GLMM) approach to
quantify the effect of exogenous variables on temporal variation in daily numbers of
sturgeon observed initiating migration and entering the UBR (i.e., dependant variable)
and the daily number of sturgeon captured as they arrived to spawning locations and
initiated spawning activity (i.e., dependant variable). Daily activity was analyzed as a
Poisson random process to account for heterogeneity of variances and non-normal
residual variation resulting from a strong positive skew with many days each year having
zero or a low number of observations or captures.

Correlations between observations, which tend to have the most serious impact on
statistical interpretation of migratory data (Tre'panier 1996 and Erkinario et al. 1999),
were accounted for using a low-rank radial smoothing algorithm applied through time
(i.e., days). Radial smoothing is a non-linear non-parametric procedure that can be
applied within the context of a parametric linear model with parameters of ﬁxed and
random components (see Gurrin et a1. 2005). Like other methods that account for
temporal autocorrelation (e.g., autoregressive models), smoothing algorithms generally
provide more accurate predictions across the range of linear covariates (Pyper and
Petennan 1998). Importantly, accounting for autocorrelation by using smoothing

algorithms also provide more precise and robust estimates of model B’s by reducing

12

irregularities (random ﬂuctuations) in the dependant variable of time series data (Paige et
al. 2009; Verbyla et al. 1999; Wang 1998). Radial smoothing algorithms were applied

though time as a random effect using the GLIMMIX procedure (SAS Institute, Cary,

NC). Terms for year (B1) and the day (By) samples were collected were also incorporated

into all statistical models. Year was treated as a random effect. The addition of a
chronically ordering day component as a ﬁxed effect provided a measure of correction
for expected intra-annual declines in spawner abundance.
Environmental eﬂect models

The behaviors of migratory animals are often dependent on covariates
experienced during previous periods (e.g., Sherril-Mix 2008). Therefore, lagged
environmental variables, estimated over a period of 1-2 days prior to daily observations,
were used to assess the inﬂuence of environmental conditions on the initiation of

migration and arrival to spawning grounds in the UBR. These variables included: 24 hr

lagged temperature (Tlag, B3), 24 hr lagged discharge (Dlag, B4), 24 hr change in

temperature (Tchange, B5) and 24 hr change in discharge (Dchange, B6). Changes in

water temperature and discharge were computed as the average difference between mean
observations collected 48 hrs apart, reﬂecting direction (positive vs. negative) and
magnitude of trends. Note that lagged variables were also highly and signiﬁcantly
correlated (r range = 0.90-0.98) with conditions at the time of observation or capture
(e. g., average daily water temperature).

Two models of increasing complexity were ﬁt to count data for the number of

individuals entering the river mouth and arriving at spawning grounds. The ﬁrst model

13

included only environmental covariates. A second model included both environmental
variables and autocorrelation parameters (i.e., smoothing) to illustrate the importance of
accounting for temporal trends. We refer to these models as Model 1- 2 (initiation of
migration) and 6-7 (arrival at spawning areas), respectively. Environmental models

generally took the form:
Y = u + B1 Year + B2 Day + B3 Tlag +B4 Dlag + B5 Tchange + B6 Dchange,

where Y is the number of lake sturgeon observed or captured each year on a given day, 11
is the intercept and the B’s are parameters for each dependent variable to be estimated.
Collinearity among independent variables was not a problem due to low correlation (r <
0.3). Preliminary analysis also indicated that daily arrival to spawning sites may be
quadratic (i.e., curvilinear with an intermediate extreme) ﬁmction of water temperature.
Thus, an additional quadratic term for lagged effects of water temperature was added to
these models. However, since lagged temperature and its quadratic effect were highly
correlated, both variables were centered by subtracting the mean of each variable from
each of its values and only the transformed variable of best ﬁt was used in the ﬁnal model
(see below).

Lunar eﬂect model

The effect of lunar periodicity was initially assessed separately using several

terms (i.e., sin 9, B7; cos 9, B3; sin 29, B9) that treated the lunar cycle as continuous

cyclical rather than discretized events (see deBruyn and Meeuwig 2001). Here, 9 was the
angular-transformed day of the lunar cycle (day/29 X 21:). A signiﬁcant cos 6 regression

coefﬁcient represents a peak occurring at full moon and new moon whereas a signiﬁcant
sin 6 regression coefﬁcient represents a peak at the ﬁrst and last quarter. The strength of

14

association and the direction of association between numbers observed and lunar
variables (i.e., full vs. new moon) depends on the sign of the coefﬁcient. We also
included the sin 29 coefﬁcient due to evidence of multiple spawning peaks per lunar
cycle (i.e., sin 20 test for peaks at the ﬁrst and last quarter as well as a full or new moon).
Each relationship describes how strongly the time series of observations tended to covary
with lunar ﬁmctions. Two models of increasing complexity were again developed for
both the daily number of individuals entering the river mouth and arriving to spawning
grounds and included: a lunar effect only model and a model with autocorrelation plus
lunar covariates. We refer to these models as 3-4 (river entry and onset of migration) and
8-9 (arrival at spawning areas and initiation of spawning activity), respectively. The full

lunar model was generally:
Y=u+B1 Year+B2Day+B7sin0+B3 cosB+B95ine 29,

where Y is the number of lake sturgeon observed or captured each year on a given day, 11
is the intercept and B’s are the parameters for each dependent variable to be estimated-
Model selection

Detecting the best ﬁtting and most parsimonious model is an inrportant analytical
step. However, the estimation methods employed by GLIMMIX (e.g., REML) do not
allow for multivariate model selection procedures such as forward, backward or stepwise
selection criteria. Thus, predictive variables ﬁ'om both environmental and lunar models
(EL) were selected based on level of signiﬁcance (P S 0.25) and then combined in
reduced and parsimonious multi-parameter environmental-lunar models. All possible
two-way interactions between selected environmental and lunar variables (e. g., Tlag X

Dlag; Tchange x Dchange; Tchange X cosine) were evaluated during this last model

15

selection phase. All interactions were included and sequentially removed (i.e., model was
re-run each time) based on levels of statistical conﬁdence (P > 0.05). An interaction was
removed if it contributed little to model ﬁt (see below). In our assessment of the number
of individuals arriving to spawning locations, we also tested for sex-based responses to
both environmental and lunar effects within in the environmental lunar model by
including two-way (e. g., sex X Dlag) and three-way interactions (e.g., sex X Dlag X
cosine). A sex X day interaction was also included to test and account for expected non-
random sex-based declines in spawner abundance.

All models were compared by examining residuals, ﬁtted values and parameter
estimates as well as a linear comparison of observed and predicted values (i.e., R2). The

ratio of the Pearson Chi-Square over degrees of ﬁ'eedom was also used to compare
information on residual model variability (The GLIMMIX Procedure, SAS Institute,
Cary, NC). A ratio close to 1 indicates that variability has been properly modeled and that
there is little remaining residual over-dispersion in the data. Other commonly used model
ﬁt statistics (e. g., AIC or BIC; Burnham and Anderson 2002) were not used because
GLMMs ﬁt using the GLIMMIX procedure in SAS estimates model parameters by
applying pseudo-likelihood techniques and also provide unreliable information when
models incorporate a random component (i.e., year and smoothing algorithms).
RESULTS
Summary of river entry

The ﬁrst individual observed at the river mouth deﬁning the onset of annual

migration was extremely consistent with respect to calendar date and thus seasonal

photoperiodism (23 April in 2000; 24 April in 2001; 23 April in 2002) but less so with

16

river water temperature (9.6 0C in 2000; 10.8 0C in 2001; 8.9 0C in 2002) and river

discharge (7.8 m3/s in 2000; 11.8 m3/s in 2001 and 9.9 m3/s in 2002). However, inter-

annual consistency in temperature and discharge at migration onset was still notable

considering strong intra-annual trends as well as ranges in temperature (range: 8.6 — 21.0

0C), and discharge (range: 5.4-23.3 m3/s) observed over the whole season (Figures 1.1

and 1.2). The onset of migration did not coincide with percentage of lunar illumination
(Range = 0-82%). Migration into the UBR lasted for a maximum of 36 days from April
to late May (29 days in 2000; 32 days in 2001 and 36 days in 2002). An average of 8
individuals moved upstream daily. However, as many as 55 individuals were observed

moving upstream over a single 24 hour period. Upstream migration occurred at mean
daily water temperature and river discharge ranging from 8-210C and 5-24 m3/s,

respectively. Inna-annual variation in the number of individuals entering the river was
characterized by discemable peaks that differed in magnitude and duration (Figure 1.4).
Exogenous correlates of river entry
A total of 757 recordings were made of lake sturgeon initiating migration (264 in

2000; 258 in 2001 and 235 in 2002). Counts of individuals entering the river were

signiﬁcantly associated with hour of the day (F 23, 46 = 31.3; P < 0.001) and this behavior

was consistent across years (F 2, 46 = 1.15; P = 0.33; Figure 1.3). Individual pair-wise

comparison across 24 hours of the day (i.e., hour 1 vs. hour 2) generally showed that
signiﬁcantly (P < 0.05) more individuals initiated upstream migrations in the early

morning (0600 though 1100) and early evening hours (1700 through 2200) compared to

17

other times of the day (Figure 1.3) suggesting a strong inﬂuence of diel photoperiodicity
on upstream movement. Pair-wise comparisons of three arbitrary categories established
based on relative levels of ambient light also show that signiﬁcantly (P < 0.001) fewer
lake sturgeon initiated upstream migration during the night ﬁ‘om 2200 to 0500 hrs (0.8
individuals 3: 0.7 SE) and during the day from 1000 to 1700 hrs (6.4 individuals :h 2.6 SE)
relative to twilight hours from 1800 to 2100 and 0600 to 0900 (27.8 individuals :1: 10.2
SE), respectively.

Environmental (i.e., Models 1-2) and lunar (i.e., Models 3-4) nrixed models
explained “-77% of the variation in numbers of sturgeon entering the river (Table 1.1).
However, environmental models that accounted for temporal autocorrelation in
observations of movement explained a relatively greater proportion of the variance. All
factors including water temperature and river discharge were signiﬁcant predictors of
daily numbers of individuals entering the river (Table 1.2). Variables characterizing the
lunar cycle were not signiﬁcant (1.2). However, a model with environmental and lunar
effects including the ﬁxed effects of time (day), lagged river discharge, lagged water
temperature, change in water temperature, change in river discharge and sine of lunar
angle explained the greatest proportion of the variance (93%) in daily migratory activity.
Model predictions and 95% conﬁdence intervals were also highly concordant with
observations (Figure 1.4). Two-way interactions between environmental and lunar
variables were non-signiﬁcant (data not shown) and were excluded from the analysis.

Regression estimates for variables included in the best-ﬁtting environmental-lunar
model as well as measures of uncertainty in model estimates (95% CI) (Table 1.4)

revealed that the number of individuals entering the river did not signiﬁcantly decrease

18

during the season (F 1, 12 = 0.01, P = 0.93). However, daily numbers of individuals
entering the river was signiﬁcantly and positively associated with lagged water

temperature (F 1, 30 = 6.75, P = 0.01), increasing by approximately 20% (95% CI: 4-39%;

Figure 1.5) for every unit increase (0C). Lagged river discharge as well as the magnitude

and direction of change in river discharge were also signiﬁcant predictors of daily
movements at the river mouth (D-lag, F1, 90 = 4.37, P = 0.04, Figure 1.5; Dchange, F130

= 3.86; P = 0.05). However, in contrast to water temperature, daily number of individuals

entering the river was predicted to decline by 7% (95% CI: 1- 12%; Figure 1.5) and 8%
(95% CI: 2 -12 %; Figure 1.5), respectively for every unit (m3/s) increase in daily UBR

discharge. Migratory activity was unrelated to the sine of lunar angle although model ﬁt
decreased substantially if this variable was removed
Summary of upstream movement and arrival to spawning sites

We recorded 1262 captures of adult lake sturgeon from 2001-2008 including
repeat observations of the same individuals over multiple years (Chapter 2). The total
number of individuals captured per season ranged from 104-226, and more observations
of males were made seasonally and throughout the study relative to females (360
females; 902 males). In years when both river entry and arrival to spawning areas were
monitored simultaneously (i.e., 2001 and 2002), variable lags between the ﬁrst individual
observed entering the UBR and the ﬁrst adult arriving to spawning locations were found

(4 days in 2001 and 14 days in 2002; Figure 1.6). Further, UBR water temperature was

2°C warmer on average and increasing faster in 2001 relative to 2002 before spawning

l9

and river discharge was also higher in 2002 on average during this time (Figure 1.1). We
also noted based on this comparison that numbers of individuals arriving to spawning
areas were more cohesive (i.e., well deﬁned groups) relative to movement at the river
mouth (Figure 1.6).

The number of individuals arriving to spawning areas and initiation of spawning
activity lasted 29 days on average (range 19-43 days) ﬁ'om 2001-2008, generally
encompassing a full lunar cycle (Figure 1.7). Six individuals were captured each day on
average. However, the number of newly arriving individuals daily ranged from 0-50

individuals (Figure 1.7). Spawning occurred at water temperatures ranging ﬁ'om 8 to 19

0C but activity peaked (i..e, quadratic) when temperatures ranged from 11-17 0C.
. . . . . 3
Spawmng actrvrty was also highest when river ﬂow volume ranged ﬁom (5-9 m /s) and

declined precipitously on days when daily river discharge exceeded 9-10 m3/s. Seasonal

patterns in upstream spawning activity were also complex and generally characterized by
2 (e.g., 2002, 2003 and 2005), 3 (e.g., 2004 and 2007) or 4 (e.g., 2001, 2006 and 2008)
distinct peaks that differed in occurrence with respect to calendar date, magnitude and
duration (Figure 1.7), though multiple captures of the same individual over multiple years
revealed that the timing of spawning was highly repeatable (Chapter 2).

The onset of arrival at spawning sites (i.e., ﬁrst individual observed) was highly

variable across years with respect to calendar date (April 20 — May 7) and river water

terrrperature (8-14.9 oC). However, river discharge at this time was generally low (8.1-

12.7 m is) relatrve to seasonal ranges. The onset of spang srte arrival did not corncrde

with a consistent percentage of lunar illumination (Range 2-87 %). However, the day of

20

ﬁrst arrival consistently occurred after the ﬁrst three lunar cycles since the beginning of
the year. Further, years when spawning site arrival was most delayed (e.g., 2002 and

2005) were generally years when the fourth full moon occurred at a later day of the
calendar year although this relationship was not found to be signiﬁcant (F 1, 7 = 2.5; P =

0.162). There was no evidence for consistent sex-based differences in ﬁrst arrival to
spawning locations (i.e., protandry) noted in other studies of lake sturgeon reproduction
and as such spawning behavior was generally observed to begin within hours to days
after arrival (Forsythe, unpubl.) based on observations of reproductive behaviors such as
male cruising of spawning areas, porpoising, nudging and tail beats (described in Bruch
and Binkowski 2002).
Exogenous correlates of spawning site arrival

Environmental (i.e., Models 6-7) and lunar (i.e., Models 8-9) models explained 2-
66% of the variation in daily spawning activity in the UBR over eight years (Table 1.1).
However, a model with lunar variables only that accounted for temporal dependency in
catch of adults explained a relatively greater proportion of the variance relative to models
the environmental effects of temperature and discharge (Table 1.3). Further, coefﬁcients
for both the cosine and 25ine of the lunar angle showed were signiﬁcant suggesting a role
of cyclic lunar conditions in affecting intra-annual variation in spawning activity of UBR
lake sturgeon. However, a model with both lunar and environmental variables best
explained our observations of spawning (Table 1.1) and in most cases was highly
predictive of captures within and across years (Figure 1.7). This environmental-lunar
model of best ﬁt included the ﬁxed effects of day, sex, lagged discharge, rate of change

in water temperature, the quadratic effect of lagged temperature, cosine and 25ine of the

21

lunar angle. Two-way interactions between time and sex as well as the interaction
between lagged discharge and water temperature (quadratic effect and the rate of change)
with cosine 0 were also included based on their contribution to model ﬁt (Table 1.4).
Regression estimates for all variables as well as measures of uncertainty in
parameter estimates (95% CI) (Table 1.4) provide strong evidence for declines in
numbers of individuals arriving at spawning areas thus smaller sizes of spawning

aggregations later in the spawning season. However, rates of decline in spawner
abundance differed signiﬁcantly bysex (F 1, 12 = 23.89, P < 0.0001), being signiﬁcantly

faster for females than males. Speciﬁcally, female abundance declined by 4% on average
(2-6 % CI) each day the season progressed whereas the number of newly arriving males
at spawning locations remained unchanged Numbers of both males and females arriving

to spawning sites were found to be equally (i.e., no sex-based interaction) and
signiﬁcantly associated with 2 sine of the lunar angle (F 1, 453 = 54.5, P < 0.0001), a

function that suggests peaks in arrival in association with the ﬁrst and last quarters of the
lunar cycle as well as at a full or new moon (Figure 1.7). However, a negative parameter
estimate for sine 20 (Table 1.4) indicates a stronger association with a quarter moon, or
roughly 25% lunar illumination.

The numbers of male and female lake sturgeon arriving to upstream spawning

areas were equally predicted (i.e., no sex-based interaction) by the main effects of cosine

of the lunar angle (F 1, 453 = 10.7, P = 0.0012), lagged discharge (F 1, 453 = 28.4; P <
0.0001), rate of change in water temperature (F 1, 453 = 57.7; P < 0.0001) and the

quadratic function of lagged water temperature (F 1, 453 = 0.02, P = 0.015) (Table 1.4).

22

However, highly signiﬁcant two-way interactions between these lunar and environmental
variables (Table 1.4; P < 0.0001) indicate that the responses of individuals in terms of

spawning behaviors were not independent. Yet, spawning activity expressed as a frmction
of all two-way interactions was consistent with the sign of individual parameter estimates

(Table 1.4), generally highlighting the trend of increased numbers of spawners arriving to

spawnrng srtes at times wrth lower rrver discharge (7-15 m /s; Figure 1.8 a), when water

temperature approached an intermediate level within its range (12-16 0C; Figure 1.8 b)

and at times when water temperature was increasing at a faster rate to this “optimum”
(Figure 1.8 c), especially when the cosine of lunar angle approached 1 or a ﬁrll or new
moon (Figures 1.8 a,b,c).
DISCUSSION

Organisms often use environmental cues to make decisions that weigh heavily in
probabilities of offspring survival. Our results revealed that exogenous environmental
and lunar covariates affected lake sturgeon behavior associated with initiation of
migration, movements in the stream during migration, and timing of arrival at spawning
areas leading up to the initiation of spawning.

River entry

The onset river entry in the UBR occurred at nearly the same Julian date (range of

two days over three years). Migration was rmtrated wrthrn a range of 1.9 C mean water

3 . . .
temperature, a range of 4 m /s mean rrver discharge but was unrelated to a phase in the

lunar cycle. Data collected during this period suggest that time-related variables (i.e., day

length in the form of seasonal photoperiodism) represent an important exogenous cue for

23

UBR lake sturgeon to initiate migration. Day length has been shown to mediate the
endogenous cycle of oocyte maturation in other migratory ﬁshes (Lee et al. 2002;
Taranger et al. 1998) including sturgeon (Doroshov et al. 1997; Webb et al. 2001) and
this may explain the remarkable inter-annual consistency in the beginning of upstream
movement. This result, however, is in contrast to other migratory ﬁshes where migration
occurs at speciﬁc temperatures regardless of the date and thus is considerably more
variable across years and between early and late spawning periods (Dahl et al. 2004;
Juanes et al. 2004). Gessner et al. (2007) also noted latitudinal gradients in the timing of
migration in other sturgeons (i.e., Green and Atlantic) at a spatial scale where water
temperature, spring discharge and photoperiodism vary greatly. We also noted earlier
arrival to spawning areas (i.e., April 20) suggesting that individuals may move into the
UBR earlier, especially in years with abnormally high water temperature (i.e., 2006;
Figure 1.1). Thus, while day length may be the strongest proximate driver in the onset of
river entry, the onset of migration for UBR sturgeon indicated by river entry may have a
degree of behavioral plasticity.

Time of day was also a signiﬁcant predictor of river entry. Signiﬁcantly more
individuals entered the river at dawn and dusk relative to other times of the day (Figure
1.3). Crepuscular activity may represent a learned behavioral response to predator (and
human) activity rather than a strict environmentally mediated behavior due to the small
and shallow river environment. Alternatively, crepuscular activity may also allow UBR
lake sturgeon to avoid the thermal stress of migrating during rapidly cooling evening
water temperatures, which is considerable in small northern temperature systems like the

UBR, while avoiding sources of mortality through reduced visibility. Warmer water

24

temperatures, such as those occurring later in the spawning season, shown to shift
migratory movements in some migratory ﬁshes away from nocturnal conditions in favor
of increased swimming endurance in laboratory experimental settings supports this
hypothesis (Binder and McDonald 2008).

Environmental models that accounted for temporal autocorrelation in observations
of daily river entry explained a large proportion (77%) of the variation in daily counts of
sturgeon entering the river and considerably more than comparable models with only
lunar effects (36%). In general, the strong negative effect of river discharge (Figure 1.5)
and positive effect of water temperature (Figure 1.5) on the number of individuals
entering the river is likely due to physiological effects of both factors on swimming
performance, especially given the poor swimming ability of sturgeons relative more to
fusiform species such as salmonids (Webb 1986; Singer et al. 1990). However, UBR
sturgeon may also respond to conditions at the river mouth as a mechanism of directly
increasing offspring survival. For example, Hare and Cowen (1997) and Rideout et al.
(2005) show that by avoiding stressful conditions (i.e., high temperature, high discharge)
migratory females can divert extra blood flow to ovaries, positively affecting the size and
quality of eggs, characteristics often related to increased survival during the egg and
larval stages. We expand on the speciﬁc effects of discharge and temperature in the next
few paragraphs.

Comparable studies of sturgeon migration over shorter time periods are
concordant with our results with respect to river discharge (Benson et al. 2007), while
numerous others have shown positive associations with river ﬂow (Paragamian and

Kruse 2001; Chapman and Carr 1995). One possibility for this difference is that studies

25

based on telemetry collect data on fewer individuals and may not detect the tendency for
individuals to select energetically eﬁicient conditions (Crossin et al. 2003) or discharge
may have a greater inﬂuence on migratory behavior in smaller and shallower systems.
Differences in the response of sturgeon to river discharge are a reﬂection of species and
location speciﬁc migration strategies as found in well-studied salmon and trout (reviewed
in Hinch and Rand 1998). For example, in systems where spawning sites are relatively
close (9 km) and few natural barriers exist to impede movement may favor a migratory
strategy that conserves energy to support current reproduction. Individuals using systems
which require considerably longer migration distances, however (e. g., Kooteni River, 247
km), may elect to swim outside of optimal conditions to conserve time in order to best
coordinate arrival with gonad maturation (i.e., occurring in route) and spawning activity
with conditions (i.e., mate type and number, water temperature) that increase
probabilities of reproductive success (Standen et al. 2004).

Positive monotonic relationships between daily numbers of individuals entering
rivers and water temperature are also common in migratory ﬁshes including sturgeons
(Dahl et al. 2004; Tetzlaff et al. 2005; Paragamian et al. 2002) as a ﬁrnction of swimming
performance. Alternatively, other studies demonstrate that maximal. sustainable speeds of
migratory ﬁshes decline sharply as water temperature increases above a thermal optimum
and thus migration activity slows as individuals seek to conserve energy and in some
cases reduce the risk of pre-spawn mortality (Hammer 1995; Mayﬁeld and Cech 2004).

For example, Salinger and Anderson (2006) show that swim speed in Chinook salmon

(Oncorhynchus tslrawytscha) increased below 16°C but decreased above this optimum.

We tested this hypothesis but found no statistical evidence of a preference for UBR

26

sturgeon to migrate at a speciﬁc water temperatures (i.e., no quadratic effect); rather,
migratory activity increased linearly until 21°C (Figure 1.5). One explanation for this

result is that individuals were avoiding stressful high-temperature conditions altogether
and thus a decline in numbers with increasing water temperature was not detectable. Lake

sturgeon have also demonstrated enhanced, sustained and prolonged swimming

performance at higher water temperatures without reaching an optimum (i.e., up to 210C)

(Peake et a1. 1997). Alternatively, warmer water temperatures may be a strong cue for
individuals at the point of river entry for reasons that enhance embryo development,
especially in systems such as the UBR where migration distances are relatively close.

A model that incorporated both environmental and lunar effect was the best ﬁt to

our data (R2 = 0.94) and was predictive of substantial intra- and inter-annual variation

observed in the daily number of individuals in the UBR (Figure 1.4). This ﬁnding
illustrates the relative importance of several features of the data. First, the improvement
of model ﬁt with addition of a single lunar parameter (i.e., sine of the lunar angle)
indicates a weak but potentially important lunar effect that three years of video-
surveillance may not have been sufﬁcient to signiﬁcantly elucidate, even with the
integration of statistical methods that can generally expose weak associations (deBruyn
and Meeuwig 2001).Assuming that additional years of data would reveal a signiﬁcant
lunar effect, lunar-induced movement may be adaptive for sturgeons in reducing
predation-driven mortality or in facilitating the passage of individuals to spawning
locations purported in other migratory ﬁshes where this phenomenon is observed

(Carlsson et a1. 2004; Wilson and Halupka 1995). However, since there is strong

27

correlation between the timing of spawning with speciﬁc phases in the lunar cycle (i.e.,
full and new moon) for reasons later discussed, a more likely explanation for a weak
lunar effect is that the association between causal and response variables at sequences in
the migratory continuum spatially and temporally removed ﬁom the agent of selection
would be weaker. Recent studies demonstrating signiﬁcant evidence of weak lunar effect
at the time of river entry in environments where its indirect effects are small like the UBR
(i.e., tides), lends support for this hypothesis and suggest that lunar-based migration is of
greater biological relevance than previously discussed (Kuparinen et a1 2009).

We tested for the combined effects of discharge and water temperature on river
entry by testing for interactions in the full lunar-environmental model because
interactions between years are often notable (Prigel and Wirth 1997; Quinn et al. 1997;
Keefer et al. 2004). However, interactions between water temperature and discharge
variables on movement upstream within a season were either non-signiﬁcant or
contributed little to model ﬁt suggesting that times with either warmer water temperatures
or decreasing river ﬂow were equally strong cues for individuals to initiate river entry
(i.e., the decision of individuals to depart at times with high temperature was independent
of river ﬂow). This result contrasts the others that have tested for interactions. For
example, Svendsen et al. (2004) found that while the probability of anadromous female
brown trout (Salmo trutta) to move upstream signiﬁcantly increased with increasing
discharge, this probability was differentially affected by water temperature. However, the
relative effect of water temperature, discharge and thus the potential for interaction on
inducing upstream movement clearly varies among systems (Jonsson 1991; Workman et

al. 2002). Further, since the UBR is small and environmental cues like temperature and

28

 

discharge vary over a shorter temporal scale, daily observations and lag effects of
conditions during periods preceding the day of observation are likely to elicit a more
immediate, independent (and detectable) response relative to larger rivers.

Finally, predictions of future environmental conditions are often based on
conditions experienced in the immediate past. In order for UBR lake sturgeon to detect
effects of conditions over previous days (lagged cues) individuals would need to be near
the source of these cues prior to the onset of migration (i.e., staging before river entry).
Staging has been reported for lake sturgeon in other larger systems (i.e., Green Bay; Bott
et al. 2005) and for numerous other migratory ﬁshes (Godinho et al. 2007; High et al.
2006). Many migratory species such as birds form pre-migratory aggregations where
conspeciﬁc social interactions represent important ones that induce migration (Lundberg
1985). Migratory ﬁsh are generally not social and yet the congregation of individuals
near river mouths before migration and evidence of repeated bouts of river entry and
withdrawl in sturgeon (Godinho et a1. 2007) may suggest that social inﬂuences are an
important component of migration in addition to the exogenous cues tested herein.

Upstream movement and arrival to spawning sites

Arrival to spawning locations was more variable with respect to calendar date
than the onset of migratory activity. Thus, time-related cues such as day length appear to
become less important after river entry in the UBR. The onset of spawning activity was
also inconsistent with lunar illumination. However, anival was delayed when the fourth

full moon occurred later in the season. Water temperature at ﬁrst documented arrival to

spawning grounds was also variable (IO-19°C) but consistent with the incubation

requirements of embryos (8—20 °C; Detlaff 1993; Van Eenennaam et al. 2005) and

29

 

similar to those reported at the onset of spawning for populations across the species
geographic range (McKinley et a1. 1998; Auer 1996; Bruch and Binkowski 2002; LaHaye
et a1 1992). Based on these similarities, we believe water temperature is a strong cue
regulating the onset of reproduction. However, arrival to spawning locations also
occurred when UBR discharge was low and decreasing. Increased lag time between the
onset of migration and arrival to spawning locations in years with higher average
discharge were also observed (2001 vs. 2002; Figure 1.6). Thus, inter-annual variation in
arrival time to spawning sites may be related to river discharge resulting from different
river residence times, as individuals stay in the river between entry and initiation of
spawning. Prigel and Wirth (1977) also noted signiﬁcant delays in spawning during
seasons of high water in the Wolf River, Wisconsin.

Counter to our expectations, models with only lunar covariates explained the
largest proportion (66%) of the observed variation in spawning activity. Times ‘
approaching a new moon, during or just after with full lunar illumination were found to
be important cues for individuals to initiate reproductive activity. Lunar cues may have
important ecological implications for sturgeon species. For instance, the incubation
duration of lake sturgeon embryos is predictable given water temperature and maternal
effects (Duong et al. 2010). Thus, synchronizing reproduction with the lunar cycle (i.e.
full moon) may be synchronized to timing of the larval dispersal (Forsythe et a1. 2010),
decreasing larval risk to nocturnal predators. Timing spawning with lunar phases
facilitates hatching and post hatching development during a period that subsequently
exposes larvae to dark lrmar phase conditions that reduce visibility during nocturnal drift

and as exogenous feeding begins (Hernandez -Leon 2008). Incubation length and the

30

duration of larval residence in the substrate prior to downstream dispersal tend to
decrease with increasing temperature as the season progresses (Crossman 2008; Duong et
al. 2010). Our study is consistent with reports indicating that spawning in both lake and
Gulf sturgeon occurred in association with a new moon (Auer and Baker 2002; Sulak and
Clugston 1998).

Models with environmental effects also explained a large proportion of the

variance in spawning site arrival (R2 = 0.63). Parameter estimates indicated that low-ﬂow

conditions (i.e., below 9 m /s) were 3 Signiﬁcant cue for rndrvrduals to arrrve at spawning

areas and initiate spawning (Table 1.3). One explanation for this relationship is that
increased river discharge is a negative cue for individuals staging downstream to move
onto spawning locations (i.e., similar to at the river mouth). Reproductive readiness of
individuals as well as critical aspects of the species’ reproductive ecology may also be
impaired or disrupted as river discharge increases (Auer 1996). Low ﬂow conditions are
likely important for increasing probabilities of egg fertilization and to increase
probabilities that eggs are deposited within desirable substrates within selected spawning
locations to increase egg survival (Auer 1996; LaHaye et a1. 1992; Paragamian and
Wakkinen 2002). However, increased abundance of reproductively active individuals
arriving to spawning sites has been found to coincide with increasing river stage for some
sturgeons (i.e., shortnose) (Tripp et al. 2009).

The number of new individuals arriving to spawning sites and initiating spawning

was best predicted by a quadratic ﬁrnction of water temperature, peaking between 12 and

16 0C, and by the rate of increase in water temperature over successive days (Table 1.4).

31

Similar qualitative associations between intra-annual variation in spawning activity and
water temperature and temperature induced behaviors have been documented in other
sturgeon species across a wide geographic range. For example, Scott and Crossman
(1973) and LaHaye (1992) reported peak spawning temperatures between 13 and 18°C.
Kempinger (1988) and Paragamian and Wakkinen (2002) noted that small declines in
water temperature (0.8 — 3°C) reduced spawning activity and caused individuals to
temporally leave spawning locations. Reproductive activity of lake sturgeon in the Des
Praries and L’Assomption Rivers also increased in association with increasing water
temperature (LaHaye et a1. 1992). However, our study is the ﬁrst to empirically test for
and quantify the relationship between spawning site arrival and temperature over multiple
spawning seasons and based on groups spawning at different times.

The lunar cycle was a relatively stronger cue for UBR lake sturgeon at the ﬁnal
stages in the migration continuum as compared to the importance of water temperature,

river, discharge or Julian date for initiation of migration at the river mouth. Yet, a model

incorporating both lunar and environmental effects best ﬁt our capture data (R2 '= 0.77)

and predictions and conﬁdence intervals were concordant with observed spawning
activity across all years (Figure 1.7), especially after accounting for temporal
autocorrelation. Parameter estimates were similar between all models (Tables 1.4 and
1.4). However, highly signiﬁcant interactions between all environmental variables with a
single lunar covariate (i.e., cosine 9) suggests that the response of individuals to
environmental variables was of secondary importance to lunar cues as adults arrived at
spawning areas and initiated spawning. Signiﬁcant interactions suggest that inter-annual

variation in timing of initiation of spawning is a reﬂection of environmentally-induced

32

modiﬁcation of behavior from times with either a full or new moon (Figures 1.8 a, b and
c). Associated work on the UBR supports this hypothesis showing that environmental
deviance is not extreme, as inter-annual variation in the timing of spawning when
adjusted to a standardized day of spawning, for individual lake sturgeon is highly
repeatable (Chapter 2).

We found little statistical support for sex-based differences in response to
exogenous conditions at the time of spawning. Sex-based responses may not be important
to initiation of spawning as observed during other stages during migration. Paragamian
and Kruse (2001) found that male white sturgeon were signiﬁcantly less likely to respond
to changes in water temperature and discharge while moving upstream than females.
Although males and females are exposed cued by the same exogenous conditions,
differences in reproductive strategies likely keep males at spawning locations longer to
enhance mating opportunities as additional females arrive at spawning sites as noted
previously (Bruch and Binkowski 2002; Van Eenennaam 1996; Douglas et al. 2009;

Carmichael et al. 1998).

33

CHAPTER 2

REPEABILTY FOR SPAWNING TIME AND SPAWNING SITE SELECTION IN
THE LAKE STURGEON

INTRODUCTION

The tendency of adults to reproduce during certain times or at speciﬁc locations is
an adaptive behavior that places offspring in ecological conditions that are conducive to
growth and survival (Norris 1993; Trippel 1998; Sheldon et al. 2003; Hipfner et al.
2010). For some animals, an important component of this adaption is long-distance
migration, often with considerable ﬁtness consequences, to habitats that prepare
individuals for reproduction (i.e., foraging areas) and areas suitable for offspring
development. In migratory species, movement during times that favor energy
conservation and reduced mortality (Rand and Hinch 1998), the capacity to perceive and
respond to environmental cues during migration so that reproductive activities are
coordinated with suitable features of the environment (Duriez et al. 2009), and the ability
to select suitable breeding areas within spatially and temporally heterogeneous
environments represent important behavioral traits (Geist and Dauble 1998).

Migratory ﬁshes that use streams for reproduction exemplify species that must
coordinate reproductive activities with speciﬁc conditions to successfully reproduce
because of high mortality of offspring during early-life stages (Bailey and Houde 1989;
Smith and Marsden 2009; Chapter 1). To achieve success, many species use
environmental features of streams (e.g., temperature, discharge, substrate size), that affect
spawning site quality (Malcolm et al. 2009), and other abiotic cues (e. g., lunar cycle) to
make choices of when and where to reproduce (Hodgson and Quinn 2002; Beall and de

Gaudemar 1999; Dahl et al. 2004; Chapter 2). Previous reproductive experience and I i

34

learning from repeated breeding events (Trippel 1998; Carr and Kaufman 2009), age/size
(Trepanier et a1. 1996; Stewart et al. 2002), alternative reproductive tactics (Morbey
2003) and social inﬂuences (Brown and Laland 2003) also affect inter- and intra-annual
spawning time and breeding site selection. For many migratory ﬁshes, the timing of
spawning and spawning site selection also have a genetic basis (Jennings et al. 1996;
Quinn et al. 2000; Stewart et al. 2002) whereby the responses of individuals to
environmental cues are regulated through endogenous cycles (i.e., gonad maturation) that
inﬂuence spawning time or to physiological (i.e., olfaction) properties that allow
individuals to ﬁnd suitable spawning habitats and thus choices are consistent over
multiple reproductive events.

Spawning time and spawning site selection may also have considerable
implications depending on the circumstances that affect them. For example, many
migratory ﬁshes are characterized by groups that reproduce at different times and often at
different locations within a single tributary with little mixing of groups (Hellawell et al.
1974; Fillatre et al. 2003; Papa et al. 2007). If environmental conditions associated with
spawning habitats affect offspring traits and ﬁtness, and if offspring share similar
tendencies, adaptive phenotypic divergence can evolve across breeding units even within
the same population, a pattern recently coined isolation by time (IBT) (Hendry and Day
2005). Examples of heritable adaptive phenotypic and life history specializations across
multi-modal distributions of temporally and spatially isolated migratory ﬁsh breeding
groups include age at ﬁrst reproduction, egg size, fecundity and embryonic incubation
time (Hendry and Day 2005). Co-variation between adult spawning time and juvenile

traits has also been quantiﬁed (Quinn et al. 2000, Einum and Fleming 2004).

35

Despite the signiﬁcance of both spawning time and spawning site selection in
migratory ﬁshes, little is known regarding individual variation (i.e., plasticity) in
response to exogenous cues that individuals experience (e.g., stream ﬂow, temperature,
lunar cycle, social inﬂuences) or whether trait expression remains consistent over
repeated reproductive events (i.e., has direct genetic basis). Further, knowledge of
whether individuals exhibiting similar tendencies for timing and location of reproduction
continue to associate with each other across reproductive episodes is limited but has
important implications, particularly for small populations. We used multi-year data
collected on a rrrigratory ﬁsh species, the lake sturgeon (Acipenserﬁrlvescens), to conduct
an individual-based analysis of spawning behavior. The speciﬁc objectives of this study
were to 1) estimate spawning periodicity (i.e., inter-annual spawning intervals) and
determine whether intervals varied as a function of sex and age, 2) measure the degree of
intra—annual consistency in spawning time within a breeding season and determine
whether time-dependant or exogenous covariates were the best predictors of when and
where spawning occurred within a small tributary and 3) test for evidence of size-based
(ontogenetic) variability in arrival time.

Lake sturgeon are a long-lived iteroparous species exhibiting aggregate mating
behavior where eggs and sperm are released over large areas of natal streams in the
absence of nest preparation and without post-ovulatory parental care (reviewed in
Peterson et al. 2007). Because of the species’ mating ecology and offspring exposure to
variable environmental conditions and predation, eggs and juvenile lake sturgeon often
experience high mortality early in life (Kempinger 1998, Chapter 4). Further, because

early-life survival represents a substantial component of variation in lifetime ﬁmess, there

36

is also opportunity for natural selection to direct adult behaviors, breeding site selection,
and offspring traits that are tied to environmental conditions at the time of reproduction.
Quantiﬁcation of the degree of sex and age differences in spawning behaviors, individual
behavioral plasticity, or evidence of IBT through non-random assortative mating
promoted through inter-annual consistency in spawning time and spawning location
selection has seldom been empirically evaluated in long-lived iteroparous ﬁshes.
METHODS
Study location

Research was conducted from 2001-2008 on a well-studied lake sturgeon
population located in Black Lake, Michigan The Black Lake system is ideal for
conducting long-term research directed at quantifying variability in individual spawning
behavior over several reproductive episodes. The Black Lake system is closed to
immigration because of dams. Lake sturgeon use a single tributary, the Upper Black
River (UBR), for reproduction (Chapter 1) and individuals spawn within a 1.5 km stretch
of the UBR that is approximately 9 km upstream from the mouth (Figure 2.1). After
arriving at spawning sites adults spawn in groups for periods of one to several days
(Chapter 1). During the spawning period, individuals are readily captured at several
accessible and wadable areas. Individuals can also be visually monitored nearly
continuously after capture without the use of telemetry or acoustics because of the narrow

stream channel (< 25 m), shallow water depth (< 1.5 m), and relatively low ﬂow volume

(daily average 6.4 m3/s). Due to ease of collection, a large proportion of adults have been

captured (i.e., 678 individuals of an estimated 850; E. Baker unpublished data) and most

individuals have been observed over multiple years.

37

Data collection

Lake sturgeon were captured at several locations in the UBR using large handled
landing nets while wading the entire section of the stream used for spawning _>_ 1 per day
through the entire spawning season each year. All individuals were sexed by examining
cloacal morphology or expelling gametes and biological information including total
length (cm) and weight (kg) was collected at the time of capture. Information on the
relative stream location (i.e., Zone; Figure 2.1) where individuals were ﬁrst observed
each year was also recorded. All individuals captured were marked with both internal PIT
(Passive Integrated Transponder) tags (Biomark, Inc.) and a unique combination of
external colored polyethylene T-Bar Anchor Tags (F LOY TAG, Inc.) placed at the base
of the dorsal ﬁn. Quantitative data on exogenous conditions that likely inﬂuence daily

and seasonal levels of migratory and spawning activity in lake sturgeon including water

temperature (0C), stream discharge (m3/s) and the lunar cycle phase were also collected

continuously through the spawning season (Forsythe et al. 2010). Water temperature was
recorded hourly using Onset Hobo temperature loggers. Because the gauging station on

the UBR was removed in 2002, daily river discharge was estimated using a statistical
model (F1 ,4056 = 9355, P < 0.0001 , R2 = 0.68) that predicted daily average discharge

based on long-term historical correlations (1950-2000) between the UBR and discharge
of the Pigeon River, a nearby tributary of Mullet Lake. Discharge data were obtained
online from the USGS National Streamﬂow Information Program
(http://water.usgs.gov/nsip). Daily data on the lunar phase in the form of the fraction of

the moon illuminated (values range from 0-1) were obtained online from the

38

Astronomical Applications Department of the US. Naval Observatory
(http://aa.usno.navy.mil/data/docs/MoonFraction).
STATISTICAL ANALYSES
Data preparation

Repeatability in individual spawning time was analyzed using summarized stream
environmental and lunar variables including: average daily water temperature, average
daily discharge, rate of change in average daily water temperature and discharge (lagged
over previous 48 hrs), 24 h lagged water temperature and discharge and the proportion of
the moon illuminated each day. Values resulting from the summarization of
environmental variables for each day were linked to capture and recapture histories for all
individuals in all years. Analyses of repeatability for spawning date were initially
conducted using the day of the calendar year. However, because the onset and duration of
spawning activity were variable across years, repeatability was also computed as a
function of standardized spawning date (SSD). SSD was estimated as the day of
spawning activity for a given year (y) divided by the total duration of spawning activity
(n) observed for that year with the ﬁrst day spawning was observed assigned a value of 1.
This computation placed the time of spawning on the same relative 0 to 1 scale across
years. Statistical analyses of individual repeatability in SSD and differences in
repeatability as a function of sex and body size (age) were conducted using linear mixed
or generalized linear mixed models in SAS (version 9.1, SAS Institute, Cary, NC). Mixed
effects models were generally appropriate for this analysis because of the ability to
account for repeated measurements on the same individual across spawning seasons

implemented as a random component. Mixed models can also account for alternative

39

(e.g., multinomial) distributions of the response variable, and provide a means to estimate
the variance attributable to a random factor which was also assessed at the individual
level.
Spawning periodicity

The interval between observed spawning events (i.e., spawning periodicity) was
computed as the difference in years between consecutive captures. If an individual was
captured on 2 2 occasions over the period 2001-2008, inter-spawning interval was
computed as the average interval between all successive captures. The effects of sex and
size on inter-spawning interval were incorporated into the model as ﬁxed effects.
Average spawning interval for each sex was estimated using least square means (P <
0.05).

Repeatability in spawning behavior

Repeatability was used to evaluate the degree of consistency in individual lake
sturgeon spawning behavior. Repeatability (r) is commonly deﬁned as the ratio of
between-individual variance to the total variance for a given trait or the proportion of all
variation expressed among rather than within individuals (Lessels and Boag 1987).
Repeatability estimates are equivalent to an intra-class correlation coefﬁcient (i.e., the
data are centered and scaled using a pooled mean and standard deviation for individuals)
and frequently used to describe the extent to which a behavior remains consistent over
multiple measrn'ements (Boake 1989; Farrell et al. 2009). In this study, repeatability in
spawning time was estimated for SSD along with environmental variables (e. g., current

and time-lagged river temperature and discharge) that deﬁne the stream environment at or

40

immediately before spawning, lunar phase, and spawning site location. Repeatability was

estimated using the following equation:

ug (1)

eg

A 2 A
where is Jug the estimated variance among individuals of sex g and O'eg is the

estimated variance within individuals of sex g (i.e., residual variance). Estimates of
among- and within-individual components of variance in repeatability were computed
separately for each dependent variable (i.e., time and environmental covariates) and were
obtained using restricted maximum likelihood estimation in the context of a null linear
model (i.e., intercept with no independent variables). Among and within estimates of

variance were provided as covariance parameter estimates in SAS. Consider two

. 2
hypothetical cases. If there was no variation within individuals ( 0' eg = 0), then r = 1

and all spawning observations for the same individual would occur on the same SSD. In

contrast, if no variation was present among individuals ( 0' ug = 0) and all variation

could be attributed to within-individual factors, then r = 0.

In addition to estimating variance components separately for each sex, we also
tested for evidence of heterogeneous variances among individuals of different size groups
under the alternative hypotheses that repeatability increases with increasing age and
spawning experience. Body length is correlated to and predictive of age in lake sturgeon
(Baker and Borgeson 1999; Bruch et al. 2009). We ﬁrst deﬁned three size categories for

both female and male lake sturgeon that placed approximately 33% of the total number

41

into each category. Approximately equal representation in all categories was necessary to
meet statistical assumptions. However, categories were adjusted in situations where
distinct gaps in size/age frequencies and thus levels of reproductive experience were
observed. Female categories for this analysis included individuals that were 142-164 cm
(28-42 yrs), 166-174 cm (43-48 yrs) and 175-189 cm (49-57 yrs) TL while male
categories included individuals that were 112-140cm (15-25 yrs), 141-152cm (26—30 yrs)
and 153-179 cm (3 5-52 yrs) TL. Estimates of within and between individual components
of variance were then computed separately for each dependent variable according to size

category and separately for each sex Thus, equation 1 was modiﬁed to:

X2
0' . (2)

A u r

r. = g

1 A 2 + A 2
O-ugi Jegi

A 2
where O'ugi is the estimated variance among individuals of sex g in length category i

A 2
and 0- egi is the estimated variance within individuals of sex g in length category i. The

improvement in ﬁt of this more complex model was evaluated relative to the primary
model (i.e., homogenous variances using Eq. 1) using likelihood ratio tests (P < 0.05).
Statistical evidence supporting a model with heterogeneous variances would indicate
size-based differences in repeatability with respect to the size categories created here.
Spawning site selection and size based arrival
Seasonal patterns in use of different river zones were tested using a generalized
linear mixed model that ﬁt river zone at the time of capture to standardized spawning

day. The dependent variable in this analysis (i.e., zone of capture) was modeled as having

42

a multinomial distribution with a grouped logit function which is appropriate for
categorical variables with a ﬁxed number of outcomes (i.e., six zones in this particular
case). Parameter estimates resulting ﬁ'om this model are interpreted as the odds of an
event occurring in the focal zone relative to a predeﬁned reference. Zone 1 was used as a
reference based on the observation that spawning groups tended to utilize spawning zones
further downstream as the season progressed. An odds ratio of 1 indicates that spawning
is equally likely to occur in both the zone of observed spawning and reference zone with
respect to SSD. An odds ratio greater than 1 indicated that the spawning was more likely
to occur in the observed zone relative to the reference.

Non-random seasonal patterns in SSD as a function of body size were evaluated
using a linear mixed mode] that ﬁt total length at each capture to the ﬁxed effects of SSD
along with the interaction between SSD and sex. The interaction term is biologically
interesting because it tests for evidence of heterogeneous slopes in the relationship
between total length and SSD for each sex. The effects of year and individual were
included as random effects to correct for annual variation in the number of individuals
spawning at different times, and to account for repeated measurements across time.

RESULTS
Spawning periodicity

A total of 1260 captures of 678 adult lake sturgeon were made over 8 consecutive '
years of sampling (2001-2008). Between 100 and 234 adult lake sturgeon were observed
spawning in the UBR each year. Dates of initiation and the duration of spawning activity
varied among years and spawning generally extended just over 40 days (range 19 to 43

days; Figure 2.2). However, spawning activity was not continuous during a spawning

43

season. Rather, distinct groups of males and females were observed arriving, spawning
and departing from spawning areas (Figure 2.2). The number of spawning groups and.
intergroup intervals (in days) also varied among years.

Of the total number of adults captured, 302 (44%), consisting of 70 females (26%
of total females captured) and 232 males (57% of total males captured), were observed
spawning during at least two years (Figure 2.3). The interval between spawning events

was signiﬁcantly longer for females (3.7 i 0.16 yrs, least square mean :1: SE) compared to

males (2.3 i 0.08 yrs, least square mean :1: SE) (F 1,299 = 49.5, P < 0.001). Intervals

between spawning observations for individuals ranged from 1-7 years for males and 2-7

years for females (Figure 2.4). There was no statistical evidence for an association

between spawning interval and average body size for combined data (F1 ,299 = 0.04, P =

0.84) or separately for female (F1 ,69 = 0.05, P = 0.84) or male (171,301 = 0.02, P = 0.78)

lake sturgeon

Repeatability estimates were sirrrilar for both day of calendar year and SSD (data
not shown). Because SSD facilitates comparisons among years, only results for SSD are
provided. Levels of repeatability in males and females spawning differed with respect to
SSD, environmental conditions and spawning zone (Table 2.1). However, the highest
levels of repeatability were consistent for both sexes and included standardized spawning
day, river zone, ﬁaction of lunar illumination and average river temperature, respectively
(Table 2.1). Repeatability was high for males and females despite the large inter-annual
variation in duration of spawning (Figure 2.2). Estimated repeatability of 0.56 and 0.44

for SSD in particular were high for females and males, respectively, and can be viewed as

44

the percent consistency in choice for spawning during a standardized time within a
season. Estimates of individual variance (standard error) in SSD were also relatively
small (~0.15 of the entire spawning season; Figure 2.5). Estimated SSD’s (mean i
expected SE) are reported in Figure 2.5 for all recaptured individuals. Although there was
limited support for differences in repeatability according to size categories for most
variables (i.e., homogenous variance model was the model of best ﬁt to SSD, lunar cycle
and temperature), statistical comparison of model ﬁt did support evidence for
heterogeneous between-and within-individual variances for some discharge variables
including average discharge, lagged discharge and the magnitude and the rate of change
in discharge,. Thus, computed estimates of repeatability were different for each size class
within each sex (Table 2.1).
Spawning site selection and size-based spawning

A signiﬁcant association was found between zone of spawning and standardized
spawning day for both male and female lake sturgeon (P < 0.001). In general, adults were
signiﬁcantly more likely to spawn in downstream zones during later periods of the
spawning season. Speciﬁcally, the odds of spawning in downstream zones, relative to
zone 1, increased by 2.9 (Zone 2), 4.3 (Zone 3), 18.7 (Zone 4), 108.7 (Zone 5) and 333.1
(Zone 6) for every 0.1 increase in standardized spawning day, respectively. Computed
odds were highly signiﬁcant (P < 0.001) with the exception of odds for zone 2 (P = 0.07).
Figure 2.6 illustrates the odds of spawning in each zone relative to zone 1 throughout the

reproductive season. A signiﬁcant positive association was also found between body size
(total length) and SSD (F1 ’569 = 9.02, P = 0.0028) indicating that individuals aniving

later in the season were signiﬁcantly larger than those arriving to spawn earlier.

45

However, no evidence for heterogeneous slopes between sexes was found based on a

non-signiﬁcant sex X SSD interaction (F 1,569 = 2.84; P = 0.092). Thus, predicted

seasonal increases in total length (i.e., 2.7 cm) for each standardized day were the same
for male and female UBR lake sturgeon (Figure 2.7).
DISCUSSION

Variation in numerous reproductive traits including offspring size, birth date and
female condition has been reported to vary among taxa and among populations in part as
an adaptive response to local ecological processes (Reznick and Ghalombor 2001).
Recent research has also indicated that reproductive traits vary among individuals within
the same population (Brown and Shine 2007) and most are repeatable over multiple
reproductive events despite variation in exogenous conditions (Farrell et al. 2009). We
show that reproductive behaviors in the form of intra- annual spawning time and breeding
site selection in a highly migratory ﬁsh species vary substantially among individuals
within a single population. However, reproductive behaviors, including inter-annual
spawning time are highly repeatable for individuals across reproductive episodes and are
relatively unaffected by exogenous conditions. We discuss the plausible ecological and
evolutionary mechanisms leading to repeatability and how these behaviors may or may
not be adaptive for the future of this imperiled ﬁsh species.

Inter-annual spawning time

Longevity and iteroparity are important for species with high mortality early in
life because individuals need to reproduce on multiple occasions to ensure successful
reproduction (Winemiller and Rose 1992). The inter-annual spawning interval for

iteroparous ﬁsh is highly dependent on factors that inﬂuence the duration of the

46

maturation cycle including sex (i.e., window is longer for females than males) and
environmental variation that affects the rate of gamete maturation (Pankhurst and King
2010). Estimates of means and ranges of spawning periodicity for male and female lake
sturgeon found in this study indicate signiﬁcant sex-based differences in the frequency of
and duration between spawning episodes (reﬂected in the frequency distribution; Figure
2.4). Our estimates of inter-annual spawning time Were concordant with those reported
for other lake sturgeon populations (Roussow 1957; Magnin 1966; Bruch et al. 2009) as
well as for green (Acipenser medirostris) (2-4 yrs; Erickson and Webb 2007) white
(Acipenser transmontanus) (3-5 yrs; Paragamian et al. 2005) and Atlantic sturgeon
(Acipenser oxyrinchus) (3 yrs; Caron et al. 2002). Importantly, based on our large sample
sizes and quantitative analysis with numerous environmental variables, differences in
spawning periodicity between sexes is likely determined by resource allocation that
dictate egg and sperm production and maturation schedules as opposed to variation in
environmental conditions (Doroshov et al. 1997; Taylor 1986). Further, because of the
lack of inﬂuence of environmental variables, inter-annual spawning interval is likely to
remain consistent over time and also will not vary as a function of age.

Differences in inter-annual spawning interval between males and females can be
important in long-lived iteroparous species, particularly species with polygamous mating
systems or in situations where breeding population size is low, due to lower probabilities
of encountering (and mating with) the same individuals over multiple reproductive
episodes. Quantitative estimates of inter-annual spawning interval provide probabilities
of encountering and mating with individuals ﬁ'om previous years, the demographic

composition of future spawning runs and annual spawner abundance (e.g., Waples 2002;

47

F igure 2.4). In the UBR, for example, a female spawning in year t would interact with
mature UBR males ranging in age from 15 to 52 years on average (predicted age-
frequency distribution for all UBR males collected to date). This female would
subsequently return to spawn with the highest probability (0.46) three years later based
on observed inter-annual spawning intervals (Figure 2.4). Assuming no repeatability for
SSD, this female would interact with approximately 42% (i.e., proportion of males
projected to return every year plus every three years) of the males encountered during the
ﬁrst spawning event based on expected male inter-spawning intervals (Figure 2.4).
Results ﬁ'om our study further suggests that this pattern would repeat throughout this
individual’s lifetime and would be largely unaffected by age (Table 2.1).

Intro-annual spawning time (repeatability)

Spawning time in many ﬁsh is affected by exogenous and genetic factors that are
adaptive for synchronizing reproduction with ecological conditions that promote
increased offspring survival during critical early life stages (Rogers et al. 2006; Cushing
1990). For many ﬁshes, especially broadcast spawning species with little or no post-
ovulatory parental care and high early life mortality, timing reproduction with appropriate
exogenous conditions linked to survival represents a critical life history feature (Webb
and McLay 1996; Quinn et al. 2000). In the UBR, lake sturgeon spawn over 23 days on
average (Chapter 1) in the early spring with peaks of spawning activity (Figure 2.2)
occurring in relation to times with rising and absolute water temperature and declining
river discharge, especially as the lunar cycle approaches a new moon (Chapter 1).
Although abiotic conditions vary throughout a breeding season, the window individual

lake sturgeon choose to spawn is considerably smaller and choices are highly repeatable

48

in time across seasons irrespective of exogenous conditions (Table 2.1). This result
appears to conﬂict with studies suggesting that lake sturgeon have evolved the ability and
use exogenous cues to guide the timing of reproduction (Paragamian and Kruse 2001;
Chapter 1). However, subtle variation in individual SSDs may represent an
environmentally mediated and behaviorally plastic response within this window to
proximal exogenous conditions including water temperature, the lunar cycle and
discharge, all of which are consistent within and across years and have higher estimates
of repeatability (Table 2.1). One potential caveat to our data is that measures of
repeatability were dominated by observations from consecutive spawning episodes. Thus,
there may be temporal autocorrelation in environmental variables that would not be
evident over the lifetime of the individual. However, the SE for repeatability in SSD for
individuals captured twice were not different from SE estimated from individuals
captured more than 2 times.

Interestingly, repeatability estimates (r) may also provide an upper limit estimate
on the heritability (hz) for a given trait because repeatability includes both genetic and

environmental sources of variation whereas heritability only reﬂects genetic differences
(Boake 1989; Falconer and Mackay 1996; Dohm 2002). However, since SSD is a
reﬂection of both genetic and environmental components, and we have estimates of
repeatability for the plausible sources of exogenous variables thought to inﬂuence
spawning time in this species (i.e., discharge, water temperature and lunar effects;
Chapter 1) we estimate that heritability for spawning time may be considerably lower and
approximately 0.27 for both females and males (i.e., SSD repeatability — repeatability for

average water temp, river discharge and the lunar cycle). Comparable estimates of

49

repeatability for breeding time in natural populations of long-lived migratory animals,
especially ﬁshes, are rare in the literature. However, levels of repeatability for spawning
time derived from lab and aquaculture studies are generally higher than found here
(rainbow trout, r = 0.72-0.83, Quinton et al. 2004) but estimates of repeatability for other
behavioral traits such as mate preference (guppy, r = 0.05-0.35, Brooks 1996;

stickleback, r = 0.65; Bakker 1993) are similar. Although not directly comparable,

estimates of heritability (h2) for spawn date based on breeding studies of other migratory

animals including the collard ﬂycatcher (0.19; Sheldon et al. 2003) and German blackcap
(0.34; Pulido et al. 2001) as well as for migratory ﬁshes such as rainbow trout (0.55;
Siitonen and Gall 1989) and coho salmon (0.26; Gall and Neira 2004) are within our
pseudo-estimates of heritability for spawning time for lake sturgeon.

Heritability for spawning time may have several important implications for lake
sturgeon populations (Hendry et al. 1999). One is that natural selection may be effective

in shifting spawning time. For example, Sheldon et al. (2003) found that estimated

heritability for egg laying date in the collared ﬂycatcher, although small (h2 = 0.19),

promoted consistent and directional selection at the level of the phenotype. Second,
heritability would indicate that groups of individuals spawning at the same time in the
UBR, and those responding similarly to environmental conditions speciﬁc to the timing
of spawning, are also likely maintained through heritable additive genetic variation Thus,
the observed occurrence of multiple and temporally segregated breeding groups (Figure
2.2) were not random mixtures of individuals but rather groups of reproductively mature
adults with the tendency to reproduce at speciﬁc times (Figure 2.5). Although the

ultimate mechanisms that induce repeatable differences in spawning time in lake sturgeon

50

are generally unknown, we posit that groups of individuals may be genetically
predisposed to differentially respond to time-related cues such as photoperiod, a known
proximate cue for lake sturgeon at the point of river entry in the UBR (Chapter 1). Third,
the tendencies for adults to spawn at speciﬁc times as a ﬁmction of different
environmental characteristics such as water temperature may promote isolation,
phenotypic traits and adaptations unique to each breeding group (Hendry and Day 2005;
Woody et a1. 2000). Covariation in temperatures at the time of mating and temperatures
experienced through the period of embryonic and early post-incubation development
have been shown to affect body size and developmental rate (Atkinson 1994) and
offspring size (Fox and Czesak 2000). Stillwell and Fox (2005) have also shown that
temperatures experienced by adults affected life history traits of offspring including age
at ﬁrst reproduction, fecundity and embryonic developmental time. Recent studies in
migratory ﬁshes including UBR lake sturgeon also support this hypothesis. For example,
larvae produced from females spawning later in the season, when water temperatures are
generally warmer (Chapter 1), had signiﬁcantly smaller body sizes than larvae from early
spawning females implying different utilization efﬁciencies (Crossman 2008). Due to
these reasons and our ﬁndings of high repeatability in spawning time (SSD) for males
and females (SSD, Table 2.1), our study minimally suggests that a full quantitative
genetic analysis based on multigenerational pedigrees is warranted.
Inna-annual spawning time (learning)

Age and sex-dependent reproductive behaviors have been reported for many

species including migratory ﬁsh (e.g., Paitz et al. 2007; Dickerson et al. 2002). Fish of

different ages have also been found to spawn at different times (i.e., maiden vs. multi-

51

year spawning ﬁsh) (N iemela et al. 2006). Some migratory species, have been found to
alter reproductive timing based on previous successful experiences (e. g., egg and
hatching survival in blue tits; Grieco et al. 2002). Because lake sturgeon are a long-lived
species and individuals have opportunity to breed multiple times during a lifetime (i.e.,
estimated 18 times for males and 15 times for UBR females based on estimated adult
survival and age structure, repeatability could increase as a function of reproductive
experience (i.e., learning). Although repeatability varied among age classes according to
some environmental variables (e.g. river discharge), we generally found no statistical
evidence that variability in repeatability for SSD or for exogenous conditions decreases
with age and reproductive experience. Without a means of reinforcement (e.g.,
observations of reproductive success or failure associated with spawning decisions) or the
ability to recognize kin, selection operating as a ﬁmction of offspring survival associated
with conditions at the time of reproduction is unlikely to control repeatability in this
species. However, some sex-based differences were found with respect to spawning date
(Table 2.1). In particular, lower repeatability in spawning time for males may indicate
greater plasticity in behavior relative to females. We posit that lower repeatability in male
spawning time may be a reﬂection of sex-based alternative reproductive strategies
whereby males achieve reproductive success by attempting to increase mate number
rather than select appropriate breeding times.

Our analysis of spawning time was also intended to test the hypothesis that arrival
to spawning sites varied as a function of body size (and age). Our hypothesis was based
on two factors. First, phenotype-dependant based arrival is common in migratory animals

and proposed in migratory ﬁshes with larger older and generally better conditioned or

52

experienced. males and females often arriving ﬁrst (Moller 1994; Keefer et a1. 2009;
Cooper et a1. 2009). Second, early arrival may be adaptive because early arriving
individuals likely experience less competition, have access to high quality mates and are
able to select the best breeding habitats; all factors contributing to reproductive success
(Quinn 1999; Dickerson et al. 2002). We found that early arriving male and female lake
sturgeon were signiﬁcantly smaller than later aniving individuals (Figure 2.7). Size-
based arrival partly conﬂicts with the results discussed previously and generally contrasts
the ﬁndings of other studies. The positive relationship between body size and SSD
documented herein may also be driven by different historical recruitment patterns of early
and late spawning groups. For instance, contemporary levels of annual recruitment to the
larval stage have consistently been higher from early groups leading to a relatively
younger (and smaller) aggregate of early spawning individuals, relative to later groups
that are comprised of larger and older individuals.
Inter-annual spawning site selection

Spawning site selection is an important component of reproductive success in
migratory ﬁsh where strong associations between stream habitat features (e.g., substrate
size, infra-gravel water ﬂow) and levels of mortality during the egg and larval stages exist
(Chapter 4). The UBR has numerous suitable spawning locations throughout a 1.5 km
section. However, despite the close proxinrity of spawning areas in the UBR, males and
females aniving earlier in the season are signiﬁcantly more likely to select locations
further upstream (Zone 1-3; Figure 1.1) relative to individuals arriving later (Zone 3-6;
Figure 1.1) (Figure 1.6). These relationships held across years despite temporally

transient changes in sedimentation, stream cover or temperature across years or variation

53

in stream hydro-geomorphology and levels of ground water infusion have likely occurred
within and among years. Similar seasonal patterns have been noted in other migratory
ﬁsh and were attributed to habitat optimization (e.g., Orians and Wittenberger 1991),
whereby earlier aniving individuals selected the best habitats which were located ﬁrrther
upstream (Doctor and Quinn 2009). Mechanisms such as social learning (reviewed in
Kieffer and Colgan 1992; Olson et al. 1978), mate copying or heritable genetic attributes
(Quinn et al. 2006) that return individuals to areas differing in intrinsic physical
properties within natal streams have also been implicated as mechanisms.

Our ﬁndings that show early and late spawning adults consistently used different
spawning sites even though all sites were available during the entire reproductive season
suggests intra-annual variation in habitat quality and the ability of individuals to select
these high quality areas that are speciﬁc to different spawning times (i.e., habitat
optimization). Alternatively, relatively high estimates of repeatability for spawning zone
(especially for females; Table 2.1) may suggest that either spawning site quality varies
seasonally or selection for hydrogeomorphological cues intrinsic to speciﬁc spawning
times and locations is heritable. Further, higher repeatability estimates in spawning site
selection for females (I‘ able 2.1) suggest that females not only dictate the relative
spawning time (SSD) but also speciﬁc locations in the UBR Regardless, temporally
structured annual spawning groups of lake sturgeon are further segregated through the
selection of different habitats even over the course of a spawning season. Although ﬁne
scale homing has been observed in several migratory ﬁsh (e. g., Quinn et al. 1999), future
work could proﬁtably explore the relative inﬂuences of behavioral ﬂexibility versus

heritability in spawning site selection.

54

We should note that our observations may be a consequence of signiﬁcant
reductions in numerical abundance of UBR lake sturgeon over the past 50-100 yrs. Yet,
the UBR has always been the primary spawning area for ﬁsh in the drainage and thus our
inferences should be robust. Our observations were also made from a single population
and may not be generalizable to other systems. However, considerable similarities and
synchrony in spawning chronology of other Great Lakes tributaries with the UBR have
been found. Further, evidence of multiple spawning runs in other systems has been
reported in support of this research (Auer and Baker 2002). Thus, it is likely that other
populations of lake sturgeon, as well as other sturgeon species, are composed of discrete
breeding groups that are structured in part due to spawning periodicity and isolation as a

function of repeatable spawning time and spawning site selection.

55

CHAPTER THREE

SPATIALLY EXPLICIT ANALYSIS OF LAKE STURGEON EGG DEPOSIT ON AND
MORTALITY IN NATURAL STREAM SETTINGS

INTRODUCTION

Many organisms seek very speciﬁc habitats at different times of the year for the
purpose of reproduction. Further, the decisions that adults make with regard to breeding
site selection often have serious ﬁtness consequences for both themselves and for their
offspring. For example, the environments that animals select may have characteristics
that signiﬁcantly inﬂuence survival, especially for species that have offspring which are
vulnerable to numerous physical and biotic sources that contribute to mortality (Donazar
et al. 1993; Stokes and Boersma 1998; Wilson 1998; Hendry et al. 2001). High and
variable mortality during early life stages is important and often tied to variation in
population levels of recruitment (Bailey and Houde 1989). Second, habitats chosen for
reproduction can impose selection on phenotypic, behavioral, and morphological traits of
offspring, most of which are also tied in some way to survival during subsequent early
life stages that have long since dispersed ﬁom breeding grounds (Jensen et al. 2008;
Shine et al. 1997; Webb et al. 2001). Thus, the process of habitat selection for the
purpose of reproduction is often under strong selective pressure and in some situations
results in the evolution of physiological or behavioral traits that allow adults to use
exogenous (i.e., physical or biotic) cues to assess the relative habitat quality for their
offspring (Danchin et al. 1998).

For some species, the process of reaching critically important breeding habitats
requires individuals to complete long-distance migrations during abbreviated periods

within a season (Alerstam and Enckell 1979; Alerstam et al.2003). Although migration is

56

necessary for reducing mortality incurred during early life periods, it is also an extremely
risky behavior that comes with tremendous ﬁtness (i.e., energetic) consequences for
adults (Cooke et al. 2004; Jonsson et al. 1997; Hinch and Rand 1998; Shmueli et al.
2000). Because of this tradeoif, there is strong selection for migratory animals to make
the best choices by differentiating and accurately selecting among habitats for
reproduction after arrival (Blanchﬁeld and Ridgway 2005; F ontaine and Martin 2006;
Kondou et al. 2001; Rodewald and Yahner 2001). However, because the environments
that migratory species seek are often spatially complex, change temporally over short
time periods, and have varying implications across different early life stages (Sillett and
Holmes 2002) this task can be challenging. Yet, if selection is strong and is expressed
similarly across different breeding sites, the same suites of environmental traits
(including habitat) should theoretically be selected by breeding adults and equally
inﬂuence survival, regardless of where or when they elect to reproduce. This hypothesis
is rarely tested.

Highly migratory ﬁsh species that utilize spatially and temporally dynamic and
structurally complex rivers for reproduction provide an excellent opportunity to study the
effect of environmental covariates on the distribution of offspring and their subsequent
survival among sites selected for reproduction. Examples of important and quantiﬁable
features of river systems of stream habitats and those empirically shown to increase
survival rates, include river substrates with interstitial spacing (Bernier-Bourgault and
Magnan 2002; Chapter 4; Smith and Marsden 2009), highly permeable riverbeds with
ample ground and surface water exchange (Malcolm et al. 2003; Malcolm et a1. 2005;

Youngson et al. 2004) , increased river gradient (Curry and MacNeill 2004) and areas

57

with suitable water velocity (Smith and Marsden 2009) that provide sediment free
substrates. Interestingly, habitats associated with spawning locations also ﬁequently form
the basis for environmentally-mediated selection during embryogenesis and on
phenotypes during early life stages that create locally speciﬁc adaptations in many
species including ﬁshes (Hendry et al. 1995). As such, it is critically important for
researchers to characterize the physical and biotic features of breeding habitats that
inﬂuence survival during early life stages of these animals (Fausch et al. 2002).
Although offspring distribution and survival as a function of non-random
spawning site selection has been intensively studied in migratory ﬁsh with very speciﬁc
life histories (e.g., nest building), few studies have attempted to map the spatial
occurrence and subsequent survival of offspring for broadcast spawning species in
natural environments where fertilized gametes are tied to the locations of adult
reproductive activity with no parental care. Further, few studies have provided a detailed
understanding of whether deposition and early life survival are mediated by spatial and
temporal variation in environmental variables over small micro- geographic scales or
across time intervals of relevance to organisms during critical life stages (e.g., during
incubation). There are many reasons for this lack of information. First, highly migratory
broadcast spawning species are rare because of recent and dramatic population declines.
Second, analyses seeking to quantify the importance of environmental covariates to egg
deposition and survival in some situations need to account for more complex spatial
dependencies. Failing to account for spatial dependence may result in inaccurate
parameter estimates and erroneous conclusions regarding associations between stream

covariates related to spawning sites and gamete and offspring survival (Hoeting 2009).

58

Further, rigorous statistical applications for jointly modeling early life survival data
within and between time periods while accounting for both spatial autocorrelation and
exogenous covariates have yet to be developed for complex heterogeneous environments
such as streams and rivers.

We address both ecological and statistical limitations using the lake sturgeon, a
long-lived, iteroparous, highly fecund (11,000 eggs/kg) and broadcast spawning (i.e.,
eggs are released without nest preparation) migratory ﬁsh. Lake sturgeon were selected
for this study in part because they exhibit an aggregate mating behavior whereby
negatively buoyant eggs and sperm are released over large areas of rock and gravel. N o
nest preparation and no post-ovulatory parental care are provided for offspring. As such,
lake sturgeon eggs experience extremely high mortality in natural settings (i.e., 80-99%)
due primarily to direct exposure of eggs to harsh environmental conditions, predation or
microbial infection, and physical processes that dislodge and transport eggs from
spawning sites (Kempinger 1988; Parsley et al. 2002). Thus, the features of sites selected
by adults for reproduction are also likely conducive to egg and offspring survival.

The speciﬁc objective of this study was to quantify relationships between
environmental covariates characterizing attributes of lake sturgeon spawning locations
that were predictive of egg deposition and loss using a novel hierarchal spatial modeling
approach. Our working hypotheses were 1) stream variables deﬁning hydro-
geomophological conditions can explain variability in egg deposition and egg survival, 2)
the relative importance of environmental covariates on deposition and survival are

consistent across spawning sites suggesting that females are focusing on similar micro-

59

geographic features in each spawning environment, and 3) rates of egg survival are
constant across spawning sites once environmental covariates are considered.
METHODS
Study location
This study was conducted on a well-studied lake sturgeon population that migrate
annually into the Upper Black River (UBR), the primary tributary of Black Lake, MI
(Figure l), and use one of several spawning locations (Chapter 2). The UBR is a unique
location for studying migratory ﬁsh and exogenous environmental features associated
with adult spawning, egg deposition, and larvae for several reasons. First, individuals are
relatively free to choose spawning habitats. In other systems, for example, traditional
spawning sites have been blocked by dams, covered with sediments or destroyed by other
anthropogenic activities (Paragamian et al. 2009). Second, the UBR is relatively narrow
and shallow with low average discharge which facilitates observations of spawning
adults, monitoring early life history stages, and quantifying environmental covariates.
Other characteristics of the UBR include seasonal increases in water temperature,
stochastic variation in river discharge as well as predictable variation in the timing of
annual migration and spawning activity (Chapter 2). Further, Forsythe (2010), and Smith
and Baker (2005) detail exogenous correlates related to reproductive activity as well as
characteristics of the adult spawning population which also aids in the logistics of the
time sensitive data collection. F orsythe (2010) also provide evidence that adults returning
to breed at repeatable times and locations in the UBR suggesting that timing of spawning
is heritable and suitability of selected habitats in different locations by adults spawning at

different times may vary.

6O

Field sampling

Data were collected at four spawning locations selected by lake sturgeon along
the UBR, Figure 3.1, on 5/19/03, 5/12/05, 5/10/05, and 4/25/06 (Sites 1-4, respectively).
Lake sturgeon were observed entering and captured while spawning using handled
trapezoid shaped landing nets (78.7 cm base, 30.5 cm top, and 70.5 cm length).
Individuals were sexed at the time of capture by examining cloacal morphology and by
physically extruding gametes. Each Site was delineated using GPS and permanent stakes
were placed along the stream bank during the spawning event. One day after spawning
was completed and adults had left the area, transects were established across the stream
channel starting ~ 1 m below the most downstream spawning group and repeated every 5
m upstream to cover the downstream extent of the spawning. Counts of eggs and
measurements of environmental covariates were collected at a 1 m interval along each
transect (Figure 2). At each sample location, the number of eggs was measured using a

standardized kick net sample. Substrate was thoroughly disturbed for 5 seconds over a

0.125 m2 area of substrate upstream of a small triangular net (305 X305 X 305 mm

opening). Eggs deposited on and within the disturbed substrate were swept into the net by
the stream current.

The status (live vs. dead) of eggs collected during sampling was determined by
visual inspection and for microbial colonization in the form of visible fungal hyphae
growth. Numbers of live and dead eggs were recorded. Data on environmental covariates
(water depth, water velocity, and substrate size) were collected immediately after surveys
of egg deposition by a separate survey crew. Stream depth was recorded in meters using a

stadia rod. Water velocity at the stream bed was recorded using an electromagnetic ﬂow

61

meter (Marsh-McBirney, Inc.). Substrate size was recorded as the average of four

randomly collected stones within the 0.125 m2 sampled area. The largest stone within the

area was also measured to estimate maximum substrate size.

For Sites 1 and 3, a secondary sampling of egg deposition was conducted 3-4 days
following the initial data collection to determine the magnitude of egg loss. The number
of eggs and survival status was recorded. The sampling grid was shifted 0.5 m upstream
during the second sampling period to mitigate the disturbance caused by the ﬁrst
sampling, but retain the proximate sampling locations and to replicate as closely as
possible the spatial dispersion of sampling sites within transects.

STATISTICAL ANALYSES

Covariate summaries

Analysis of variance (ANOVA) and Kolomogorov—Smimov (KS) tests were used to
summarize and compare the frequency distributions of environmental covariates within
and among sites. The KS test was used to determine whether the frequency distribution of
sites' substrate size (average and maximum), water velocity, and water depth came from
the same population while AN OVA was used to test for signiﬁcant differences between
the means of these variables among the sites. Means were differentiated using a Tukey
HSD that adjusts for multiple comparisons. Surface plots used for exploratory analysis of
egg counts and environmental covariates were generated with multilevel B-splines using

the R (www.r-project.org) package MBA.

Embryo deposition
A Poisson regression model was used to quantify associations between egg counts and

environmental covariates measured at the same sites. This model is adequate in the

62

absence of correlated variation beyond what may be explained by possible covariates.
However, because lake sturgeon are broadcast spawners and eggs are extruded into a
spatially complex ﬂuvial environment, counts of eggs are expected to exhibit strong
anisotropic spatial dependency along the down-stream and across-stream sampling axes
(i.e., similar egg counts in the proximate locations) even after accounting for the
information provided by environmental covariates. This residual spatial correlation was

modeled as a random spatial effect. Given observations arise over a set of locations, say S
= {s1,. . .,sn} where s is a sample location’s spatial coordinates, we specify the log-link

model:

y(Si) ~ P0iSSOﬂ(Il(Si)); 10g (”(80) = X(S)'l3 + W(S),

where y(si) is the i-th sample location count, which is assumed to be distributed as a

Poisson variable with location speciﬁc conditional expectation 11(51). This conditional
expectation comprises vectors of environmental covariates and associated slope
parameters, x(s1) and B respectively, and w(s) which provides local adjustment (with

structured dependence) and captures the effect of unmeasured or unobserved spatial
covariates. Here, we assume the random spatial effects follow a Gaussian Process (GP),
where w(s) ~ GP (0, 02 p (°; 0)) with variance 0'2, and correlation function p (s, s'; 0). The
vector of parameters 0 control the correlation function’s behavior. For a collection of 11

locations the n X 1 vector of random effects follows a Multivariate Normal distribution, w

~ MVN (0, 0'2 R(0)), where R(0) = [p (31, Si; 0] 214:1 is the correlation matrix.

63

Embryo survival
To address the embryo survival, we would ideally use live egg count at sample time 1
and 2 to calculate survival rate at each location. This rate could be related to the predictor
variables and random spatial effects through a binomial generalized linear model.
However, because of the time 1 and 2 sampling grid offset and egg transport by the
stream current, our data have several locations where there are more eggs recorded in
time 2 than time 1. As a result, we cannot directly model the survival rate. Therefore we

opt to model the change in live egg counts as:

 

10g(y(si,t2) + 1) _ '
(log(y(si,t1)+1))— N (X(5) 13+ W(S)+ e(s) (2)

where y(si,t 1) and y(si,t2) are the i-th location's egg count at time 1 and time 2,

respectively. Here, we made the assumption that time 1 and 2 sampling grids coincide.
By modeling the outcome as a normally distributed continuous variable, we are able to

include the error term which is assumed to follow an independent and identically

distributed Normal distribution, i.e., for the n locations e(s) ~ N(0,t21), where I is the

. . . 2 . .
identity matrix. Here, I mcludes the measurement error and variance not captured by the

covariates and the spatially structured residuals, and could also include micro-scale

spatial variability that occurs at a resolution ﬁner than the sampling grid. In the
geostatistical literature 12 is referred to as the nugget, whereas, 02 is viewed as the partial

sill.

Spatial distribution and Bayesian implementation

64

Spatial correlation functions of varying complexity are available for deﬁning p(s,

s*; 0) (e. g. Cressie 1993). A common and more simplistic choice is an isotropic
exponential, p (s, 5*; 0) = exp (-(D||si — 51'”), where (I) is the spatial range parameter. Here,

it is assumed that spatial dependence is the same in all directions. Given the dynamics of
stream ﬂow, this is a rather strong assumption. To explore the possibility for a more

complex spatial structure we also use and anisotropic form where p(s, 5*; 0) = exp [(5 —
5*) [G((p)A2 G'(<p)]-l (s — s*)]. For this model, the rotation matrix G(<p) controls

directional dependence and the positive diagonal matrix A deﬁnes the range of spatial
dependence.

When describing a spatial process it is useful to report the distance at which
spatial dependence is negligible. This distance is referred. to as the effective spatial range,
and is deﬁned here at the distance at which the spatial correlation drops to 0.05. For the
isotropic exponential correlation function this is - log(0.05)/<D and for the anisotropic
form this is the — log(0.05)it, where it is the diagonal value in A. Note that there are two
effective spatial ranges for each anisotropic process, one oriented with q), and the other
perpendicular to (p.

A prior was assigned to each parameter to complete the model Bayesian
speciﬁcation. The B’s received ﬂat prior distributions and the spatial variance parameters
were assigned inverse-Gamrrm (1G) with hyperparameters IG (2, -). With a shape value of
2, the IG distribution has inﬁnite variance and centered on the scale value, which is
dataset speciﬁc. The spatial range parameters 1 follow a Uniform prior, which are chosen
to support a spatial range from 0 to the maximum inter-location distance in the dataset.

The rotation parameter (p also receives a Uniform prior that is data set speciﬁc.

65

Joint modeling, cross-correlation and egg survival
The previous models (1) and (2) allow us to relate egg deposition and subsequent
change in live egg count to a suite of environmental covariates. To address the third
hypothesis, we could model time 1 and time 2 live egg count jointly as a frmction of the
environmental covariates and respective spatial residual processes. Speciﬁcally, we could
assess the degree of location speciﬁc association between the residuals to assess evidence
of a non-stationary multivariate residual spatial process. A model to explore this question

is:

(108043, 12) + 1)) ~ N ((X(S) ® by B + W(S) + e (8)) (3)

1080151, t1) + 1)

where, again (si,t1) and y(si,t2) are the i-th location's egg count at time 1 and time 2,

respectively. This model is a non-stationary version of the linear model of co-
regionalization derived by Gelfand et al. (2004). Here, we assume the process w(s) = A(s)
v(s) to be a linear transformation of v(s), where A(s) is a space-varying matrix transform

that is non-singular for all 5. Speciﬁcally, for a given location A(s) is a lower-triangular

matrix with elements a1,1(s), a1,2(s) and 32,2 (5) modeled a independent GPs with

. . . 2 2 2 .
corresponding spatial variances 0 1,1, 0 1, 2, o 2, 2 , and spatial ranges (131,1, (D1, 2, and
(D22 (for details see Finley et al., 2010). The non-spatial residual process, e(s), is
assumed to follow N (0, Diag(r21, :22». The space-varying correlation is deﬁned by v(s)

= v1(s), v2(s))’ which is a 2 X 1 process with independent zero-centered spatial processes

66

with unit variance; that is, v1(s) and v2(s) follow a GP(0,p1(-,(I>v1)) and GP(0,p1(',(1>v2)),

respectively.
Given the complexity of this model and the paucity of data, we make two

simplifying assumptions. First, the GP’s associated with A(s) share a common spatial
range parameter (DA. Second, a common (Dv is used for the v(s) processes. These

concessions improve model identiﬁability of parameters and greatly reduce computing

time (Finley et al. 2010). Importantly, for any given location, we can take [(5) =
A(s)A(s)' as the residual t1, t2 covariance matrix. To recognize how missing spatial
environmental covariates might impact live egg count over the sampling interval we can

look at a surface of 1'1, 2 (s) or the corresponding correlation i.e., p1, 2(5) =

132(8)
J 13,1 (8) 132(8)

 

. Also, with access to the ﬁrll posterior distribution of p1, 2(5) we can

formally test the signiﬁcance of this correlation at any given location.
Model selection and implementation

To compare several alternative models with varying degrees of covariate and
spatial richness, we use the Deviance Information Criterion (DIC) (Spiegelhalter et al.,
2002). Letting Q be the generic set of parameters being estimated for each model
(including random effects), we compute the expected posterior deviance D(Q) = E(Q|Y)
[-210gL(Data | Q) ], where L(Data | Q) is the ﬁrst stage Gaussian likelihood from the
respective model and the effective number of parameters (as a penalty) as pD= D(Q)-

D(Q), where Q is the posterior mean of the model parameters. The DIC is then given by

D(Q) + 131) and is easily computed from the posterior samples with the lower values

67

indicating better models. For all models, three MCMC chains, with unique starting
values, were run for 50,000 iterations. The CODA package in R (www.r-project.org) was
used to diagnose convergence by monitoring mixing using Gelman-Rubin diagnostics
and autocorrelations (see, e. g., Gelrnan et al. 2004, Section 11.6). For all analyses,
acceptable convergence was diagnosed within 10,000 iterations (which were discarded as
bum-in). The sampler was coded in CH and Fortran and leveraged Intel's Math Kernel
Library threaded BLAS and LAPACK routines for matrix computations. This code will
be available in version 0.1-7 of the spBayes R package and also submitted along with the
illustrative data sets to Ecological Archives.
RESULTS
Summary of spawning groups and environmental covariates
The spawning locations used by adults exhibited similar ranges in substrate sizes

(1-180 mm), maximum substrate size (2-300 mm), water depth (0.1-1.2 m) and water

velocity (0.1-1.5 m3/s). However, the KS test indicated that all environmental covariates

had signiﬁcantly different frequency distribution across sites (P < 0.0001; Figure 3.2).

Sites also differed signiﬁcantly in mean values for average substrate size (Site 1: 40.8;

Site 2: 88.4; Site 3: 60.7; Site 4: 70.7; F 3, 647 = 32.9, P < 0.001), maximum substrate size
(Site 1: 64.1; Site 2: 111.9; Site 3: 83.5; Site 4: 109.3; F3, 647 = 24.1, P < 0.001), and
water velocity (Site I: 0.54; Site 2: 0.65; Site 3: 0.63; Site 4: 0.48; F 3, 647 = 13.4, P <

0.001) but not in water depth (Site 1: 0.52; Site 2: 0.56; Site 3: 0.56; Site 4: 0.51; F 3, 647

= 2.18, P = 0.09). Frequency distributions and pair-wise comparison of means indicated

that sites 2 and 4 had signiﬁcantly larger substrates and sites 2 and 3 had signiﬁcantly

68

greater water velocity at the streambed. Environmental covariates also exhibited spatial
heterogeneity within and among locations, see e.g., Figure 3.3 (a-d).
Embryo deposition and environmental covariates

A total of 9,426 eggs were collected across all spawning locations from 651 kick
net samples. Egg count was location-dependent and generally concordant with increased
female abundance (Site 1 = 2,509, Site 2 = 564, Site 3 = 2,111, Site 4 = 4,242). Within
each Site, eggs were primarily deposited along the downstream axis on the side of the
river spawning took place and ﬁ'om 0 to 60 m directly downstream from locations
selected by spawners. However, egg deposition exhibited non-random and spatially-
dependent patterns (see e. g., Figure 3.3 (e)), often occurring in discrete patches that
ranged in count from 0-345 eggs.

Both the non-spatial and spatial versions of Model (1) were ﬁt using the 145, 61,
250, and 199 observations collected at Sites 1-4, respectively. Tables 3.1 - 3.4 provide

candidate models' parameter estimates for each site. Looking ﬁrst at the spatial model's
effective spatial range parameter estimates, 7111 and 11, we see there is relatively strong

but variable spatial dependence among the residuals. For example, Site 1 exhibits the

shortest effective spatial range of ~ 5 m, whereas Site 4 shows the longest range of ~ 54
m. For Sites 1, 3, and 4, the disparity between 710 and 7L1 suggests there are strong
anisotropic patterns in spatial dependence. In each of these sites (p describes the
orientation of the major spatial range parameter axis M. Here we offer (p in both radians

and degrees, where in both cases zero is perpendicular to the west river bank. Not

surprisingly parameter estimates suggest a shorter range of spatial dependence

69

perpendicular to water ﬂow. Sites with the maximum dispersal distance (i.e., Site 3) also
had the highest average water velocity.

For all sites, most of the non-spatial models' parameters associated with the
environmental covariates are signiﬁcant at the 0.05 level (i.e., the 95% credible interval
does not include zero). This is not the case for the spatial models. Recall a key
assumption of generalized linear models is independent and identically distributed
residuals. The interpolated surface plots of the non-spatial model residuals for all sites
reveal substantial spatial dependence, see e. g., Figure 3.4 (a), which suggests we are
violating this model assumption and must accommodate the lingering spatial dependence
through a random spatial effect. Ignoring this spatial dependence results in parameter
credible intervals that are too narrow and, as a result, we might draw incorrect
conclusions about the impact of associated environmental covariates. Including the
random spatial effect helps satisfy the model assumptions and subsequently provides a
more realistic view of the statistical relationship between egg deposition and the
environmental covariates few explain a substantive amount of variability in egg
deposition.

Beyond meeting the model assumptions, the addition of a random spatial effect to
the mean greatly improves model ﬁt. This resulted in lower DIC values, Tables 3.1- 3.4,
and visually comparing the observed and ﬁtted deposition surfaces, Figure 3.2 (e) and
Figure 3.4 ((1), respectively. Note, in comparing Figure 3.2 (e) to Figure 3.4 (d) we see
the non-spatial model fails to approximate the observed deposition, which further
emphasizes its inappropriateness for these data. The same conclusion can be drawn by

looking at the surfaces from the other three sites.

70

Considering the spatial models’ parameter estimates, egg count is positively
associated with increasing average or max substrate size at three of the four spawning
sites. Water depth was also a positively associated with egg deposition at 2 of the 4
locations (i.e., Sites 1 and 4). Although egg deposition was positively related to water
depth at both sites the expected increase (slope) in egg deposition for every unit in
increase in water depth was much greater at Site 4 versus Site 1. The regression
coefﬁcient associated with water velocity was also statistically different from zero at Site
4, which suggests egg count may increase with increasing in velocity.

Embryo survival and environmental covariates

Analyses were performed to determine degree of spatial dependency and effects
of environmental covariates on changes in live egg count after 3 days following
deposition. At Site 3, a total of 643 viable eggs were collected during the second survey
representing a 70% reduction in the total nrnnber of viable embryos collected three days
prior in the ﬁrst sampling. At Site 4, 1,276 viable eggs were collected dining the second
sampling period representing a similar 69% reduction in the number of viable eggs
collected relative to the ﬁrst sampling. Within each site, the non-spatial and. spatial
versions of the postulated model (2) were to those locations that had at least 1 egg
recorded in time 1. Again, this location subset was used because our focus was on loss
over the sampling interval.

Parameters' posterior summaries and model diagnostics are offered in Tables 3.5
and 3.6. Here we see the environmental covariates explain only a marginal amount of the
variability in the log difference in egg count between time 1 and 2. However, average

substrate size was a signiﬁcant predictor of the change in egg abundance between

71

sampling periods for Site 4 but not Site 3 (Tables 3.5 and 3.6). Speciﬁcally, the negative
parameter estimate indicates that the ratio between time 1 and time 2 egg count becomes
smaller (i.e., fewer in time 2 relative to time 1) as substrate size decreased. Also, again
due to spatial autocorrelation, the non-spatial model's credible intervals are falsely
precise. The addition of the random effects allows us to apportion total residual
variability into spatial and non-spatial components. Here we see that for Site 3, ~ % 62 of
the total residual variability can be explained by spatial proximity i.e., 0.62 = 1:32 / (0.81
+ 1.32), whereas for Site 4 the residual variance is dominated by the nugget (i.e., there is
relatively more non-spatial variation either due to measurement error or variations in the
data that relate to shorter ranges than the sampled data spacing). Similar to egg
deposition, log difference in egg count shows strong anisotropic effects, with a longer
range of spatial dependence coinciding with the direction of stream ﬂow which is also an
indicator of the importance of water velocity to covariance in numbers of live eggs
between the 2 sampling periods. The lower DIC values also support the inclusion of the
random spatial effects.
Cross-correlation and joint modeling of egg survival

Joint modeling of live log egg counts at time 1 and 2 allows us to assess the
consistency of change over time after accounting for the environmental covariates. If the
rate of change is consistent over locations then we would expect a common correlation
between the time 1 and 2 spatial random effects over the Site, i.e., a stationary cross-
covariance. The results of this multivariate model (3) for Site 3 and 4 are offered in Table
3.7. Here we have environmental covariate regression parameters for time 1 and 2 log

egg counts. With the exception of depth in Site 3, the covariates are not contributing

72

much to our understanding of deposition and subsequent loss. Note, again, we are using a

. . 2
subset of locatrons where the egg count at time 1 was at least 1. The three 0 parameters

are the GP's spatial variances that compose the components of A(s). Their common
spatial range parameter (Da has an effective spatial range of 25 and 16 for Sites 3 and 4,
respectively, and relatively narrow 95% credible intervals suggest some degree of spatial
non-stationarity across the two sites.

The spatial range parameter associated with the cross-correlation, (Dv, also shows
a relatively deﬁned range of dependence within and between the processes. We can look
to surface plots of the correlation between wt=l (s) and wt=2(s) to visually assess the
variability of residual change over the sites. These surface plots are offered in Figures 3.5
and 3.6 for Sites 3 and 4, respectively. Here we see that after accounting for the
environmental covariates, there are location-speciﬁc and unexplained differences in the
correlation of the random spatial effects. For instance, Figures 3.5 (e) and 3.6 (e) show
correlations ranging from ~ 0.2 — 0.8. This pattern suggest that mortality is random with
respect to location or we are missing spatially varying physical or biological covariates
that effect rates of egg survival over the sites.

DISCUSSION

We provide an empirical and analytical framework that is conducive to studies of
many organisms that inhabit and reproduce in spatially and temporally complex
environments. We show that linking spatially explicit point pattern analysis is necessary
to understand and generalize (across sites) the contribution of factors affecting spawning
site selection and to identify associations between habitat and early life survival at small

spatial scales.

73

Selected spawning locations

Migratory animals including birds, turtles and ﬁshes have evolved the ability to
use abiotic (i.e., temperature, vegetation, substrate) and biotic cues (i.e., social
inﬂuences) for selection of breeding areas (Knapp and Vredenburg 1996; Melles et al.
2009; Pike 2008). Thus, it seems reasonable that lake sturgeon also use exogenous cues
to select spawning areas that provide reliable information regarding probabilities of
fertilization success and embryo survival during incubation (Kempinger 1988). Lake
sturgeon aggregate and spawn in relatively deep pools immediately upstream from
raceway segments characterized by increasing water velocity and substrate size but
decreasing water depth lends support for this hypothesis (Figure 3.3).

Conditions at the site of spawning may not have been predictive of stream
features downstream where eggs were deposited Downstream locations were found to
differ signiﬁcantly with respect to substrate size and water velocity. However, while
spawning locations may have differed from each other it is possible that their physical
properties were actually more similar than unselected locations. We did not measure the
physical properties of other UBR segments. However, recent work by Chiotti et al.
(2008) demonstrated that locations where lake sturgeon eggs were deposited, and thus
locations individuals had selected to spawn, had signiﬁcantly greater heterogeneity in
substrate size compared to non-selected spawning locations. Other studies have also
noted speciﬁc preference of sturgeons to select substrate free of sediments within a range
of water depth and water velocity (LaHaye et al. 1992; Harkness and Dymond 1961; Fox

et al. 2000; Collins et al. 2000)

74

Alternatively, spawning site selection may be driven by physiological or genetic
components as opposed to environmental variation (i.e., individuals are returning to their
place of birth). For example, Stewart et al. (2004) found that sockeye salmon displaced
from spawning areas returned to former locations, suggesting the importance of olfactory
ones as opposed to habitat features in selection. There is also evidence of site-ﬁdelity in
sturgeon species. F orsythe (2010) documented a signiﬁcant tendency for adult lake
sturgeon to non-randomly select spawning locations based on spawning time (i.e.,
individuals spawning early selected locations further upstream). Paragamian et al. (2009)
found that while the cobbles and gravel substrates of historical spawning areas were
buried with post-dam sediment, white sturgeon still used these areas repeatedly for
spawning.

' While our data do not unravel the relative importance of environmental and
genetic inﬂuences on selection of spawning locations, we believe that both forces are
signiﬁcant and future studies should address this question for several reasons. First, if
female choice of spawning habitat also affects probabilities of survival of eggs and
larvae, then the relative environmental-genetic contribution of spawning site selection
will be predictive in how sturgeons will respond to anthropogenically altered systems.
Second, our analysis of stream hydrology and substrate also provides evidence for a
signiﬁcant degree of difference between spawning environments experienced by eggs and
larvae of adults spawning at different times. Thus, differences in the physical features of
stream associated with spawning habitat may result in adaptive phenotypic variation
among sturgeon populations but among breeding groups within the same population that

are reproductively isolated in time (Hendry et al. 1995).

75

Spatial dependency, egg deposition and environmental covariates

Depositional proﬁles of lake sturgeon eggs 1 day after spawning were
signiﬁcantly clustered at all spawning locations. Strong spatial dependency is likely due
to the spawning behavior and the physical system (Bruch and Binkowski 2001). Eggs are
released and subjected to site-speciﬁc ﬂuvial dynamics and geomorphic heterogeneity
near the streambed that non-randomly determine the direction of embryo drift and the
distance travelled. The effect of ﬂow to generate ‘hot spots’ in other ecological processes
such as the drift and colonization of aquatic macroinvertebrates is frequently observed in
lotic systems (Bond and Downes 2003). Our results are concordant with others showing
that reproductive behaviors such as nesting in birds or redd site selection in migratory
ﬁshes have spatially explicit (i.e., clustered or patchy) properties due to both abiotic (i.e.,
habitat) and biotic (i.e., social) mechanisms related to the species reproductive ecology
(Melles et al. 2009; Hendry et al. 2001).

Lake sturgeon egg depositional proﬁles are spatially autocorrelated and
anisotropic. The spatial range of egg deposition was unidirectional and always greater in
the downstream vs. cross stream direction. The tendency for UBR lake sturgeon to select
areas and spawn to one side of the UBR and for natural stream ﬂow to disperse eggs
longitudinally likely drives anisotropy. Model ﬁt was also improved with the addition of
environmental covariates, supporting our prediction that hydro-geomophological
environmental stream variables will also explain variability in egg deposition within a
location. Speciﬁcally, egg deposition was signiﬁcantly associated with increasing

substrate size (i.e., max and average).

76

One possibility for this relationship is that turbulence, generated by interactions
between river ﬂow and substrate, acts as a mechanism to aggregate eggs in areas of larger
substrate (Cardinale et al. 2002). Recently deposited eggs may also become trapped by
interstitial spacing of larger rocks serving to reduce subsequent transportation before eggs
become adhesive. Water velocity and depth were also found to be signiﬁcant predictors
of egg deposition for similar reasons. Our results are concordant with others suggesting
that stream variables including water depth, water velocity and substrate size are
predictive of egg deposition for migratory ﬁshes (Mulli and Wilzbach 2007 ; Knapp and
Preisler 1999; Thurow and King 1994). LaHaye (1996) and Chiotti et al. (2008) also
reported similar and consistent associations between numbers of lake sturgeon eggs
deposited with substrate size, depth and velocity across the species geographic range.

Survival and environmental covariates

Relationships between environmental covariates, particularly substrate size, with
number of eggs deposited were also consistent across spawning areas (Tables 3.1 - 3.4).
Differences in the strength and the sign of parameter estimates in aspatial models would
have provided contradictory and thus misleading conclusions (i.e., 15 out of 18
environmental covariates measured across Sites 1-4 were found to be signiﬁcant
predictors of egg deposition). Results show several important points. First, while the
locations adults selected to spawn may be different ﬁom each other in their physical
properties, and spawning site selection varies by females spawning at different times with
different ecological conditions (Chapter 2), the environmental characteristics associated
with places eggs are deposited are similar. Thus, the speciﬁc habitats selected among

spatially complex habitats for reproduction are likely to set into motion similar biological

77

and physical processes that inﬂuence offspring survival and growth. Our ﬁnding of a
similar 70% reduction in viable egg abundance between sampling periods for both
locations supports this hypothesis.

The environmental features we measured were selected because of known
relationships between these features and adult spawning site selection. The same features
were assumed to also be predictive of deposition rates and embryo survival. Results
revealed that relationships were weaker than anticipated However, average substrate size
was a signiﬁcant predictor in the magnitude of change in egg number over three days.
Fewer viable eggs remained after 3 days as a function of decreasing substrate size (i.e.,
the ratio between time 1 and time 2 decreases with decreasing substrate size). One
explanation for this relationship is that relatively smaller and perhaps more
heterogeneous substrates with increased porosity reduce the level of predation on eggs by
predators inhabiting interstitial spaces in stream substrate (Chotkowksi and Marsden
1999; Chapter 4; Palm et a1. 2009). Smaller substrate sizes may also have reduced loss
due to physical disturbances transporting eggs out of the spawning area (Biga et al. 1998;
F itzsirnons et al. 2007). However, environmental covariates were only signiﬁcant for one
location, Site 4, a spawning area with the largest average substrate size suggesting that
this pattern is not generalizable to all spawning areas.

The rate of change in egg loss provided by estimates of the correlation between
time 1 and time 2 (Figures 3.5 and 3.6) also suggested that loss was non-stationary (i.e.,
rates of loss were different across locations) and relatively independent of the
environmental covariates measured This result was surprising given recent research

showing that variation in incubation conditions (i.e., water ﬂow, substrate size) in

78

 

IE

St

511
et
in
lm1
H01

em-

experimental settings signiﬁcantly altered rates of mortality during incubation (Chapter
4). Although water depth, water velocity and substrate size are characteristics
traditionally measured to quantify spawning habitats selected by migratory ﬁshes

(W ollebaek et a1. 2008), these variables are only predictive of where individuals spawn.
Others variables that deﬁne habitats at ﬁne spatial scales including hyporheic
temperature, substrate porosity, dissolved oxygen, pH, bedform migration and the
presence of groundwater spring or vertical hydraulic gradients may be better predictors of
egg survival (Geist and Dauble 1998). Information on these later variables is often
required to estimate and to accurately quantify the amount of habitat available to
spawners (Geist and Dauble 1998), predict the locations that may be selected by adults
(Mulli and Wilzbach 2007) and estimate survival to hatch.

Additional unmeasured variables are likely important to egg survival. Of the
abiotic factors mentioned previously we tend to focus on the hyporheic zone for several
reasons. First, ground-surface interactions are highly variable, temporally, and at
extremely small spatial scales in streams (Boulton et al. 2010; Hakenkamp et al. 1993).
Second, egg survival of numerous migratory ﬁsh is directly proportional to the amount of
dissolved oxygen (DO) available at the substrate surface (i.e., high (DO) high egg
survival; Malcolm et al. 2003) vs. groundwater (i.e., low DO low egg survival; Malcolm
et al. 2004). Inter-annual spatial and temporal variability in ground-water surface water
interactions and hydrological events that affect hyporheic water quality have been
implicated in driving variation in annual recruitment in salmon (Malcohn et al. 2005).
However, variable loss may not be a function of the physical properties of the

environments selected. One example includes aspects of the biotic predator community

79

inclt
inter

of SI

including type, size and density. Forsythe et al. (2010) showed that high densities of
interstitial predators could consume 99% of lake sturgeon eggs over 24 hours regardless
of substrate size. Moderate levels of predation have been implicated as driving the
recruitment of many ﬁshes especially at low population abundance (i.e., predation driven

Allee effects; Gascoigne and Lipcius 2004).

80

Sl

cemr
vane
POPt
cont
al. 21
base.
1931K
(Fair
prog:

Wisc

dtClli
role 1
high 1
Kant;
Wrath
Houde

Proces

CHAPTER 4
SOURCE-BASED ASSESSMENT OF LAKE STURGEON EGG MORTALITY IN
NATURAL AND EXPERIMENTAL STREAM SETTINGS
INTRODUCTION

Identifying the sources of mortality and estimating the magnitude of loss are
central goals in population ecology that offer critical insight into naturally occurring
variation in levels of annual reproduction (Caselle 1999), provide projections of future
population trends and dynamics (W iegand et al. 1998) and indicate the relative
contribution of different life history stages and their traits to future generations (Gross et
al. 2002; Martin 1995; Gaillard and Yoccoz 2003). A detailed understanding of source-
based mortality at different life stages also provides insight into how populations will
respond to natural or human- induced changes in environmental conditions or to harvest
(Fahrig 2002), both important factors in developing science-based adaptive management
programs for species that are exploited or of conservation concern (Doak et al. 1994;
Wisdom et al. 2000).

Mortality occurs throughout an organism’s life cycle at varying rates that often
decline with development and growth. Early life mortality however plays an important
role in population dynamics of many species because mortality is often disproportionally
high (Congdon et al. 1999; Donovan et al. 1995; Gosselin and Qian 1997; Katvala and
Katitala 2001; Muths et al. 2003) and variation in a single source (e.g., predation,
weather) can generate large and predicable ﬂuctuations in annual abrmdance (Bailey and
Houde 1989; Houde 1987; Lubow and Smith 2004; Osman and Whitlatch 1995).

Processes occurring early in life can also “cascade” through subsequent life stages,

81

rile:
phen

2006

time

that

cite
P0P”
lhﬂi
neu‘
Char
senir
gear I
sue.
moutl
preda
genen
2003;
II101131
Cousin

muchc

affecting not only abundance and levels of annual reproduction but variation in
phenotypic characters (e. g., body size) tied to probabilities of survival (Vandenbos et al.
2006). High early life mortality also produces a host of important evolutionary
adaptations (e.g., ageing, reproductive behaviors, and parental investment) that enhance
ﬁtness, particularly when parents contribute differentially to annual reproduction and
when traits are heritable (Roff 2002).

In many ﬁsh species, early life mortality during the egg and larval stages is
excessively high, often exceeding 95%, and plays a major role in setting the limit on
population abundance (Bouwes and Luecke 1997; F itzsirnons et al. 2007; Smith and
Marsden 2009). Thus, knowledge of the sources and estimates of early life mortality are
viewed as essential elements in predicting future levels of annual reproduction
Quantifying some-based early life mortality however is complicated in natural aquatic
settings due to the obscurity of early life stages and the technical limitations of sampling
gear (Houde 1987). Fish species also display an array of early life history traits (e.g., egg
size, size at hatch, incubation time), reproductive behaviors (e. g., nest construction,
mouth brooding, egg fanning), physiological defenses (e.g., induced hatching) and anti-
predator mechanisms that may compensate for excessive loss, which tends to make
generalizations difficult (Utne-Palm 2001; Taborsky and Foerster 2004; Knouft et al.
2003; Wedekind and Muller 2005). The ecological and evolutionarily role of early life
mortality in short-lived species exhibiting very speciﬁc behavioral adaptations (e. g., nest
construction in Salmonids) or life histories that minimize loss, for instance, are likely

much different for relatively long-lived iteroparous species (e. g., walleye, sturgeons and

'82

pad

(ll:

3C0

lea1

ads
rep

ind

19‘

the

p01
ind

ind

ﬁel
and
em;
bee:

liter

paddleﬁsh) that don’t construct nests nor provide post-ovulatory parental care
(Winemiller and Rose 1992).

Adding further complexity, early life dynamics in ﬁsh also vary substantially
across populations due to adaptations that have evolved as a function of site-speciﬁc
features of spawning locations that include gradients of substrate size, oxygen availability
and predator sizes/types (de Leaniz et a1. 2007 ; Quinn et al. 1995). Similar ecological
adaptations can also evolve within segments of the same population which are
reproductively isolated (i.e., adaptation by time; Hendry and Day 2005) because
individuals and their offspring encounter ecological conditions that select for traits
unique to each reproducing segment even over short time scales (Hendry and Quinn
1997; Kinnison et al. 1998). To this end, the magnitude and sources of mortality across
the range of habitats experienced by ﬁsh eggs and larvae of a single species are largely
unknown. Stage-based estimates of mortality, where available, are also typically provided
in isolation of the many suspected sources predicted to vary across space ande time;
potentially underestimating the magnitude and variability of loss and leaving little
indication whether the sources of mortality vary directly (i.e., compensatory) or
independently (i.e., additive) across reproductive locations and times.

The main goal of this study was to address this uncertainty using a combination of
ﬁeld-based manipulations and laboratory experiments designed to identify the sources
and magnitude of egg mortality in the lake sturgeon (Aczpenserﬁrlvescens). We place
emphasis on the egg stage because of our unique access to this brief early life phase and
because embryonic dynamics are relatively underappreciated in the ﬁsheries ecology

literature, limiting our understanding of ﬁsh population dynamics especially for species

83

with
inch
addr
part
sire;

Van

pan
yea
iter
fast
rep
(We
can
spa

(Cl.

gen
Mo:
indi
it al

P0pl

with unique life history traits (e. g., longevity and iteroparity) and reproductive behaviors
including broadcast spawning and no parental care. The following objectives speciﬁcally
addressed were to: (I) obtain estimates of mortality/survival during the egg stage, (2)
partition mortality into sources that may vary spatially and temporally across natural
stream incubation environments and (3) quantify the effects of abiotic and biotic
variables such as substrate and predator density on the magnitude of egg consumption.
Lake sturgeon are an interesting species for studying early life mortality due in
part to their unique life history; i.e., lake sturgeon are a long-lived (ages may exceed 100
years), large-bodied (1 1-100 kg), late-maturing (12—25 yrs of age depending on sex),
iter0parous ﬁsh species with high fecundity (11,000 eggs/kg) that migrate to sections of
fast-ﬂowing rivers for reproduction in the spring (reviewed in Peterson 2007). Dming
reproduction, eggs and sperm are released (i.e., broadcasted) into the stream environment
over rocks and gravel in the absence of nest site preparation. Despite little post-ovulatory
care, numerous aspects of this species’ biology and behavior such as highly synchronized
spawning events (F orsythe et al. 2010), non-random selection of stream benthic habitats
(Chiotti et al. 2008), adhesive properties of eggs (Bruch et al. 2006) and relative large
embryos (2.7-3.8 mm) with short incubation times (5—14 days) potentially enhance (or
generate variation in) probabilities of embryonic survival (Kempinger 1988).
Mortality during the egg stage is high and variable with an estimated 80 — 99% of
individuals dying through incubation (Carofﬁno et al. 2010; Johnson et a1. 2006; Nichols
et a1. 2003). Thus, the egg stage is likely a signiﬁcant component of the species

Population dynamics (Gross et al. 2002; J ager et al. 2002).

84

Mortality can be attributed to several sources including (Parsley et al. 2002): l)
developmental/physiological mortality occurring when eggs die due to genetic or
developmental abnormalities that arise during incubation, 2) predation by a wide range of
vertebrate and invertebrate benthivores (Kempinger 1988) and 3) physical processes in
the form of water currents that may dislodge and transport healthy eggs ﬁ'om rearing
environments. However, the relative magnitude of mortality sources has not yet been
quantiﬁed in natural settings. Without this knowledge, it is difﬁcult to determine whether
substantial variation in recruitment documented between groups that reproduce at
different times and/or utilize geographically different and unique habitats for
reproduction (Chapter 3) can be attributed to dynamics occurring during the egg stage.

METHODS
Study site

This research was conducted using eggs from a well studied population of lake
sturgeon in Black Lake, MI (Smith and King 2005). The Black Lake population consists
of approximately 1, 100 sexually mature adults, upper 95% conﬁdence (Baker and
Borgeson 1999). The Upper Black River (UBR), a fourth-order stream, is the largest
tributary of Black Lake and is approximately 91.7 km long with 11 km of staging and
unrestricted spawning habitat downstream from Kleber Dam (ca. 1949). However,
shallow and wadable locations within a 1.5 km section of the UBR 9 km upstream from
the river mouth are used for spawning annually (Chapter 2). Reproduction occurs
exclusively in the UBR from late April through early June (Chapter 1). Approximately
100-250 adults spawn annually and variation is observed in individual spawning site

arrival time, the chronology and duration of spawning activity, adult sex ratio, size and

85

condition of breeding adults and in spawning site selection (Chapter 2; Smith and Baker
2005).
Mortality and sources of loss

During the spring of 2005, we quantiﬁed the magnitude and sources of lake
sturgeon egg mortality using replicated exclosure boxes of varying mesh sizes.
Treatments of different mesh sizes were designed to simulate an increasing probability
(small to large mesh) that eggs would be encountered and consumed by increasingly
diverse range of predators (interstitial vs. epibenthic) or removed from the incubation
substrate (e.g., de-adhesion owing to higher water velocities). Sources of mortality
leading to egg removal were collectively categorized as exogenous loss/removal. Water
velocity and co-variation in other factors such as dissolved oxygen concentration
associated with varying ﬂow conditions also differed as a function of treatment (large to
small mesh). In addition to altering water ﬂow, exclosure treatments may also have
lowered levels of light exposure but this was not quantiﬁed.

Exclosure boxes were constructed using welded heavy metal rebar of sufficient

weight to withstand high velocity water currents and stream debris (Figure 4.1). Metal

rectangular boxes (0.093 m2, 0.305 m high) were enclosed with metal cloth of three

different sizes (Large 5.08 cm, Small 1.27 cm and Fine 2 mm). A fourth box was left
open to simulate the ambient stream environment (unimpeded ﬂow and predator access).
Our working hypothesis was that egg mortality rates experienced across mesh size
treatments would be comparable but the sources of mortality would vary signiﬁcantly
(i.e., compensatory mortality). Speciﬁcally, we predicted that treatments would result in:

1) Open Cage — high rates of exogenous loss and a low incidence of developmental

86

mortality, 2) Large Mesh — moderate levels of exogenous loss and low mortality due to
sources that arrest development, 3) Small Mesh — a reduction in exogenous loss but
higher incidence of developmental mortality, and 4) Fine Mesh — low probabilities of
exogenous removal but high rates of developmental mortality because of low water
velocity.

Eggs were obtained from spawning lake sturgeon captured on the UBR Two-

hundred eggs were counted, fertilized, and allowed to adhere to one brown circular

(0.073 m2) porous ﬁlter (3M Worldwide, Inc. Bufﬁng and Polishing Pads) in each

exclosure. This egg density is within the range commonly observed in the UBR under
natural spawning conditions (Chapter 2). Filters were randomly assigned to an exclosure
mesh size treatment. Four different treatments were then randomly placed (i.e., blocked
together) on a heavy metal rebar base (Figure 4.1), secured with plastic tie-downs and
lowered to the bottom of the stream By creating homogenous blocks (i.e., placing all
four treatment in close proximity), we hoped to remove the effect of important nuisance
variables inherent to this study, namely the range of abiotic conditions (e. g., depth, water
velocity and substrate size) present in sections UBR where lake sturgeon naturally
reproduce. Blocks were placed at least 5 m apart and eight replicates of each group (i.e.,
block) of four treatments were deployed simultaneously for 5 days.

Exclosures were cleaned of stream debris daily to maintain consistent water
velocities. Flow conditions inside each replicate were measured using a March-
McBirney ﬂow meter at the end of the experiment. We also avoided attributing hatching
to egg mortality by terminating the experiment once larvae became active inside the

embryo and chorions softened (120 hours post fertilization). Numbers of viable and dead

87

eggs were counted every 24 hours. Viable eggs (i.e., those exhibiting proper
development) were identiﬁed by the formulation of the blastodisc and thickening at the
animal region pole (Dettlaff et al. 1993). Eggs that died between observations could
therefore be partitioned into different mortality sources including developmental failure
(i.e., number of new dead eggs) and exogenous loss (i.e., decline in viable egg abundance
subtracted from new developmental failures).
E ﬂeet of predator density, predator size and substrate size

A complementary replicated laboratory experiment was conducted in 2006 and
2007 to further document the importance of predator density and stream substrate on egg
mortality. A single predator species (rusty crayﬁsh Orconectes msticus) was used along
with substrates commonly found at spawning locations. Crayﬁsh predators were
collected from sections of the UBR where spawning had not occurred and therefore
previous experience encountering, handling, or foraging on lake sturgeon eggs was
unlikely. Predators were housed in ﬂow-through tanks for approximately 48 hours before
used and were not fed while in captivity. Eggs were obtained ﬁ'om adult lake sturgeon
and temporally stored in incubation trays ﬁlled with river water at a rate of approximately

15 l/min.
Fiberglass rectangular tanks (bottom surface area = 2.8 m2) were divided into

three sections of equal size, but not partitioned so that crayﬁsh could freely access all
tank sections. Tanks were ﬁlled with equal and regulated amounts of water ﬁ'om the
UBR. Three substrates were randomized within each tank prior to the start of a trial (1
per section) including: 1) large (mean 63.5 :1: 18.4 mm), 2) medium (mean 24.4 i 5.87

mm) and 3) open (mixture of sand and ﬁberglass bottom). A predator density (5, 10, 15

88

or 30 individuals) was randomly assigned to a raceway and crayﬁsh were placed in equal
proportions in each substrate. After 15 minutes of acclimation, 200 fertilized eggs (i.e.,
600 per tank) were randomly scattered across each section. After 48 hours, predators
were immediately removed and measured for carapace length (mm). Tank sections were
then completely partitioned to minimize egg drift, substrates were sifted and remaining
lake sturgeon eggs were counted.

Confounding effects of crayﬁsh behavior and habitat preference/usage on egg
survival (particularly as it relates to substrate size) were partly accounted for in two ways.
First, we recorded the proportion of crayﬁsh occupying each substrate type and the
general activity patterns (i.e., foraging, searching) of predators at discrete intervals (every
4 hours) during diurnal and nocturnal periods. Second, a small complementary laboratory
experiment was conducted in 2006 where both predator movement and predator density
were ﬁxed across substrates. In this experiment, only one of the three substrate categories
previously described were placed in individual ﬂow-through tanks (1 m X 0.5 m X 0.5 m)
provided with ﬁltered water taken from the UBR Flow-through tanks were blacked out
so that crayﬁsh movements and foraging behavior were not affected by outside
movements. Five novel crayﬁsh predators (also housed for 24 hours and unfed) were
randomly chosen and placed in each tank. Two-hundred eggs were released over the
substrate after 15 minutes of predator acclimation. Each trial was conducted for 48 hours
after which time the crayﬁsh were removed and measured for total carapace length (mm).
Substrates were then removed from each tank, sifted and remaining eggs were counted.

STATISTICAL ANALYSES

89

All analyses were performed using PROC MIXED in SAS Version 9 (SAS
Institute 2003). Data from the ﬁeld-based experiment were analyzed as a randomized
complete block design with repeated measures across time in the context of a general
linear mixed model (GLMM) (i.e., a model that contains both ﬁxed and random effects).
The GLMM is commonly deﬁned as y = XB +Zv + e where X is a known design matrix
for the ﬁxed effects, Z is a known design matrix for the random effects, B and v are
vectors of unknown ﬁxed and random effects parameters and e is a vector of unobserved
random errors. Mixed modeling is the preferred approach for repeated measures designs
largely because 1) the user can deﬁne a number of alternative covariance structures for
experimental data in which observations are likely correlated (i.e., traditional ANOVA
assumes independent errors) and 2) by treating a variable as a random effect (block in
this case), and thus estimating variance for a population distribution plus a mean,
inference can be generalized and extended to all possible levels of the effect.

We used a GLMM to examine the effect of time and exclosure treatment on three
separate dependent variables (y) including: a) percent of eggs remaining at each
observation, b) percent of total mortality due to developmental failure and c) percent of
mortality due to exogenous loss. Each dependent variable was modeled separately as:

y = u + Time + Treatment + Treatrnent*Time + Block + a

where the ﬁxed main eﬁ‘ect factors in the model were exclosure treatment (Ho: no mean
difference), time (Ho: no change overtime), and the interaction between treatment and

time (H0: no difference in the shape of the response to treatment overtime (i.e., no

difference in slopes)). u was the general location parameter of the model (i.e., the mean)

90

and 8 represents the residual variance. Block was treated as a random effect and thus
represents a random sample of all potential experimental sites in the spawning area. An
auto-regressive error of the ﬁrst order, AR( 1 ), was used to model temporal
autocorrelation among observations collected across time periods. AR(l) denotes that
correlations between observations between adjacent time periods are larger than for those
made farther apart, a common phenomenon in repeated measures designs (Littel et al.
1996). All pair-wise comparisons were investigated using least square means and post
hoc differences were computed using Tukey’s HSD (a = 0.05). All dependant variables
were arcsine square root transformed to meet normality assumptions. Differences in
average ﬂow rates between exclosure treatments were tested separately using traditional
analysis of variance (ANOVA) in SAS (PROC GLM).

Data from the primary laboratory experiment was analyzed as a whole-plot split-
plot design also in the context of a GLMM. Here, predator density (i.e., 5 (n = 9), 10 (n =
9), 15 (n = 18) and 20 (n = 12)) was treated as the whole plot factor (i.e., a random
crayﬁsh density was applied to each tank) and substrate category (i.e., large, medium and
open) was treated as the subplot treatment factor as all three substrates were
simultaneously evaluated in a single simulated system. The proportion of eggs remaining
(arcsine square root transformed) aﬁer 48 hours (y) was then modeled as:

Y = u + Density + Substrate + Substrate*Density + Predator Size + 3

where the primary ﬁxed factors were predator density (Ho: no effect of predator density),
substrate category (Ho: no effect of substrate), and the interaction between density and

substrate (Ho: egg consumption for each density is constant across substrates). u was the

91

general location parameter of the model (mean) and 8 represents the residual error
variance. Average predator size was used as a covariate to partly control for any
unwanted and non-random differences in the crayﬁsh predators selected for individual
trials. Although not a mixed model (i.e., no random effect), the residual maximum
likelihood (REML) estimators currently employed with the Mixed Procedure in SAS
(opposed to Type III sums of squares) are most appropriate for estimating effect size in
experiments with unequal replications. Pair-wise comparisons were investigated using
least square means and post hoc differences were computed used Tukey’s HSD (a =
0.05).

Lastly, a traditional ANOVA in the form of a general linear model (PROC GLM)
was used to test for differences in the proportional use of substrates by crayﬁsh predators
in the primary laboratory experiment. A general linear model AN OVA was also used to
address the effect of substrate size on egg survival in the complementary laboratory
experiment where predator density was ﬁxed (5 individuals) and movements were
restricted to individual tanks. In this model, substrate category was the only ﬁxed effect
and predator size was again used as a covariate. In both cases, pair-wise comparisons of
arcsine square-root transformed proportions were calculated for each main effect using
least square means and tested for statistical separation with post-hoe tests (a = 0.05).

RESULTS
Magnitude of egg mortality and the components of loss

Exclosure treatments exposed eggs to different predator guilds as evidenced by

size- and species-speciﬁc differences in colonizing benthic organisms including

plecoptera (stoneﬂies), ephemeroptera (mayﬂies), diptera (chironomids), decopoda

92

(crayﬁsh) and perciformes (darters) (personal observation). Exclosures also contributed
to signiﬁcant variation in stream water velocity. As expected, water velocity increased

with increasing mesh size, and with the exception of the Fine Mesh and Small Mesh

treatments, differences were signiﬁcant (F 3’ 23 = 25.5, P < 0.05; Open: 0.60 :1: 0.076 m/s,

Large Mesh: 0.30 :1: 0.04 m/s, Small Mesh: 0.12 i 0.021 m/s, Fine Mesh: 0.12 d: 0.013
m/s).

Egg mortality was high during the ﬁrst 120 hours of incubation. In total, 5793
eggs (91% of the starting abundance) died before hatching with an average of 36 (:t 48
SD) mortalities occurring between each observation period Mortality was also
inconsistent during incubation; i.e., mortality remained below 2% of the original
abundance during the ﬁrst 48 hours post fertilization (4 individuals/day), increased
substantially to 57% between 72 and 96 hours (133 individuals/day on average) and
eventually tapered to less than 10% (20 individuals/day) prior to hatch. Time to 50%
mortality occurred between 72 and 96 hours after fertilization. Substantial mortality

leading to rapidly declining levels of viable egg abundance was indicated by a signiﬁcant

main effect of time in the mixed effects model (F4’ 101 = 476.9, P < 0.001).

There was no signiﬁcant difference in average mortality/survival between
exclosure treatments (Average mortality i Range: Fine Mesh = 89 d: 22%, Small Mesh =

92 :1: 19%, Large Mesh = 90 :h 21%, Open = 89 i 12%) due in part to the large variability
in the magnitude of loss (Treatment effect: F 3, 31 = 5.7, P = 0.374). Mortality however

was responsive to exclosure treatment as indicated by a signiﬁcant time-by-treatment

interaction (i.e., means were signiﬁcantly changing overtime but at different rates) (F 1 2,

93

101 = 5.7, P <0.001). Speciﬁcally, time-dependant pair-wise comparison illustrated

signiﬁcantly lower levels of loss (depicted in Figure 4.2a as the percent of egg surviving)
in the Open and Large Mesh treatments from 48 through 96 hours post fertilization.
There was no evidence for signiﬁcant differences in mortality among exclosure
treatments either before 48 or at 120 hours after fertilization.

Total mortality was portioned into two sources including exogenous removal and

developmental failure (Figures 4.2b and 4.2c). Mortality attributed to each source

increased signiﬁcantly between observations through time (i.e., Exogenous removal: F3,

61:60.6, P <0.001; Development failure: F3, 61:45.1, P <0.001). Developmental failure

(N=4848, 84%) was the dominant source of total mortality across treatments compared to
removal (N = 946, 16%). The relative magnitude of each mortality source however varied
depending on the number of days into incubation (Figures 4.2b and 4.2c). For instance,
100% (N = 272) of the total mortality during the ﬁrst 48 hours prior to fertilization was
attributed entirely to exogenous removal. Developmental mortality increased
substantially between 48 and 96 hours, accounting for 89% of the total loss or 29
(average) new mortalities per day during this developmental stage.

Higher levels of exogenous removal and a lower incidence of developmental
mortality were predicted to occur with increasing exclosure mesh size. Signiﬁcant mean

differences in source-based mortality between exclosure treatments (Exogenous removal:
F 3, 43 = 3.9, P < 0.0]; Development failure: F 3, 43 = 4.8, P < 0.005) and signiﬁcant mean
pair-wise comparisons through time were consistent with these expectations (i.e.,

treatment-by-time interaction: Exogenous removal: F 3, 61=45.7, P = 0.08; Development

94

failure: F 3, 57 =59.7, P = 0.17). Pair-wise evaluation for instance illustrates that

developmental failure was a signiﬁcantly smaller component of total mortality during
incubation as mesh size and water velocity increased (Figure 4.2b). Further, levels of
exogenous removal were consistently, and in many cases signiﬁcantly greater, in
treatments relatively open to ambient conditions and thus more susceptible to water
currents as well as larger macro-invertebrates and ﬁsh (Figure 4.2c).

The role of predator density, predator size and substrate
We conducted 16 laboratory trials with four predator densities. Approximately 7718 eggs
(80% of the total) were consumed by 255 crayﬁsh predators over 48 hours (17-104
eggs/predator on average). Mortality was generally higher as predator density increased

and substrate size decreased (Figure 4.3). Predator density was a signiﬁcant predictor of
egg mortality (F 3, 11 = 6.9, P < 0.005) which was expected to increase by approximately

' 57%, based on least square mean estimates, for every increase in predator density.
Substrate size did not explain a signiﬁcant amount of the variation in mortality (F 3, 24 =
0.9, P = 0.39). Interpretation of this main effect however was partly confounded given the

presence of a marginally signiﬁcant density-by-substrate interaction (F 6, 24 = 3.5, P =

0.05), rejecting the null hypothesis that mortality for each density was consistent across
substrates. Pair-wise evaluation of simple effects (i.e., substrate within each density)
indicates signiﬁcantly lower mortality/higher survival in the Large Rock treatment
compared to the Gravel and Open treatments when predator densities were at the
extremes (5 and 30 individuals). At intermediate densities (10 and 15 individuals),

substrate had little or even the opposite effect on egg mortality (Figure 4.3).

95

Additional analyses/experiments enhanced and supported these general results.
For instance, predator movements in the main laboratory experiment were non-random
with respect to substrate type. Although mortality was lowest in Large Rock treatments,

individuals spent signiﬁcantly more time (32 out of 48 total hours on average) associated

with these areas than other substrates tested (F 2, 6g= 28.1, P <0.001). Levels of predation

were also consistently high and concordant when predator density and movements were
ﬁxed (i.e., 5 individuals). For instance, a total of 2495 eggs representing 69% of the total

(65% in the main experiment when n=5) were consumed by the 90 crayﬁsh predators (28

eggs/ predator). Substrate size (F 2, 15 = 3.85, P = 0.045) was also a signiﬁcant predictor

of egg mortality in this complementary laboratory study and mortality was signiﬁcantly
lower in the Large Rock (62 %) and Gravel (63 %) treatments relative to conditions

lacking any benthic heterogeneity (Open = 89 %). Finally egg consumption was related to

predator size in both laboratory experiments (F 1, 11 = 15.0, P < 0.01). Tanks with larger

I crayﬁsh predators on average had generally consumed more eggs (Figure 4.4).
DISCUSSION

Understanding the processes that bound species between limits of abundance
requires precise estimates of mortality, especially during early life stages. However, early
life mortality of many organisms is commonly overlooked, underappreciated and seldom
quantiﬁed for many taxonomic groups, especially during the egg stage in ﬁshes. Here, we
estimate mortality of lake sturgeon eggs to average 91% during incubation in natural
stream environments. Further, we estimate substantial variation in mortality across

incubation conditions comprised of differences in ﬂow velocity and susceptibility to

96

predation to range from 75-97%, translating into nearly a 10 fold difference in the
number of larvae produced at hatch.

Our estimates are generally concordant with those reported for other sturgeons
(83 — 99%) (Nichols et al. 2003, Johnson et al. 2006; Caroffino et al. 2010) and for other
broadcast spawning species that have unprotected demersal eggs including white sucker
(97%; Scott and Crossman, 1973 cisco (99%; Bouwes and Luecke 1997), walleye (87%;
Johnson 1961) and smelt (99%; Rupp 1965). However, levels of egg mortality found in
this study where considerably higher in comparison to species that construct nests (i.e.,
bass and salmonids), provide parental care, have protective mechanisms or use vegetation
for attachment (reviewed in Dahlberg 1979). For example, Clady (1975) reported over
95% survival of naturally produced yellow perch eggs which are laid in mass, protected
by unique gelatinous envelopes and attached to vegetation.

Developmental mortality

Developmental mortality was the dominant source of loss during incubation
representing 84% of the total mortality across exclosure treatments on average. Several
explanations may be offered to explain this result. First, developmental failures may have
been a ﬁmction of endogenous properties of eggs that negatively interacted after
fertilization. For instance, the ability of ﬁsh embryos to survive in nature depends on both
nutritional (e. g., yolk volume) and genetic effects (e. g. DNA and hormones) (Brooks et
al. 1997). Thus, if properties such as adult condition, the timing of spawning (degree of
ripeness) and parentage are not matched well, egg quality will be reduced and may
decrease probabilities of survival (Brooks et al. 1997). Developmental mortalities may

have also occurred because of polyspermy, or simultaneous fertilization by multiple

97

spermatozoa (Gilkey 1981). Polyspermy is found in many ﬁsh species and is especially
prevalent in sturgeon because of numerous large openings (i.e., micropyles) in the egg
envelope (Dettlaff et al. 1993; Linhart and Kudo 1997). Eggs fertilized by multiple
sperm have additional chromosomes and centrioles and thus cause developmental defects
and mortality by the ﬁrst cleavage division (Gilbert 1997). Polyspermy may have been
enhanced during our experiment due to artiﬁcial fertilization (Psenicka et al. 2010)

Developmental mortality on average was also signiﬁcantly greater as mesh size
and water velocity decreased. Water velocity is often related to lower levels of dissolved
oxygen (DO) and D0 is directly tied to spawning site selection in ﬁsh and is considered
the primary factor in determining the fate of eggs developing in natural conditions
(Y oungson et al. 2004). Thus, a portion of developmental mortality here could be a result
of suboptimal DO concentrations during incubation. Further, the delayed onset of
developmental mortality observed in this study (i.e., after 48 hours after incubation) may
also reﬂect developmental stages with an increasing demand for DO. Peaks in D0
consumption and periods of susceptibility to mortality are common in ﬁsh embryos,
especially during the formation of the animal pole and during hatching (Hamor and
Garside 1977). However, in addition to DO, it is also possible that toxins (e.g., PCBs),
mechanical disruptions that rupture eggs, ﬂuctuation in water temperature and ultraviolet
solar radiation were exogenous contributors to endogenous developmental malformations
(Parsley et al. 2002). However, these are unlikely in the UBR and in our experimental
conditions.

Eggs recorded as developmental mortalities in the ﬁeld were also covered with

microbes, a phenomenon commonly observed in ﬁsh culture practices and more recently

98

in natural settings for several ﬁsh species including sturgeon (Kempinger 1988; Parsley et
a1 2002; Rach et al. 1995; Kitancharoen et al. 1997). Anders and Beckman (1992) for
instance found that 2-45% of white sturgeon eggs collected were infected with water-
borne microbes. Microbial infection observed here and elsewhere may be a result of
colonization post-mortality (i.e., eggs died due to something else). However, signiﬁcant
differences in levels of infection were also noted among treatment groups in this study.
Differences in infection rates of eggs have also been found to be a function of water
temperature (Knotek and Orth 1998), egg density (Cote and Gross 1993) and water
velocity (Anders and Beckman 1992). Thus, it is plausible that developmental mortality
is partly driven by wide-spread predatory-type infection by aquatic microbes that kill
eggs by preventing oxygen transport through the perivitelline space. If so, microbial
infection may have severe ecological consequences for sturgeon by placing constraints on
seasonal levels of recruitment (Knootek and Orth 1998), altering behavioral and
reproductive decisions (Knouft et al. 2003; Takahashi et al. 2004) and inducing hatching
(W edekind 2002). Interestingly, microbial infection has lead to the evolution of
antimicrobial properties of eggs (Paxton and Willoughby 2000) and antimicrobial
substances that are conferred to eggs by adults (Giacomello et al. 1996) in several ﬁshes.
Exogenous removal

Exogenous sources leading to the removal of eggs in the ﬁeld accounted 16% of
the total mortality in this study. Further, differences between exclosure treatments
indicated higher levels of mortality as a function of increasing water velocity. A portion
of removal therefore may be a function of currents that dislodge and translocate embryos.

The relative impact of water velocity on ﬁsh egg mortality, particularly when reaching

99

critical sheer stress, has been well documented in natural conditions (Lapointe et al.
2000; Vending-Schwank and Livingstone 1994). Mortality is signiﬁcantly elevated for
translocated individuals due to sediments that bury eggs and create hypoxic conditions
(V entling-Schwank and Livingstone 1994) or mechanical stressors that rupture the
plasma membranes encasing embryos (Bunn et al. 2000). However, probabilities of
secondary drift can be reduced by features of the stream that tend to mediate water ﬂow
including substrate size (V ending-Schwank and Livingstone 1994). Comparable
estimates of egg mortality show that the magnitude of loss due to removal is smaller than
reported here. For instance, Mills (1981) and Bouwes and Luecke (1997) estimated
translocation to be a small fraction (2%) of mortality in dace and cisco, generally citing
abiotic factors that reduced the impact of water currents on egg de-adhesion. LaHaye et
al. (1992) also found that a small number of lake sturgeon eggs in drift were in advanced
stages of development which is consistent with the transportation hypothesis. However,
Fitzsimons et al. (2007) reported that physical disturbances generated by water ﬂow
contributed to higher egg mortality than other plausible sources in lake trout (i.e.,
predation) combined.

Another explanation for exogenous removal is predation by interstitial predators.
Several factors support this hypothesis including a signiﬁcantly consistent elevation in
removal during incubation across treatments as well as signiﬁcantly higher levels of
mortality in the Large Mesh and Open treatments. Predation on ﬁsh embryos often
represents a signiﬁcant source of mortality and is a strong regulatory component of ﬁsh
populations because of the broad diversity of benthic ﬁsh and invertebrate predators with

acute visual and chemosensory cues that allow egg predators to forage effectively within

100

the substrata of spawning sites (Bailey and Houde 1989; Bouwes and Luecke 1997;
Dittrnan 1998). In broadcast spawning ﬁsh species, predation is predicted to have severe
ecological consequences through depensatory population dynamics (i.e., Allee effects),
especially if populations are reduced past thresholds (Gascoigne and Lipcius 2004).
Empirical evidence also implicates predation on eggs as the primary cause of regionally
poor recruitment and in the decline (or failed reintroductions) of many ﬁsh species of
conservation concern (Savino et al. 1999).

Although signiﬁcant levels of predation on ﬁsh eggs for species including
sturgeon have been documented (Miller and Beckman 1996; Kempinger 198 8), predation
in our ﬁeld-based research represented a small proportion of the total mortality relative to
developmental sources. However, levels of predation may have been underestimated
because of time lags in the colonization of exclosure treatments. Fortunately, the impact
of predation during the egg stage can also be gauged by the number of eggs consumed by
a single benthic predator (DeBlois and Leggett 1991). In laboratory trials, we determined
that average consumption by crayﬁsh predators ranged from 17-104 eggs over 48 hours.
Similar research reports much lower consumption rates (mean individual consumption
ranged from 0.1 to 5.5) on lake and rainbow trout eggs (Ellrott et al. 2007), levels thought
to hamper restoration activities implemented for natural populations of these species
(e. g., Fitzsimons et al. 2006). Given the high rates of consumption found in this study,
combined with much lower levels of interstitial predation found in others, we believe
predation may indeed represent a major source of mortality during the egg stage in lake

sturgeon.

101

Our laboratory research ftuther indicates that levels of predation in lake sturgeon
are strongly inﬂuenced by both abiotic and biotic conditions. Speciﬁcally, we found
higher levels of egg mortality as the density of crayﬁsh predators and substrate size
increased (Figure 4.3). Aspects of the aquatic environment that inﬂuence consumption
rates of embryos including predator type, predator density, river substrate composition
and water temperature have been studied extensively (Fitzsimons et a1. 2006). Levels of
egg consumption generally vary depending on predator foraging guilds (Savino et al.
1999) and decline with increased predator density (Ellrott et al. 2007) and increasing
substrate complexity because of the formation of deep interstitial spaces that reduce egg
delectability (Chotkowski and Marsden 1999; Biga et al. 1998). We believe the
composition of river substrates is critical for protecting lake sturgeon eggs ﬁ'om predators
given signiﬁcantly lower levels of mortality in substrates of larger size, regardless of

whether predators spend signiﬁcantly more time associated with these preferred habitats.

102

SYNTHESIS AND CONCLUSIONS
Chapter 1

Sturgeon migration should be viewed as a continuum whereby individuals
respond to exogenous cues before and after river entry and process different sources of
information leading up to and during reproduction. Further, the relative importance of
variables varied during different stages of migration. The inrportance of temperature and
discharge to sturgeon behavior during each stage of migration indicates that perturbations
affecting stream environments may pose a threat to the viability of sturgeon populations.
Speciﬁcally, changes in river discharge and water temperature above or below status quo
conditions, due to hydroelectric operations for example, may cause changes in exogenous
cues which likely elicit non-adaptive responses. The consequences of delayed migration
are potentially severe and range from a reduction in energy reserves that exceeds the
capacity for reproductive success, a decrease fertilization success by increasing the time
spent in pre-spawning condition, uncoupling arrival and breeding (Ahola et al. 2004) or
reducing egg and larval survivorship.

Stress from continued global climate change in the form of increased/decreased
magnitude and timing of spring ﬂows, increased in water temperatures and habitat
fragmentation are likely to threaten ﬁrture population viability of this species (Carpenter
et al. 1992; Stefan and Sinokrot 1993). Most importantly, our study suggests that river
discharge and water temperature cues will decrease the likelihood that individuals will be
able to coordinate spawning activities with periods of the lunar cycle we believe are
important for enhancing larval survival. Thus, the choices of individuals, guided by

exogenous cues, are no longer adaptive (i.e., ecological traps; Schlacpfer et al 2002;

103

Stanford et a1 1996). Recent work has also shown that individual lake sturgeon have a
signiﬁcant tendency to reproduce at different times and different places within the same
river system and that these reproductive behaviors are highly repeatable (F orsythe et al.
2010). If repeatability of spawning time indicates non-zero heritability, each group may
have evolved unique life history adaptations (e.g., egg size) and requirements (i.e.,
incubation temperatures) related to offspring growth and survival (Hendry and Day
2002). Conﬂicting exogenous cues may also increase probabilities of inter-group
breeding, thereby disrupting adaptive life history or genetic diversity (Sugg et al. 1996
Crandall et al. 2000). Data presented herein provide a means of projecting the rate and
direction by which changes in environmental regimes will act upon migratory and
spawning behavior as a function of spatially and temporally changing environments (e. g.,
Williams et al. 2009).
Chapter 2

High repeatability in spawning time and spawning site selection indicates that
individuals have a predisposition to spawn associated with exogenous environmental and
lunar cues that covary with time within a spawning season and thus both may be heritable
behavioral traits. These data, along with recent work demonstrating evidence for the co-
evolution of early life traits shown to predictably vary as a ﬁinction of spawning time
(i.e., body size; Crossman 2008), suggest adaptive phenotypic and genetic differentiation
may develop over ecological time frames in different spawning groups. Adaptation to
conditions at such a ﬁne micro-geographic scale can allow a population to maintain
variability which is increasingly irrrportant to facilitate adaptation in response to future

anthropogenic change (Hilbom et al. 2003).

104

Repeatability for spawning time may also indicate future challenges for sturgeon
populations. For instance, repeatable assortative mating indicates that effective breeding
population size will be lower because not all individuals reproduce at the same times or
even the same year. Minimal mate avoidance due to aggregate breeding and an increased
life span would tend to increase the probability of encountering and breeding with related
individuals. Decreased effective population size will potentially lead to more rapid
accrual of coancestry. Heritability for spawning ﬁrrther indicates that sturgeon are
susceptible to human sources of selection such as angler harvest and hatchery practices
that can inadvertently shift migratory time and reduce phenotypic variability in spawning
time (Quinn et al. 2000; Quinn et al. 2002; Quinn et al. 2007). Streams and rivers are also
rapid integrators of terrestrial and atmospheric processes and thus are greatly impacted by
human disturbance (Williamson et a1. 2008). The gradual warming of rivers due to global
climate or regulatory activities that modify spawning habitats may risk decoupling
important associations between the timing of reproduction and environmental conditions
and co-evolved traits linked to offspring survival (Portner and Farrell 2008). The
disruption of environmental regimes that have given rise to coevolved traits and
temperature-dependant developmental trajectories (Kingsolver and Huey 1998) and
genetic responses to alternation of the chronology of thermal regimes are expected to be
reﬂected by the loss of variation within generations of animal populations (Angilletta et
al. 2006).

Studies reporting reproductive isolation and evidence for adaptive divergence
among spawning groups have typically been conducted over spatial scales large enough

where differences in environmental regimes are expected. Studies which have attempted

105

to document the degree of spatial and temporal segregation and local adaptations of
breeding groups at comparatively small scales (e.g., over distance of < 2 km as reported
herein) have been hampered by systems with relatively homogenous abiotic structure
(Doctor and Quinn 2009) and lack of long-term data. Although the section of river used
by UBR lake sturgeon contains several spawning locations that are relatively close
together, subtle variation in physical features including substrate size, water depth, and
velocity inﬂuence the natural deposition of embryos and inﬂuence levels of early life
mortality (Chapter 3). Further, signiﬁcant variation in environmental conditions (e. g.,
water temperature) is common between ﬁsh spawning only a few days apart. Thus, we
offer that studies attempting to quantify locally adapted traits at ﬁne micro-geographic
scales should be encouraged to increase the ability to predict population responses.
Chapter 3

Our study reinforces the need to employ appropriate statistical methods in order to
properly document relationships between response and predictor variables. The primary
goal of this study was to use data collected on a unique migratory species to illustrate the
need to account for spatial dependencies when possible, especially in spatially complex
environments such as streams and rivers. Data provide compelling evidence that common
assumptions of spatial models (e. g., isotropy) are often violated. Alternative analytical
methods as described here provide a rigorous ﬁamework to model spatially complex
ecological problems. Ecological applications including modeling spatially explicit data
within and between time periods provide a powerful approach to identify exogenous

features associated with recruitment of stream ﬁshes.

106

Analyses also have important applications for lake sturgeon and for other species
sharing a similar reproductive ecology. For example, lake sturgeon, like many other
ﬁshes, have suffered dramatic population declines largely because of anthropogenic
activities that have restricted, reduced or destroyed spawning areas (Duke et al. 1999;
Opperman et al. 2005; Rochard et al. 1990) In response, managers have attempted to
mitigate the effects of habitat loss by supplementing historically important and degraded
spawning areas with new gravel or by creating new habitats that mimic natural areas in
form (Johnson et al. 2006; Kondolf et al. 1996; Zeh and Donni 1994). Indeed, the
addition of gravel to areas has been shown to promote successful reproduction for some
species (e. g., Atlantic salmon and Brown trout), even when constructed areas are
suboptimal relative to natural spawning grounds (Barlaup et al. 2008). However, little
information has been available on the quantiﬁable features (i.e., gravel size, placement
and ﬂow requirements) of natural spawning habitats that sturgeons use for reproduction
leaving little guidance for restoration.

We believe our summaries of spawning habitats used by UBR lake sturgeon can
inform this type of management action. Still, while our data and others suggest that lake
sturgeon will likely use man-mad structures (Caswell et al. 2004; Johnson et al. 2006),
we caution that spawning site enhancement will have a negligible impact on recovery
through recruitment until all exogenous variables affecting rates of survival are identiﬁed
and thus can appropriately mimic function (Hester and Gooseff 2010) as well as form.

Chapter 4
Egg mortality after fertilization in lake sturgeon represents a complex

compensatory-type (i.e., decreasing percent of exogenous removal leads to increased

107

levels of developmental mortality) relationship between multiple sources. Important
relationships between the sources themselves and environmental conditions indentiﬁed
here may serve to provide projections of loss that can guide conservation/management
activities including construction, restoration and modiﬁcation of spawning locations. For
example, assuming levels of microbial infection are site-speciﬁc and stream microbes
infect and subsequently kill otherwise healthy ﬁsh eggs (Wedekind 2002), individuals
deposited in locations, or at times with a greater diversity and abundance of microbes,
will likely experience higher rates of infection. Further, eggs deposited in riverbed
locations that facilitate the delivery of dissolved oxygen will experience lower morality
(Greig et al. 2007). Areas consisting of relatively large substrate size will serve to
mitigate translocation by reducing levels of critical sheer stress that lift eggs ﬁrm the
stream bottom and will protect individuals through the complexity of interstitial voids.
The magnitude of loss and the sources of mortality may also have important
evolutionary/ecological consequences. At the population level for instance periodic ﬁsh
species (i.e., delayed maturation, large body size, high fecundity and low investment per
offspring) such as lake sturgeon are predicted to exhibit substantial inter-annual variation
in levels of annual recruitment tmrelated to the abundance of spawning adults
(W inemiller 2005). Several explanations for the lack of density-dependant compensation
in the number of offspring produced have been offered including mortality during early
life stages, stochastic abiotic/biotic variation (Myers 1998) and the availability of food
for offspring (Cushing 1990). Preliminary empirical spawner-recruitrnent data collected
for lake sturgeon populations support this prediction (F orsythe et al unpublished data).

Further, our research showing high and variable egg mortality across ecological gradients

108

supports the relative contribution of early life stages to population regulation and
reinforces the ecological signiﬁcance of mortality during the egg stage in long-lived ﬁsh
species.

Bailey and Houde (1989) predicted that variation in a single source, predation,
resulted from spatial and temporal variation in predator-prey distribution, predator
abundance and the availability of alternative prey would ultimately lead to variation in
levels of recruitment. Our data also supports this prediction. However, we also offer that
heterogeneity in all abiotic and biotic factors leading to variation in mortality of embryos
may contribute to variation in levels of sturgeon recruiting to the population both within
and between spawning locations, across segments of populations that breed at different
times and annually. This prediction is based on relationships presented in this paper as
well as quantiﬁed differences in water velocities, depths and substrate sizes of spawning
habitats (Chapter 3) often correlated with variation in the abundance and distribution of
benthic egg predators (Crowl et al. 1997), water temperatures, the direction and velocity
of water currents and surface—subsurface DO exchange (Boulton et al. 1998) and the
abundance and diversity of the aquatic microbial community (e.g., bacteria and fungi).

Finally, in times of human-mediated change, early life morality sources not
common to the evolutionary history of ﬁSh populations including pollution, microbial
diseases or parasites, the alternation of ﬂow regimes through regulated river management
and the introduction of exotic interstitial predators are likely to occur. Variation in abiotic
and biotic mortality sources has been shown to shape population or breeding segment-
sDﬁciﬁc adaptations during the egg stage including egg size, resistance to microbial

infection and the timing of transition to the larval stage (Quinn et a1. 1998). Philopatric

109

tendencies of lake sturgeon and repeatability for spawning at the same time and place
may exacerbate these differences (Chapter 2). Yet, these questions have not yet been
explored for sturgeon populations. Knowledge of the ‘biocomplexity’ that accompanies
the evolution of diversity during the egg stage will be critical for understanding sturgeon
sustainability in the face of spatial and temporal heterogeneity in environmental
conditions exacerbated by human activity and climate change (Hilbom et al. 2003).

Overall synthesis and wrap-up

Migration for the purpose of reproduction is a behavior that sets into motion a
cascade of processes that often have considerable ﬁtness consequences such as early life
mortality. For many ﬁsh species, natural selection has led to the evolution of adaptations
that allow individuals to select the best time and location (i.e., spawning site) for
reproduction. Research presented in this dissertation suggests that individual lake
sturgeon are highly repeatable for spawning time and for spawning location (Chapter 2).
This ﬁnding suggests the importance of mechanisms that have a physiological (i.e.,
olfaction or response to time-related cues) or genetic basis and thus promote a degree of
consistency in the choice of spawning time and location within and across seasons.
Future studies should seek to address the relative role of these mechanisms and the
degree of reproductive isolation resulting from repeatability in the UBR and elsewhere.
However, variation was also observed in both spawning time and location at the
individual level (Figures 2.4 and 2.5). Further, results ﬁom Chapter lindicate non-
random associations between occurrences of groups of individuals (that have a tendency
to migrate at the same time) at all stages in the migration continuum including river entry,

upstream movement and arrival to spawning sites with numerous exogenous variables.

“0

Variation in individual reproductive time and location, therefore, may be a function of
population-wide responses to exogenous cues that may confer survival advantages to
offspring (i.e., lunar cycle). Research seeking to tie egg and larval survival to exogenous
conditions selected by adults are encouraged in other systems. In Chapters 2 and 3, we
see that adult lake sturgeon have a tendency to select different spawning areas as a
function of spawning time (i.e., early spawning individuals select locations further
upstream than those spawning later; Figure 2.6) and that areas selected for spawning
have signiﬁcantly different abiotic characteristics including substrate size and water ﬂow.
I believe future research would proﬁt ﬁom seeking to determine whether spawning site
selection is mechanisticlly tied to seasonal differences in stream habitat quality,
physiologial adaptations (i.e., olfaction) or whether selection has a genetic basis.
Research seeking to tie spawning site selection to adaptive phenotypic differences of
offspring at hatch (i.e., larval size) would also be proﬁtible. Despite these differences,
however, spatially-explicit patterns in egg deposition and the environmental covariates
(e. g., substrate size) associated with increased numbers of eggs were relativey consistent
across locations. Yet, abiotic factors measured in this study were relatively poor
predictors of mortality within locations despite remarkably similar estimates of mortality
between (i.e., 70%; Chapter 3). However, space-varying loss may be consistent with the
two sources of mortality including developmental malformations and exogenous removal
Predicted to cause high and variable loss after deposition (Chapter 4). I believe research
seeking to further breakdown mortality components and quantify their role in
Contributing to space-varying loss of eggs observed in natural settings (Chapter 3) will be

extrernley important for conservation and rehabilitation of this species and others.

111

APPENDIX A: TABLES AND FIGURES

112

Table 1.1. Generalized chi-square (xZ/degrees of freedom) and R-square (R2) ﬁt statistics
for models explaining daily numbers of individuals entering the river mouth and arriving
to upstream spawning areas in the Upper Black River, Michigan and those that account
for temporal autocorrelation in counts (AC), environmental (ENV) and lunar (LUNAR)
parameters or a combination of all three.

 

 

Dependent variable Model xz/df R2
River entry Null (AC) 4.69 0.38
l (ENV) 7.44 0.11
2 (AC + ENV ) 2.13 0.77
3 (LUNAR) 7.13 0.08
4 (AC + LUNAR) 4.86 0.36
5 (AC + ENV + LUNAR) 1.33 0.94
Spawning site arrival Null (AC) 2.86 0.41
6 (ENV) 7.72 0.02
7 (AC + ENV) 1.99 0.63
8 (LUNAR) 7.19 0.12
9 (AC + LUNAR) 1.93 0.66

10 (AC + ENV + LUNAR) 1.75 0.77

113

Table 1.2. Parameter estimates, parameter conﬁdence intervals (lower and upper 95%

conﬁdence intervals), and test statistics for environmental (Dlag = 24 hr lagged

discharge, Tlag = 24 hr lagged temperature, Tchange = 24 hr change in temperature and

Dchange = 24 hr change in discharge) and lunar effect (9 = lunar angle) models

describing daily observations of individuals entering the mouth of the Upper Black River,
Michigan, 2000-2002.

 

 

Model Effect [3 95% LCL 95% UCL F Prob > F
Environmental Day -0.04 -0.06 -0.01 7.63 0.009
Dlag -0.07 -0. 12 -0.02 7.94 0.009
Tlag 0.26 0.16 0.36 4.10 < 0.001
Tchange 0.21 0.04 0.38 6.30 0.015
Dchange -0.07 -0.13 -0.02 5.67 0.013
Lunar Day 0.01 -0.01 0.03 2.46 0.152
sine 0 -0.08 -0.22 0.06 1.33 0.256
cosine 0 -0.01 -0.17 0.15 0.02 0.888
sin 20 -0.03 -0. 15 0.10 0.19 0.664

 

114

Table 1.3. Parameter estimates, parameter conﬁdence intervals (lower and upper 95%
conﬁdence intervals), and test statistics for environmental (Dlag = 24 hr lagged
discharge, Tlag = 24 hr lagged temperature, Tchange = 24 hr change in temperature and
Dchange = 24 hr change in discharge) and lunar effect (9 = lunar angle) models
describing daily numbers of individuals aniving to spawning locations in the Upper
Black River, Michigan 2001-2008.

 

 

Model Effect [3 95% LCL 95% UCL F Prob > F
Environmental Day -0.30 -3.92 3.30 0.03 0.854
Sex 19.35 < 0.001

Day >< Sex . . . 23.89 < 0.001

Dlag -0.06 -0. 10 -0.03 12.31 0.001

Dchange -0.01 -0.05 0.04 0.02 0.898

Tquad 0.01 0.01 0.05 10.38 0.001

Tchange 0.29 0.17 0.40 24.54 < 0.001

Lunar Day -0.36 -3.95 3.25 0.02 0.8362
Sex 19.35 < 0.001

Day X Sex . . 23.89 < 0.001

sin 0 -0. 15 -0.36 0.06 1.85 0.174

cosine 0 0.59 0.40 0.78 38.91 < 0.001

sin 20 -0.80 -l.02 -0.55 40.94 < 0.001

 

115

Table 1.4. Parameter estimates, parameter conﬁdence intervals (lower and upper 95%
conﬁdence intervals), for combined environmental (Dlag = 24 hr lagged discharge, Tlag
= 24 hr lagged temperature, Tchange = 24 hr change in temperature and Dchange = 24 hr
change in discharge) and lunar effect (9 = lunar angle) models describing daily numbers

of individuals entering the mouth of the Upper Black River, Michigan and arriving to
spawning locations, 2001-2008.

 

 

Variable Effect [3 95% LCL 95% UCL

River entry Day -0.11 -2.77 2.56
Dlag -0.08 -0. 12 -0.01
Tlag 0.18 0.04 0.33
Tchange -0.05 -0.28 0.17
Dchange -0.09 -0. 1 3 -0.02
sin 9 -0.21 -9.05 8.63

Spawning site arrival Day -0.42 -4.74 3.75
Sex
Day >< Sex . . .
Dlag -0. 12 -0. l 7 -0.08
Tchange 0.50 0.36 0.62
Tquad 0.01 0.00 0.02
cosine 0 1.08 0.43 1.73
sine 20 -1.01 -l.28 -0.75
Dlag X cosine 0 -0.15 -0.21 -0.10
Tquad >< cosine 0 0.01 0.00 0.01
Tchange >< cosine 0 -0.60 -0.83 -0.37

 

116

:—.—2ooo
_—o—2001
.—-—2002

 

4/21 4/28 5/5 5/12 5/19 5/26 6/2 6/9

24- +2003
22- +2004
20- +2005

Water temperature 0C

 

 

4/21 4/28 5/5 5/12 5/19 5/26 6/2 6/9

26 1

24 + 2006
22 + 2007
201 —-— 2008

 

4ﬁi.

4/21 4/2 5/5 5/12 5/19 5/26 6/2 6/9
Date

Figure 1.1. Measurements of daily water temperature (°C) of the Upper Black River,
Michigan collected from April 22 through June 5 for 2000-2002 (upper panel), for 2003-
2005 (middle panel) and 2006-2008 (lower panel).

117

45-
40~
35-
30-
25-
20‘
15'
10-

 

—o— 2000
+ 2001
—-— 2002

    

 

5
4/21
45

40

mo)
cur

N
O

Water discharge m3/ S
3‘. 8

g—a
O

 

5

4/28 5/5 5/12 5/19 5/26 6/2 6/9

—-0— 2003
—0— 2004
—-— 2005

 

 

 

45-
40~
354
30-
25-
20-
15-
10«

 

5 ..

4/21 4/28 5/5 5/12 5/19 5/26 6/2 6/9

+ 2006
—o— 2007
—-— 2008

    

I
"“ * ~h.-u.‘. “3 '~ ﬁ‘

4/21 4/28 5/5 5/12 5/19 5/26 6/2 6/9

Date

Figure 1.2. Measurements of daily river discharge (313/ S) 0f the Upper Black River,
Michigan collected from April 22 through June 5 for 2000-2002 (upper panel), for 2003-
2005 (middle panel) and 2006-2008 (lower panel).

118

55 ‘ —— 2000
50 .. .......... 2001
--—- 2002

Number of lake sturgeon observed

 

 

 

Time (24 hours)

Figure 1.3. Number of adult lake sturgeon observed entering the Upper Black River,
Michigan 2000-2002 as function of the hour of the day.

119

5° ' 2000 — 95% C1
45 ' 0 Observed
Predicted

 

301

  

 

 

 

0 3 6 912151821242730

35‘

 

Number of lake sturgeon observed
N
O

 

 

 

0 3 6 9121518212427303336
Dayofseason

Figure 1.4. Observed, predicted and 95% conﬁdence intervals for the daily number of
lake sturgeon entering the Upper Black River, MI ﬁom 2000-2002.

120

SOT
45'

35-
30-
25-
20-
15-
10-

5.

 

2“... a

 

4
50-

45‘
40.
35i
30-
25‘
20‘
154
10-

Number of lake sturgeon observed

 

10 12 14 16 18 20 22 24 26
Lagged river discharge (m3/s)

 

8 10 12 14 16 18 20 22
Lagged water temperature (° C)

Figure 1.5. Relationship between the daily number of lake sturgeon observed entering the
mouth of the Upper Black River, Michigan with 1 day lagged river discharge and 1 day
lagged water temperature.

121

Number of lake sturgeon observed or captured

40q
354
30.

25-

N
O
1

ﬂ
VJ!
l

H
O
I

M
l

 

0
4/22

Ur
O

401

D)
O
4

N
O
r

10r

 

0

4/22

2001

 

4/26 4/30 5/4

‘ 2002

4/26 4/30 5/4

---- Migratory activity
—— Spawning activity

5/8 5/12 5/16 5/20 5/24 5/28

 

 

 

J, ‘1

5/8 5/12 5/16 5/20 5/24 5/28
Date

ﬂ

Figure 1.6. The number of lake sturgeon observed entering the Upper Black River or
captured at upstream spawning location as a function of the date in 2001 and 2002.

122

 

99999?
ONAmmO

Proportion of lunar illumination

 

15 18 21 0 3 6 9 12 15 18 21 24

-w
MM
.m-

’40-

  
 
   
 

Number of sturgeon captured
O
w
0
\O
5

 

 

  

12 16 20 24 28 32 36

 

 

 

- ‘ 0-
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40 45
Day of season Day of season

Figure 1 7. Observed, predicted and 95% conﬁdence intervals for the daily number of
lake sturgeon captured at spawning locations the Upper Black River, MI ﬁ'om 2001-2008
and descriptions of the proportion of lunar illumination for each day.

123

o8
>

Number of lake sturgeon
8 U.)

H

O
\‘

r
v * "

é...

\

.0

Number of lake sturgeon
O

 

Figure 1.8. Relationship between daily numbers of lake sturgeon captured at spawning
locations on the Upper Black River, Michigan and lunar angle with A) 24 hour lagged
l’lver discharge, B) 24 hour lagged water temperature and C) the magnitude and change in
water temperature over the previous 48 hours.

124

Table 2.1. Estimates of repeatability by sex for three size classes of lake sturgeon in
relation to standardized spawning day, river zone and environmental variables of the
Upper Black River, MI. Variables are rank ordered by repeatability. Asterisks indicate
variables where statistical evidence supported a model with heterogeneous variances and
thus size/age-based differences in repeatability.

 

 

 

 

Repeatability
Female sturgeon Male sturgeon
142-164 166-174 175-189 112-140 141-152 153-179
Variables cm cm cm cm cm cm
Standardized day 0.56 0.56 0.56 0.42 0.42 0.42
River zone 0.16 0.16 0.16 0.04 0.04 0.04
Lunar illumination 0.15 0.15 0.15 0.00 0.00 0.00
Lagged water temp 0.13 0.13 0.13 0.02 0.02 0.02
Average temp 0.12 0.12 0.12 0.14 0.14 0.14
Average discharge“ 0.01 0.05 0.01 0.02 0.01 0.01
Temp change 0.00 0.00 0.00 0.00 0.00 0.00
Lagged discharge“ 0.00 0.08 0.00 0.00 0.00 0.00
Discharge change“ 0.00 0.00 0.00 0.04 0.00 0.00

 

125

  

Upper Black River

 

 

-

 

 

 

 

 

‘e

 

\‘Zone 2

 

i Zone 1

A

  

L
Q

 

1

 

0km
:-

 

lake sturgeon.

Frgure 2 1 Map of the Upper Black River, MI and spawning areas (zones) used by adult

126

— Males
=23 Females

 

 

2003

 

Number of adult lake sturgeon
o u: E a 8 a
o '5 '5’ i5 3
. . . :
8
A

 

 

  

 

60 2005 50 i 2006
50
4O
30
20 .
10
0
40 1 50
2007 12008
30_ 40
30
20-
20
10- 10
‘1
0 4 r ' l . . . 0
0 51015202530354045 0 51015202530354045
Day of season

Figure 2.2. Number of male and female lake sturgeon captured during each day of the
season in each of eight consecutive years (2001-2008).

127

225

20' B Female
I Male

Number of adult lake sturgeon
U-ﬁ F. H hi N
d 8 8 ‘o” d 8

N
U!
r

 

 

 

 

 

1 2 3 4 5 6 7 8
Times captured

Figure 2.3. Distribution of total number of capture and recapture events (N=1260) for
individual male and female lake sturgeon (N=678) collected over eight consecutive years
(2001-2008) in the Upper Black River, MI.

128

Female lake sturgeon

0.15“
0.01
0.26
0.46
0.10
0
q ©©®©®
0.49
0.24
0.03
0.03*
0.18*
Male lake sturgeon

Figure 2.4. Frequency distribution of observed inter-annual spawning intervals in years
for female (top) and male (bottom) lake sturgeon spawning in the Upper Black River,
Michigan. The data are presented as the probability of returning to spawn in year t + x
given that spawning occurs in year t. For example, there is an 18% chance that a male
spawning in year t will spawn the next year. Asterisk indicates upwardly biased
probability estimates due to individuals likely missed between captures.

129

1.0 '
0.9 ‘
0.8 ‘
0.7 ‘
0.6 "
0.5 '-
0.4 '
0.3 ‘
0.2 ‘
0.1 ‘

0.0

Predicted stanardized spawning day

0.3
0.2
0.1

0.0 -

 

 

0 5 1015 20 25 3035 4045 5055 60 65 70
1.0-

0.9 -
0.8 ‘
0.7 i
0.6 ‘
0.5 -
0.4 ‘

- '11} 1111111111111

75
Female lake sturgeon
’1 l

1111111 ”11

11111 . 111
1 1111111111 “111 "11111111 1111111

”11111111111111!”I'M 1

 

iiiii

 

100 125 150 175 200 225
Male lake sturgeon

0 25 50 75

Figure 2.5. Predicted standardized spawning day (i 1 SE) for recaptured male (N = 232)
and female (N = 70) lake sturgeon captured on 3 2 occasions during 8 consecutive years

(2001-2008). Individuals are presented in rank order by predicted standardized spawning

day.

130

 

70‘ ----Zone2
-'-Zone3
60 - Zone4
"" ....... ZoneS
0 or
g 50 "- Zone6
N
8
o 40 r
.2
‘66
T) 30 -
H
U)
“U
“o
O

 

 

 

0 0.25 0.5 0.75 1
Standardized spawning day

Figure 2.6. Odds of lake sturgeon spawning one of 5 different river zones in the Upper
Black River, MI relative to the most upstream zone (Zone 1) in relation to standardized
spawning day.

131

     

to... 0.00."... 000 o. O

 

200 -

 

11

A83 596— gob

 

100

0.4 0.5
Standardized sp awning day

0.7 0.8 0.9

0.6

0.2 0.3

0.1

Figure 2.7. Distribution of male and female lake sturgeon body size as a function of

standardized spawning day for the Upper Black River, MI population. Data represents
captures of all individuals collected from 2001-2008, including recaptures.

132

Table 3.1. Summary of egg deposition non-spatial and spatial models’ parameter
estimates for Site 1 (o = variance of the random spatial effect; (p = spatial range

orientation, Ag and 1 = effective range parameter). Parameter posterior credible intervals
i.e., 50 (2.5, 97.5) percentiles are provided. Those B parameters for environmental
covariates [i.e., Depth = water depth (m); Velocity = water velocity (m3/s); Angub =
average substrate size (mm); MaxSub = maximum substrate size (mm)] that are
signiﬁcant at the 0.05 level are bolded. Model ﬁt in the form of Deviation Information
Critera (DIC) and model’s effective number of parameters (pD) are also provided for
comparison.

 

 

 

Site 1

Parameters Non-Spatial Spatial
Intercept 1.86 (1.71, 1.99) -l.94 (-4.61, 0.43)
B Depth -0.44 (-0.61, -0.28) 1.89 (0.11, 3.84)
B Velocity -0.04 (-0.l9, 0.15) 1.18 (-0.21, 2.60)
B Angub 0.014 (0.012, 0.015) 0.004 (-0.005, 0.013)
B MaxSub ' 0.003 (0.002, 0.004) 0.007 (0.002, 0.012)
02 - 5.19 (2.96, 9.56)
q) radians - 0.18 (-0.63, 0.68)
(p degree - 10.78 (-36.31, 39.46)
10 - 10.96 (5.00, 16.10)
711 - 5.66 (2.94, 11.65)
M) eff. range - 32.85 (14.98, 48.25)
H eff. range - 16.97 (8.81, 34.91)
DIC -8710 -14115

JD 5.0 117.4

 

133

Table 3.2. Summary of egg deposition non-spatial and spatial models’ parameter
estimates for Site 2 (o = variance of the random spatial effect; (9 = spatial range

orientation, Ag and 1 = effective range parameter). Parameter posterior credible intervals
i.e., 50 (2.5, 97.5) percentiles are provided. Those B parameters for environmental
covariates [i.e., Depth = water depth (m); Velocity = water velocity (m3/s); Angub =
average substrate size (mm); MaxSub = maximum substrate size (m)] that are
signiﬁcant at the 0.05 level are bolded. Model ﬁt in the form of Deviation Information
Critera (DIC) and model’s effective number of parameters (pD) are also provided for
comparison.

 

 

 

Sﬁe2

Parameters Non-Spatial Spatial
Intercept 1.42 (0.90, 1.93) -0.27 (-2.54, 2.04)
B Depth 0.96 (0.33, 1.39) 0.87 (-2.18, 3.44)
B Velocity -0.79 (—l.29, 41.35) -0.20 (-2.84, 2.25)
B Angub 41.012 (41.014, -0.009) -0.015 (-0.034, 0.001)
B MaxSub 0.015 (0.012, 0.018) 0.022 (0.008, 0.039)
02 - 2.42 (1.36, 4.93)
(p radians - -O.25 (-0.75, 0.75)
(p degree - -14.49 (-4319, 43.08)
710 - 2.23 (0.37, 9.51)
1.1 - 1.83 (0.39, 9.29)
3.0 eff. range - 6.68 (1.13, 28.50)
711 eff. range - 5.48 (1.17, 27.85)
DIC -1 548 -2139

in 5.7 48.5

 

134

Table 3.3. Summary of egg deposition non-spatial and spatial models’ parameter
estimates for Site 3 (o = variance of the random spatial effect; (p = spatial range

orientation, Ag and 1 = effective range parameter). Parameter posterior credible intervals
i.e., 50 (2.5, 97.5) percentiles are provided. Those B parameters for environmental
covariates [i.e., Depth = water depth (m); Velocity = water velocity (m3/s); Angub =
average substrate size (mm); MaxSub = maximum substrate size (m)] that are
signiﬁcant at the 0.05 level are bolded. Model ﬁt in the form of Deviation Information
Critera (DIC) and model’s effective number of parameters (pD) are also provided for

 

 

 

comparison.

Parameters Non-Spatial Spatial
Intercept 0.02 (-0.12, 0.25) -3.49 (-5.79, -0.41)
B Depth -0.34 (41.58, 41.18) 0.68 (-l.36, 2.82)
B Velocity 1.71 (1.57, 1.87) 0.22 (-1.02, 1.50)
B Angub 0.016 (0.013, 0.019) 0.019 (0.005, 0.035)
B MaxSub -0.001 (-0.003, 0.001) -0.009 (-0.020, 0.002)
0’2 - 10.43 (6.33, 17.35)
(p radians - 0.10 (-0.13, 0.34)
(p degree - 5.88 (-7.44, 19.98)
7.0 - 18.12 (10.84, 22.05)
711 - 6.30 (3.71, 10.40)
7.0 eff. range - 54.29 (32.48, 66.07)
711 eff. range - 18.89 (ll.l3,3l.18)
DIC -6163 -11080

pD 5.0 117.1

 

135

Table 3.4. Summary of egg deposition non-spatial and spatial models’ parameter
estimates for Site 4 (o = variance of the random spatial effect; (0 = spatial range

orientation, 210 and 1 = effective range parameter). Parameter posterior credible intervals
i.e., 50 (2.5, 97.5) percentiles are provided. Those B parameters for environmental

covariates [i.e., Depth = water depth (m); Velocity = water velocity (m3/s); Angub =
average substrate size (mm); MaxSub = maximum substrate size (m)] that are
signiﬁcant at the 0.05 level are bolded. Model ﬁt in the form of Deviation Information
Critera (DIC) and model’s effective number of parameters (pD) are also provided for

 

 

 

comparison.

Site 4
Parameters Non-Spatial Spatial
Intercept 0.87 (0.72, 1.01) -4.20 (-7.02, -0.81)
B Depth 0.81 (0.62, 0.93) 7.26 (2.05, 9.16)
B Velocity 1.78 (1.58, 1.96) 2.64 (0.42, 4.42)
B Angub 0.009 (0.008, 0.010) 0.002 (-0.004, 0.009)
B MaxSub 0.003 (0.003, 0.005) -0.002 (-0.007, 0.003)
02 - 5.84 (3.54, 10.51)
(p radians - 0.29 (-0.l4, 1.12)
q, degree - 16.94 (-8.51, 64.34)
7.0 - 10.33 (5.87, 12.14)
711 - 5.045 (2.83, 10.31)
7.0 eff. range - 30.97 (17.60, 36.37)
7.1 eff. range - 15.11 (8.47, 30.89)
DIC -22051 -28620
pD 5.1 95.4

 

136

Table 3.5. Summary of egg count change non-spatial and spatial models’ parameter
estimates for Site 3 (t = measurement error; a = variance of the random spatial effect; (p

= spatial range orientation, 710 and 1 = effective range parameters). Parameter posterior
credible intervals i.e., 50 (2.5, 97.5) percentiles are provided. Those B parameters for
environmental covariates [i.e., Depth = water depth (m); Velocity = water velocity

(ma/s); Angub = average substrate size (mm); MaxSub = maximum substrate size
(m)] that are signiﬁcant at the 0.05 level are bolded Model ﬁt in the form of Deviation
Information Critera (DIC) and model’s effective number of parameters (pD) are also

 

 

 

provided for comparison.

Site 3
Parameters Non- Spatial Spatial
Intercept 0.27 (-0.62, 1.16) 0.21(-1.05, 1.49)
13 Depth 0.15 (-0.98, 1.29) -0.66 (-244, 1.14)
8 Velocity -0.89 (-2.26, 0.47) -121 (-2.58, 0.14)
13 Angub -0.007 (-0014, 0.001) -0004 (-001, 0.002)
B MaxSub -0.003 (-0.008, 0.002) 0.001 (-0.004, 0.006)
r2 1.94 (1.55, 2.49) 0.81 (0.31, 1.30)
6 - 1.32 (0.62, 2.64)
(p radians - 0.20 (-025, 1.21)
(p degree - 11.71 (-l4.64, 69.76)
10 - 6.81 (2.11, 11.76)
711 - 4.23 (1.54, 10.56)
1.0 eff. range - 20.40 (6.32, 35.24)
711 eff. range - 12.69 (4.62, 31.64)
DIC 246.3 160.5
pD 5.9 54.5

 

137

Table 3.6. Summary of egg count change non-spatial and spatial models’ parameter
estimates for Site 4 (r = measurement error; a = variance of the random spatial effect; (p

= spatial range orientation, Ag and 1 = effective range parameters). Parameter posterior
credible intervals i.e., 50 (2.5, 97.5) percentiles are provided. Those B parameters for
environmental covariates [i.e., Depth = water depth (m); Velocity = water velocity

(m3/s); Angub = average substrate size (mm); MaxSub = maximum substrate size
(m)] that are signiﬁcant at the 0.05 level are bolded Model ﬁt in the form of Deviation
Information Critera (DIC) and model’s effective number of parameters (pD) are also
provided for comparison.

 

 

 

Site 4
Parameters Non-Spatial Spatial
Intercept 0.27 (-017, 0.72) 0.10 (-0.63, 0.84)
8 Depth 0.04 (-0.43, 0.52) -0.16 (-0.82, 0.48)
B Velocity -0.69 (-1.13, -o.24) -0.06 (-0.66, 0.55)
p Angub 41.01 (-0.02, -o.002) 41.01 (-0.019, ~o.002)
B MaxSub 0.004 (-0002, 0.01) 0.004 (-0.001, 0.010)
r2 1.05 (0.89, 1.27) 0.73 (0.48, 0.97)
02 - 0.47 (0.22, 1.06)
(p radians - -0.01 (-0.64, 0.37)
(p degree - -1.02 (-36.73, 21.38)
710 - 14.86 (4.96, 21.70)
1.1 - 4.13 (0.89, 14.30)
710 eff. range — 44.51 (14.87, 65.01)
7d eff. range - 12.39 (2.69, 42.86)
DIC 269.6 228.3
JD 5.9 59.6

 

138

Table 3.7. Summary of cross-validation multivariate spatial models’ parameter estimates
for Sites 3 and 4 (02 = variance of space-varying matrix elements 1, 2 and 3; (p = spatial

range parameter for matrix elements (a) and the space-varying correlation (v); r =
residual measurement error for time 1 and 2 covariance matrix. Parameter posterior
credible intervals i.e., 50 (2.5, 97.5) percentiles are provided. Those B parameters for
environmental covariates [i.e., Depth = water depth (m); Velocity = water velocity

(m3/s); Angub = average substrate size (mm); MaxSub = maximum substrate size
(mm)] collected during each sampling period (t =1, t = 2) that are signiﬁcant at the 0.05

 

 

level are bolded.

Parameters Site 3 Site 4
Intercept 0.36 (-1.33, 2.002) 1.15 (0.02, 2.11)
B Depth, t = 1 1.84 (0.02, 4.00) -0.33 (-1.35, 0.77)

B Velocity, t= l
B Angub, t = 1
B MaxSub, t = 1
Intercept, t = 2
B Depth, t = 2

[3 Velocity, t = 2
B Angub, t = 2

1.41 (-003, 3.07)
0.003 (-0003, 0.008)
0.001 (-0.004, 0.006)

-1.24 (-2.60, 0.06)

2.69 (0.93, 4.57)

0.71 (-0.82, 2.07)
-0.001 (-0.008, 0.005)

0.26 (-0.54, 0.97)
0.01 1 (0.002, 0.027)
.0005 (-0015, 0.003)

0.05 (-0.81, 1.44)

-027 (-1.47, 0.64)

0.44 (-033, 1.27)
0.001 (-0009, 0.017)

0 MaxSub, t = 2 0.003 (0002, 0.009) -0.003 (-0.014, 0.004)
62,1 0.26 (0.12, 0.65) 0.24 (0.12, 0.69)
02,2 0.43 (0.18, 1.20) 0.50 (0.20, 1.46)
02,3 0.34 (0.13, 1.23) 0.54 (o. 17, 3.51)
(p, eff. Range 24.83 (6.35, 33.57) 15.78 (1.19, 61.11)
6,, eff. Range 24.79 (10.51, 33.51) 49.40 (21.16, 63.31)
r2 t=1 0.60 (0.38, 0.93) 0.51 (0.34, 0.75)

r2 t=2 0.50 (0.28, 0.81) 0.22 (0.10, 0.612)

 

139

Upper Black River

s

 

 

 

 

 

  

e.

 

 

 

 

 

 

IWNM

 

 

spawning activity (Right).

Figure 3.1. Map of the Upper Black River and the location of four Sites used for

140

  
    
  

0.45

 

   

Eisner
0'40 lSitc2
(a 0-35 DSite3
9.2 0.30 ISite4
5 025
a 0.20
0.15
o
a 0.10
0.05
0.00 . "
<30 30- 60- 90- 120- 150- >180 <40 40- 80— 120- 160- 200- >240
59.9 89.9 ll9.9l49._9179.9 79.9 ll9.9159.9199.9239.9
0 45 Average substrate srze (mm) 050 Max substrate size (mm)
0.45
’3‘ 0.40
g 0.35
.5 0.30
1: 0.25
8. 0.20
E 0.15
0.10
0.05
0.00
<0.2 0.20-0.40-0.60-0.80-1.0- >12 < 0-25-0-50-0-75- 1-0- 125->15
039 059 ()_79 099 1_19 0.25 0.49 0.74 0.99 1.24 1.49
Water depth (111) Water velocity (m3/s)

Figure 3.2. Frequency histograms for average substrate size, maximum substrate size,
water depth and water velocity for Sites 1-4.

141

Stream width (m)

 

 

0 10 20 30 40 50 60
Streamlength(m)

Figm'e 3.3. Interpolated surfaces of water depth (m), velocity (m3/s), average substrate
and maximum substrate (mm) variables and egg count (top to bottom) for Site 3.

142

 

 

Stream width (m)

 

 

 

l I T l l l l l l l I I l

0 1 0 20 3O 40 50 60
Stream length (m)

Figure 3.4. Interpolated surfaces of the non-spatial model residuals, non-spatial ﬁtted

values, spatial random effects and ﬁtted values of the spatial model (top to bottom) for
Site 3.

143

26

1 11

26

11

26

Stream width (m)
l l 26 1

1

11

 

l

O 10 20 30 40 50 60
Stream 1ength(m)

Figure 3.5. Interpolated surfaces of live eggs (time 1 and 2), random spatial effects (w(s)
time 1 and 2), and residual spatial correlation (p Fl, t=2 (s)) from model (3) for Site 3
(top to bottom). Points represent locations where there was at least one live egg in time 1.

 

0 10 20 30 40 50 60

    

 

. _ ’Ii" in,

0 10 20 30 40 50 60

Stream width (m)
l

 

 

F‘_‘—‘1-“—I—“—-I‘—ﬁ—-"_'T—'—l
0 10 20 30 4O 50 60

 

 

r—r—'—I—1—‘T—_—__T—'\
0 10 20 30 40 50 60
Stream length (m)

Figure 3.6. Interpolated surfaces of live eggs (time 1 and 2), random spatial effects (w(s)
time 1 and 2), and residual spatial correlation (p t=1, t=2 (5)) from model (3) for Site 4
(top to bottom). Points represent locations where there was at least one lrve egg in trme 1.

145

 

 

11'
- :-

 

 

T Exogenous Removal (Predation and Currents)

11
it

1 Developmental Failure (Flow)

Figure 4.1. Experimental design used to estimate egg mortality and partition components
of loss (Top). Four treatments of in situ exclosures and hypotheses regarding the source,
direction and magnitude of exogenous removal (i.e., predation and water currents) and
the magnitude of mortality due to developmental mortality (Bottom).

146

ggs remaining

Percent of e

Percent of total mortality

Percent of total mortality

1.0 -

.0
)0

0.8 -

.o .0
ON \I
l I

0.5 -

(3.0.09
i—Nw-A
..

_ Small Mesh
‘ ~~~ Fine Mesh

   

A) Survival

 

— Open
----- Large Mesh

 

~o
'o'o

0.9 -
0.8 -
0.7 r
0.6 -
0.5 r
0.4 -
0.3 -
0.2 -
0.1 ‘
0.0

0.9 -

.9 _o .o .o .0 .°
W & LII O‘ \l m
r r r r 1 r

 

P
N
r

.9
_r_

0.0 I

0 24 48 72

B) Developmental failure

0

 

C) Exogenous removal

 

'72 96
Time since fertilization (hrs)

0 24 48 120

Figure 4.2. Least square mean differences in the A) percent of eggs remaining, B)
percent of eggs dead due to developmental failure and C) percent of eggs dead due
sources leading to exogenous removal among Open, Large mesh, Small mesh and Fine
mesh exclosure treatments during 120 hours of incubation in natural stream conditions.
Treatment means with overlapping conﬁdence intervals generally indicate signiﬁcant
pair-wise differences (a=0.05).

147

0.70 -
0.65 -
0.60 ~
0.55 -
0.50 -
0.45 r
0.40 -
0.35 -
0.30 -
0.25 -
0.20 -
0.15 -
0.10 -
0.05 i
0.00

 

D Op en
I Gravel
I large Rock

Percent remaining (avg 4: C1)

 

 

   

 

 

 

 

5 l 0 15 30
Predator density

Figure 4.3. Least square mean proportion of eggs surviving across three substrate sizes
(Open, Gravel and Large Rock and four different simulated densities of crayﬁsh

predators.

148

0.80 -
0.70 -
0.60 -
0.50 - . .
0.40 -
0.30 -
0.20

0.10
O .

0.00 r r '4. . . ﬁ
18 21 24 27 30 3 3 36
Average crayﬁsh length (mm)

Percent of eggs remaining
0
O
O
O

 

 

 

Figure 4.4. Proportion of eggs consumed as a ﬁmction of mean predator size in each of
four density treatments.

149

LITERATURE CITED

150

LITERATURE CITED

Ahola, M., T. Laaksonen, K. Sippola, T. Eeva, K. Rainio, and E. Lehikoinen. 2004.
Variation in climate warming along the migration route uncouples arrival and
breeding dates. Global Change Biology 10: 1610-1617.

Alerstam, T. and P. H. Enckell. 1979. Unpredictable habitats and evolution of bird
migration. Oikos 33:228-232.

Alerstam, T., A. Hedenstrom, and S. Akesson. 2003. Long-distance migration: evolution
and determinants. Oikos 103:247-260.

Alexander, R. M. 1998. When is migration worthwhile for animals that walk, swim or
ﬂy? Journal of Avian Biology 29:387-394.

Andrewartha, H. G., and L. C. Birch. 1954. The distribution and abundance of animals.
University of Chicago Press, Chicago, Illinois, USA.

Angilletta, M. J., A. F. Bennett, et al. 2006. Coadaptation: A unifying principle in
evolutionary thermal biology. Physiological and Biochemical Zoology 792: 282-
294.

Atkinson, D. 1994. Temperature and organism size — a biological law for ectotherrrrs.
Advances in Ecological Research, V0125. London, Academic Press Ltd. 25: 1-5 8.

Auer, N. A 1996. Importance of habitat and migration to sturgeons with emphasis on

lake sturgeon. Canadian Journal of Fisheries and Aquatic Sciences. 53
Supplement 1: 152-160.

Auer, N. A. 1996. Response of spawning lake sturgeon to change in hydroelectric facility
operation. Transactions of the American Fisheries Society 125: 66-77.

Auer, N. A. and E. A. Baker. 2002. Duration and drift of larval lake sturgeon in the
Sturgeon River, Michigan. Journal of Applied Ichthyology 18: 557-564.

Bailey, K. M. and E. D. Houde. 1989. Predation on eggs and larvae of marine ﬁshes and
the recruitment problem. Advances in Marine Biology 25: 1-82.

Baker, E. A. and D. J. Borgeson. 1999. Lake sturgeon abundance and harvest in Black

Lake, Michigan, 1975-1999. North American Journal of Fisheries Management
19:1080-1088.

Bakker, T. C. M. 1993. Positive genetic correlation between female preference and
preferred male ornament in sticklebacks. Nature 3636426: 255-257.

Ball, G. F. and E. D. Ketterson. 2008. Sex differences in the response to environmental

151

cues regulating seasonal reproduction in birds. Philosophical Transactions of the
Royal Society B-Biological Sciences 3631490: 231-246.

Baril, M. and P. Magnan. 2002. Seasonal timing and diel activity of lacustrine brook
charr, Salvelinusfontinalis, spawning in a lake outlet. Environmental Biology of
Fishes 641-3: 175-181.

Barlaup, B. T., S. E. Gabrielsen, H. Skoglund, and T. Wiers. 2008. Addition of spawning
gravel - A means to restore spawning habitat of Atlantic salmon (Salmo salar L.),
and anadromous and resident brown trout (Salmo trutta L.) in regulated rivers.
River Research and Applications 24:543-550.

Bateman, A. J. 1948. Intra-sexual selection in drosophila. Heredity 23: 349-368.

Beall, E. and B. de Gaudemar. 1999. Plasticity of reproductive behaviour in Atlantic

salmon Salmo salar Salmonidae in relation to environmental factors. Cybium
231: 9—28.

Benson, R. L., S. Turo, and B. W. McCovey. 2007. Migration and movement patterns of
green sturgeon (Acipenser medirostris) in the Klamath and Trinity rivers,
California, USA. Environmental Biology of Fishes 79:269-279.

Bemier-Bourgault, I. and P. Magnan 2002. Factors affecting redd site selection,
hatching, and emergence of brook charr, Salvelinusfontinalis, in an artiﬁcially
enhanced site. Environmental Biology of Fishes 64:333-341.

Biga, H., J. Janssen, and J. E. Marsden. 1998. Effect of substrate size on Lake Trout egg
predation by Mottled Sculpin. Journal of Great Lakes Research 24:464-473.

Binder, T. R and D. G. McDonald. 2008. The role of temperature in controlling diel
activity in upstream migrant sea lampreys Petromyzon marinus. Canadian Journal
of Fisheries and Aquatic Sciences 656: 1113-1121.

Bizzotto, P. M., A L. Godinho, et a1. 2009. Inﬂuence of seasonal, diel, lunar, and other
environmental factors on upstream ﬁsh passage in the Igarapava Fish Ladder,
Brazil. Ecology of Freshwater Fish 183: 461-472.

Blanchﬁeld, P. J. and M. S. Ridgway. 2005. The relative inﬂuence of breeding
competition and habitat quality on female reproductive success in lacustrine brook

trout (Salvelinus fontinalis). Canadian Journal of Fisheries and Aquatic Sciences
62:2694-2705.

Boake, C. R. B. 1989. Repeatability — Its role in evolutionary studies of mating behavior.
Evolutionary Ecology 32: 173-182.

Bond, N. R. and B. J. Downes. 2003. The independent and interactive effects of ﬁne

152

sediment and ﬂow on benthic invertebrate communities characteristic of small
upland streams. Freshwater Biology 48:455-465.

Boulton, A J., S. Findlay, P. Marmonier, E. H. Stanley, and H. M. Valett. 1998. The
functional signiﬁcance of the hyporheic zone in streams and rivers. Annual
Review of Ecology and Systematics 29:59-81.

Boulton, A. J., T. Datry, T. Kasahara, M. Mutz, and J. A. Stanford. 2010. Ecology and
management of the hyporheic zone: stream-groundwater interactions of running
waters and their ﬂoodplains. Journal of the North American Benthological
Society 29:26-40.

Bouwes, N. and C. Luecke. 1997. The fate of Bonneville Cisco eggs in Bear Lake:
Evaluating mechanisms of egg loss. Transactions of the American Fisheries
Society 126:240-247.

Brooks, R 1996. Copying and the repeatability of mate choice. Behavioral Ecology and
Sociobiology 395: 323-329.

Brooks, 8., C. R. Tyler, and J. P. Sumpter. 1997. Quality in ﬁsh: what makes a good egg?
Reviews in Fish Biology and Fisheries 7:387-416.

Brown, C. and K. N. Laland 2003. Social learning in ﬁshes: a review. Fish and Fisheries
43: 280-288.

Brown, G. P. and R Shine. 2007. Like mother, like daughter: inheritance of nest-site
location in snakes. Biology Letters 32: 131-133.

Bruch, R M. and F. P. Binkowski. 2002. Spawning behavior of lake sturgeon Acipenser
ﬁtlvescens. Journal of Applied Ichthyology 18: 570-579.

Bruch, R M., G. Miller, and M. J. Hansen. 2006. F ecundity of lake sturgeon (Acipenser

fulvescens, Raﬁnesque) in Lake Winnebago, Wisconsin, USA. Journal of Applied
Ichthyology 22:116-118.

Bruch, R. M., S. E. Campana, et al. 2009. Lake sturgeon age validation using bomb

radiocarbon and known-age ﬁsh. Transactions of the American Fisheries Society
1382: 361-372.

Bruch, R M., T. Dick, et al. 2001. A ﬁeld guide for the identiﬁcation of stages of gonad
development in lake sturgeon, Acipenserﬁtlvescens Raﬁnesque, with notes on
lake sturgeon reproductive biology and management implications. Publ. WI Dept.
Nat. Res. Oshkosh and Sturgeon for Tomorrow. Appleton, WI, Graphic
Communications Center, Inc.

Bunn, N. A., C. J. Fox, and T. Webb. 2000. A literature review of studies on ﬁsh egg

153

mortality: Implications for the estimation of spawning stock biomass by the
annual egg production method. Centre for Environment, Fisheries and
Aquaculture Science 111.

Anderson, D. R. and K. P. Burnham. 2002. Avoiding pitfalls when using information-
theoretic methods. Journal of Wildlife Management 66:912-918.

Cardinale, B. J., M. A. Palmer, C. M. Swan, S. Brooks, and N. L. Poff. 2002. The
inﬂuence of substrate heterogeneity on bioﬁhn metabolism in a stream ecosystem.
Ecology 83:412-422.

Carofﬁno, D. C., T. M. Sutton, R. F. Elliott, and M. C. Donoﬁio. 2010. Early life stage
mortality rates of lake sturgeon in the Peshtigo River, Wisconsin. North American
Journal of Fisheries Management 30:295-304.

Caron, F ., D. Hatin, et al. 2002. Biological characteristics of adult Atlantic sturgeon
Acipenser oxyrinchus in the St. Lawrence River estuary and the effectiveness of
management rules. Journal of Applied Ichthyology 18: 580-585.

Carpenter, S. R, S. G. Fisher, et al. 1992. Global change and ﬂesh-water ecosystems.
Annual Review of Ecology and Systematics 23: 119-139.

Carr, J. P. and L. Kaufman. 2009. Estimating the importance of maternal age, size, and

spawning experience to recruitment of Atlantic cod Gadus morhua. Biological
Conservation 1423: 477-487.

Carlsson, J ., K Aarestrup, F. Nordwall, I. Naslund, T. Eriksson, and J. E. L. Carlsson.
2004. Migration of landlocked brown trout in two Scandinavian streams as
revealed from trap data. Ecology of Freshwater Fish 13:161-167.

Carmichael, J. T., S. L. Haeseker, and J. E. Hightower. 1998. Spawning migration of
telemetered striped bass in the Roanoke River, North Carolina. Transactions of
The American Fisheries Society 127:286-297.

Caselle, J. E. 1999. Early post-settlement mortality in a coral reef ﬁsh and its effect on
local population size. Ecological Monographs 69: 177-194.

Caswell, N. M., D. L. Peterson, B. A. Manny, and G. W. Kennedy. 2004. Spawning by
lake sturgeon (Acipenserﬁrlvescens) in the Detroit River. Journal of Applied
Ichthyology 20: 1-6.

Chapman, F. A and S. H. Carr. 1995. Implications of early-life stages in the natural-
history of the Gulf of Mexico sturgeon, Acipenser Oxyrinchus De Soto].
Environmental Biology of Fishes 432407-413.

Chiotti, J. A., J. M. Holtgren, N. A. Auer, and S. A. Ogren. 2008. Lake sturgeon

154

spawning habitat in the Big Manistee River, Michigan. North American Journal
of Fisheries Management 28: 1009-1019.

Chotkowksi, M. A. and J. E. Marsden. 1999. Round Goby and Mottled Sculpin predation
on Lake Trout eggs and ﬂy: Field predictions from laboratory experiments.
Journal of Great Lakes Research 251: 26-35.

Clady, M. and B. Hutchinson. 1975. Effect of high winds on eggs of yellow perch, perca-
ﬂavescens, in Oneida Lake, New York. Transactions of the American Fisheries
Society 104:524-525.

Collins, M. R., S. G. Rogers, T. 1. J. Smith, and M. L. Moser. 2000. Primary factors
affecting sturgeon populations in the southeastern United States: Fishing mortality
and degradation of essential habitats. Bulletin of Marine Science 66:917-928.

Congdon, J. D., R. D. Nagle, A. E. Durrham, C. W. Beck, 0. M. Kinney, and S. R.
Yeomans. 1999. The relationship of body size to survivorship of hatchling
snapping turtles (Chelydra serpentine): an evaluation of the "bigger is better'"
hypothesis. Oecologia 1211224-235.

Cooke, S. J., S. G. Hinch, A. P. Farrell, M. F. Lapointe, S. R. M. Jones, J. S. Macdonald,
D. A. Patterson, M. C. Healey, and G. Van Der Kraak. 2004. Abnormal migration
timing and high en route mortality of sockeye salmon in the Fraser River, British
Columbia. Fisheries 29:22-33.

Cooper, N. W., M. T. Murphy, et a1. 2009. Age- and sex-dependent spring arrival dates of
Eastern Kingbirds. Journal of Field Ornithology 801: 35-41.

Cote, I. M. and M. R. Gross. 1993. Reduced disease in offspring - a beneﬁt of coloniality
in sunﬁsh. Behavioral Ecology and Sociobiology 33:269-274.

Couzin, I. D., J. Krause, et al. 2005. Effective leadership and decision-making in animal
groups on the move. Nature 4337025: 513-516. '

Crandall, K. A., B-E. O.R.P., et al. 2000. Considering evolutionary processes in
conservation biology. Trends in Ecology & Evolution 157: 290-295.

Cressie, N. 1993. Regional mapping of incidence rates using spatial Bayesian models.
Medical Care 31:YS60-YS65.

Crossin, G. T., S. G. Hinch, et al. 2003. Pink salmon Oncorhynchus gorbuscha migratory
energetics: reSponse to migratory difficulty and comparisons with sock eye
salmon Oncorhynchus nerka. Canadian Journal of Zoology-Revue Canadienne De
Zoologie 8112: 1986-1995.

Crossman, J. A. 2008. Evaluating collection, rearing and stocking methods for Lake

155

 

Sturgeon Acipenserﬁtlvescens restoration programs in the Great Lakes. Doctoral
Dissertation, Michigan State University, East Lansing.

Crowl, T. A., C. R. Townsend, N. Bouwes, and H. Thomas. 1997. Scales and causes of
patchiness in stream invertebrate assemblages: Top-down predator effects?
Journal of the North American Benthological Society 16:277-285.

Curry, R. A. and W. S. MacNeill. 2004. Population-level responses to sediment during
early life in brook trout. Journal of the North American Benthological Society
23:140-150.

Cushing, D. H. 1990. Plankton production and year-class strength in ﬁsh populations — an
update of the match mismatch hypothesis. Advances in Marine Biology 26: 249
293.

Dahl, J ., J. Dannewitz, et al. 2004. The timing of spawning migration: implications of
environmental variation, life history, and sex. Canadian Journal of Zoology-
Revue Canadienne De Zoologie 8212: 1864-1870.

Dahlberg, M. D. 1979. Review of survival rates of ﬁsh eggs and larvae in relation to
impact to assessments. Marine Fisheries Review 41 :1-12.

de Leaniz, C. G., I. A. Fleming, S. Einum, E. Verspoor, W. C. Jordan, S. Consuegra, N.
Aubin-Horth, D. Lajus, B. H. Letcher, A. F. Youngson, J. H. Webb, L. A.
Vollestad, B. Villanueva, A. Ferguson, and T. P. Quinn. 2007. A critical review of
adaptive genetic variation in Atlantic salmon: implications for conservation.
Biological Reviews 82:173-211.

Danchin, E., T. Boulinier, and M. Massot. 1998. Conspeciﬁc reproductive success and
breeding habitat selection: Implications for the study of coloniality. Ecology
79:2415-2428.

Deblois, E. M. and W. C. Leggett. 1991. Functional response and potential impact of
invertebrate predators on benthic ﬁsh eggs — analysis of the calliopius-

laeviusculus-capelin (mallotus-villosus) predator-prey system. Marine Ecology-
Progress Series 69:205-216.

deBruyn, A. M. H. and J. J. Meeuwi g. 2001. Detecting lunar cycles in marine ecology:
periodic regression versus categorical AN OVA. Marine Ecology-Progress Series
214: 307-310.

Dettlaff, T. A., A. S. Ginsburg, et al. 1993. Sturgeon Fishes. Developmental Biology and
Aquaculture. Berlin.

Dickerson, B. R., T. P. Quinn, et al. 2002. Body size, arrival date, and reproductive

156

success of pink salmon, Oncorhynchus gorbuscha. Ethology Ecology &
Evolution 141: 29-44.

Dittman, A. H., G. H. Brown, and C. J. Foote. 1998. The role of chemoreception in
sahnon-egg predation by coastrange (Cottus aleuticus) and slimy (C. cognatus)
sculpins in lliamna Lake, Alaska. Canadian Journal of Zoology 76:406-413.

Doak, D., P. Kareiva, and B. Kleptetka. 1994. Modeling Population Viability for the
Desert Tortoise in the Western Mojave Desert. Ecological Applications 4:446-
460.

Doctor, K. K. and T. P. Quinn. 2009. Potential for adaptation-by-time in sockeye salmon
Oncorhynchus nerka: the interactions of body size and in-stream reproductive life
span with date of arrival and breeding location. Canadian Journal of Zoology-
Revue Canadienne De Zoologie 878: 708-717.

Dohm, M. R. 2002. Repeatability estimates do not always set an upper limit to
heritability. Functional Ecology 162: 273-280.

Donazar, J. A, F. Hiraldo, and J. Bustamante. 1993. Factors inﬂuencing nest-site
selection, breeding density and breeding success in the bearded vulture (Gypaetus
barbatus). Journal of Applied Ecology 30:504-514.

Donovan, T. M., F. R Thompson, J. Faaborg, and J. R. Probst. 1995. Reproductive

success of migratory birds in habitat sources and sinks. Conservation Biology
9:1380-1395.

Doroshov, S. 1., G. P. Moberg, et al. 1997. Observations on the reproductive cycle of

cultured white sturgeon, Acipenser transmontanus. Environmental Biology of
Fishes 481-4: 265-278.

Douglas, S. G., G. Chaput, J. Hayward, and J. Sheasgreen. 2009. Prespawning,
Spawning, and Postspawning Behavior of Striped Bass in the Miramichi River.
Transactions Of The American Fisheries Society 138: 121-134.

Duriez, 0., S. Bauer, et al. 2009. What decision rules might pink-footed geese use to

depart on migration? An individual-based model. Behavioral Ecology 203: 560-
569.

Einum, S. and I. A. Fleming. 2000. Selection against late emergence and small offspring
in Atlantic salmon Salmo salar. Evolution 542: 628-639.

Einum, S. and I. A. Fleming. 2004. Does within-population variation in egg size reduce

intraspeciﬁc competition in Atlantic Salmon, Salmo salar? Functional Ecology
181: 110-115.

157

Einum, S. and I. Fleming. 2000. Selection against late emergence and small offspring in
Atlantic sahnon Salmo salar. Evolution 54: 628-639.

Ellrott, B. J ., J. E. Marsden, J. D. F itzsimons, J. L. Jonas, and R. M. Claramunt. 2007.
Effects of temperature and density on consumption of trout eggs by Orconectes
propinquus and 0. rustic-us. Journal of Great Lakes Research 3327-14.

Erickson, D. L. and M. A. H. Webb. 2007. Spawning periodicity, spawning migration,
and size at maturity of green sturgeon, Acipenser medirostrz’s, in the Rogue River,
Oregon. Environmental Biology of Fishes 793-4: 255-268.

Erkinaro, J ., F. Okland, et al. 1999. Return migration of Atlantic salmon in the River
Tana: the role of environmental factors. Journal of Fish Biology 55506-516: 506-
5 16.

Evans, D. O., and C. H. Olver. 1995. Introduction of lake trout (Salvelinus namaycush) to
inland lakes of Ontario, Canada: Factors contributing to successful colonization.
J oumal of Great Lakes Research 21 (Supplement 1):30-53.

Fahrig, L. 2002. Effect of habitat fragmentation on the extinction threshold: A synthesis.
Ecological Applications 12:346-353.

Falconer, D. S. and T. F. Mackay. 1996. Introduction to Quantitative Genetics, 4th edn.
Longrnan, Harlow.

Farrell, T. M., P. G. May, et al. 2009. Repeatability of Female Reproductive Traits in
Pigmy Rattlesnakes Sistrurus miliarz'us. Journal of Herpetology 432: 332-335.

F ausch, K. D., C. E. Torgersen, C. V. Baxter, and H. W. Li. 2002. Landscapes to
riverscapes: Bridging the gap between research and conservation of stream ﬁshes.
Bioscience 52:483-498.

Fillatre, E. K., P. Etherton, et al. 2003. Bimodal run distribution in a northern population
of sockeye salmon Oncorhynchus nerka: life history and genetic analysis on a
temporal scale. Molecular Ecology 127: 1793-1805.

F itzsimons, J. D., B. Williston, G. Williston, G. Bravener, J. L.‘Jonas, R. M. Claramunt,
J. E. Marsden, and B. J. Ellrott. 2006. laboratory Estimates of Salmonine Egg
Predation by Round Gobies (Neogobius melanostomus), Sculpins (Cottus
cognatus and C. bairdi), and Crayﬁsh (Orconectes propinquus). J oumal of Great
Lakes Research 32:227-241.

Fontaine, J. J. and T. E. Martin. 2006. Habitat selection responses of parents to offspring
predation risk: An experimental test. American Naturalist 168:811-818.

Fox, C. W. and M. E. Czesak. 2000. Evolutionary ecology of progeny size in arthropods.

158

Annual Review of Entomology 45: 341-369.

Frank, K. T. and W. C. Leggett. 1985. Reciprocal oscillations in densities of larval ﬁsh
and potential predators — a reﬂection of present of past predation. Canadian
Journal of Fisheries and Aquatic Sciences 4211: 1841-1849.

Gaillard, J. M. and N. G. Yoccoz. 2003. Temporal variation in survival of mammals: A
case of environmental canalization? Ecology 84:3294-3306.

Gall, G. A. E. and R. Neira. 2004. Genetic analysis of female reproduction traits of
fanned coho salmon Oncorhyncus kiSutch. Aquaculture 2341-4: 143-154.

Gascoigne, J. C. and R. N. Lipcius. 2004. Allee effects driven by predation Journal of
Applied Ecology 41 :801-810.

Geist, D. R and D. D. Dauble. 1998. Redd site selection and spawning habitat use by fall
chinook salmon: The importance of geomorphic features in large rivers.
Environmental Management 22:655-669.

Gelfand, A. E., A. M. Schmidt, S. Banerjee, and C. F. Sirmans. 2004. Nonstationary
multivariate process modeling through spatially varying coregionalization. Test
13:263-294.

Gessner, J., J. P. Van Eenennaam, and S. I. Doroshov. 2007. North American green and
European Atlantic sturgeon: comparisons of life histories and human impacts.
Environmental Biology of Fishes 79:397-411.

Giacomello, E., D. Marchini, and M. B. Rasotto. 2006. A male sexually dimorphic trait
provides antimicrobials to eggs in blenny ﬁsh. Biology Letters 2:330-333.

Gilbert, SF. 1997. Developmental Biology, 5th edn. Sunderland, MA: Sinauer
Associates

Gilkey, J. C. 1981. Mechanisms of fertilization in ﬁshes. American Zoologist 21:359-
375.

Godinho, A. L., B. Kynard, et al. 2007. Migration and spawning of female sumbim
Pseudoplatystoma corruscans, Pirnelodidae in the Sao Francisco river, Brazil.
Environmental Biology of Fishes 804: 421-433.

Gosselin, L. A. and P. Y. Qian. 1997. Juvenile mortality in benthic marine invertebrates.
Marine Ecology-Progress Series 146:265-282.

Greig, S. M., D. A. Sear, and P. A. Carling. 2007. A review of factors inﬂuencing the

availability of dissolved oxygen to incubating salmonid embryos. Hydrological
Processes 21:323-334.

159

Grieco, F., A J. van Noordwijk, et al. 2002. Evidence for the effect of learning on timing
of reproduction in blue tits. Science 2965565: 136-138.

Gross, M. R., J. Repka, C. T. Robertson, D. H. Secor, and W. Van Winkle. 2002.
Sturgeon Conservation: Insights from elasticity analysis. American Fisheries
Society Symposium: 183-200.

Gurrin, L. C., K. J. Scurrah, et al. 2005. Tutorial in biostatistics: spline smoothing with
linear mixed models. Statistics in Medicine 2421: 3361-3381.

Hakenkamp, C. C., H. M. Valett, and A. J. Boulton. 1993. Perspectives on the hyporheic
zone — Integrating hydrology and biology - concluding remarks. Journal of the
North American Benthological Society 12:94-99.

Hammer, C. 1995. Fatigue and exercise tests with ﬁsh. Comparative Biochemistry and
Physiology a-Physiology 1 12:1-20.

Hamor, T. and E. T. Garside. 1977. Size relations and yolk utilization in embryonated
ova and alevins of Atlantic salmon Salmo-solar L in various combinations of
temperature and dissolved oxygen. Canadian Journal of Zoology-Revue
Canadienne De Zoologie 55: 1 892-1898.

Hare, J. A. and R K. Cowen. 1997. Size, growth, development, and survival of the
planktonic larvae of Pomatomus saltatrix (Pisces: Pomatomidae). Ecology
78:2415-2431.

Harkness, W. J. and J. R. Dymond 1961. The lake sturgeon. Ontario Department of
Lands and Forests, Fish and Wildlife Branch, Toronto, Ontario.

HellawelJm, H. Leatham and Williams, G.I. 1974. Upstream migratory behavior of
salmonids in River, F rome Dorset. Journal of Fish Biology 66: 729-&.

Hendry, A. P. and T. Day. 2005. Population structure attributable to reproductive time:
isolation by time and adaptation by time. Molecular Ecology 144: 901-916.

Hendry, A. P. and T. P. Quinn. 1997. Variation in adult life history and morphology
among Lake Washington sockeye salmon (Oncorhynchus nerka) populations in
relation to habitat features and ancestral afﬁnities. Canadian J omnal of Fisheries
and Aquatic Sciences 54:75-84.

Hendry, A. P., F. E. Leonetti, and T. P. Quinn. 1995. Spatial and temporal isolating
mechanisms — The formation of discrete breeding aggregations of sockeye-

salmon (Oncorhynchus nerka). Canadian Journal of Zoology-Revue Canadienne
De Zoologie 73:339-352. ‘

160

Hendry, A. P., J. E. Hensleigh, and R. R Reisenbichler. 1998. Incubation temperature,
developmental biology, and the divergence of sockeye salmon (Oncorhynchus
nerka) within Lake Washington. Canadian Journal of Fisheries and Aquatic
Sciences 55:1387-1394.

Hendry, A. P., O. K. Berg, and T. P. Quinn. 2001. Breeding location choice in salmon: c
auses (habitat, competition, body size, energy stores) and consequences (life span,
energy stores). Oikos 93:407-418.

Hendry, A. P., O. K. Berg, et al. 1999. Condition dependence and adaptation-by-time:
breeding date, life history, and energy allocation within a population of salmon.
Oikos 853: 499-514.

Hemandez-Leon, S. 2008. Natural variability of ﬁsheries and lunar illumination: a
hypothesis. Fish and Fisheries 92: 138-154.

Hester, E. T. and M. N. Gooseff. 2010. Moving Beyond the Banks: Hyporheic
Restoration Is Fundamental to Restoring Ecological Services and Functions of
Streams. Environmental Science & Technology 44: 1521-1525.

High, B., C. A. Peery, et al. 2006. Temporary staging of Columbia River summer
steelhead in coolwater areas and its effect on migration rates. Transactions of the
American Fisheries Society 1352: 519-528.

Hilbom, R., T. P. Quinn, et al. 2003. Biocomplexity and ﬁsheries sustainability.
Proceedings of the National Academy of Sciences of the United States of America
1001 1: 6564-6568.

Hinch, S. G. and P. S. Rand. 1998. Swim speeds and energy use of upriver-migrating
sockeye salmon Oncorhynchus nerka: role of local environment and ﬁsh

characteristics. Canadian Journal of Fisheries and Aquatic Sciences 558: 1821-
l 83 1 .

Hipfner, J. M., L. A. McFarlane-Tranquilla and B. Addison. (2010). Experimental
evidence that both timing and parental quality affect breeding success in a
zooplanktivorous seabird. Auk 1271: 195-203.

Hodgson, S. and T. P. Quinn. 2002. The timing of adult sockeye salmon migration into
fresh water: adaptations by populations to prevailing thermal regimes. Canadian
Journal of Zoology-Revue Canadienne De Zoologie 803: 542-555.

Hoeting, J. A. 2009. The importance of accounting for spatial and temporal correlation in
analyses of ecological data. Ecological Applications 19:574-577.

Houde, E. D. 1987. Fish early life dynamics and recruitment variablity. American
Fisheries Society Symposium 2: 1 7-29.

161

Jager, H. 1., W. Van Winkle, J. A. Chandler, K. B. Lepla, P. Bates, and T. D. Counhan.
2002. A simulation study of factors controlling White Sturgeon recruitment in the
Snake River. American Fisheries Society Symposium 28:127-150.

Jennings, M. J ., J. E. Claussen, et al. 1996. Evidence for heritable preferences for

spawning habitat between two Walleye populations. Transactions of the American
Fisheries Society 1256: 978-982.

Jensen, L. F ., M. M. Hansen, C. Pertoldi, G. Holdensgaard, K. L. D. Mensberg, and V.
Loeschcke. 2008. Local adaptation in brown trout early life-history traits:
implications for climate change adaptability. Proceedings of the Royal Society B-
Biological Sciences 275:2859-2868.

Johnson, P. H. 1961. Walleye egg survival during incubation on several types of bottom
in Lake Winnibigoshish, Minnesota, and connecting waters. Transactions of the
American Fisheries Society 90:312-322.

Johnson, J. H., S. R. LaPan, R M. Klindt, and A. Schiavone. 2006. Lake sturgeon
spawning on artiﬁcial habitat in the St Lawrence River. Journal of Applied
Ichthyology 22:465-470.

Johnson, J. H., S. R. LaPan, R M. Klindt, and A. Schiavone. 2006. Lake sturgeon
spawning on artiﬁcial habitat in the St Lawrence River. Journal of Applied
Ichthyology 22:465-470.

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

Jonsson, N., L. P. Hansen, et al. 1991. Variation in Age, Size and Repeat Spawning of
Adult Atlantic salmon in Relation to River Discharge. Journal of Animal Ecology
603: 937-947.

Jorgensen, C., E. S. Dunlop, A. F. Opdal, and O. Fiksen. 2008. The evolution of

spawning migrations: Stage dependence and ﬁshing-induced changes. Ecology
89:3436—3448.

Juanes, F ., S. Gephard, et al. 2004. Long-terrn changes in migration timing of adult
Atlantic salmon Salmo salar at the southern edge of the species distribution.
Canadian Journal of Fisheries and Aquatic Sciences 6112: 2392-2400.

Karppinen, P., J. Erkinaro, et a1. 2004. Return migration of one-sea-winter Atlantic
salmon in the River Tana Journal of Fish Biology 645: 1179-1192.

Katvala, M., and A. Kaitala. 2001. Egg performance on an egg-carrying bug.
Experiments in the ﬁeld. Oikos 93:188-193.

162

Keefer, M. L., C. A Peery, M. A. Jepson, and L. C. Stuehrenberg. 2004. Upstream
migration rates of radio-tagged adult Chinook salmon in riverine habitats of the
Columbia River basin. Journal of Fish Biology 65:1126-1 141.

Keefer, M. L., C. A. Peery, M. A. Jepson, K. R. Tolotti, T. C. Bjomn, and L. C.
Stuehrenberg. 2004. Stock-speciﬁc migration timing of adult spring-summer
Chinook salmon in the Columbia River basin. North American Journal of
Fisheries Management 24:] 145-1162.

Keefer, M. L., M. L. Moser, et al. 2009. Effects of body size and river environment on
the upstream migration of adult Paciﬁc lampreys. North American Journal of
Fisheries Management 295: 1214-1224.

Keefer, M. L., M. L. Moser, et al. 2009. Variability in migration timing of adult Paciﬁc
lamprey Lampetra tridentata in the Columbia River, USA. Environmental
Biology of Fishes 853: 253-264.

Kemp, P. S., D. J. Gilvear, et al. 2006. Variation in performance reveals discharge-related
energy costs for foraging Atlantic salmon Salmo salar parr. Ecology of
Freshwater Fish 154: 565-571.

Kempinger, J. J. 1988. Spawning and early life history of lake sturgeon in the Lake
Winnebago system, Wisconsin. American Fisheries Society Symposium 5: 110-
122.

Kieffer, JD. and PW. Colgan. 1992. The role of learning in ﬁsh behaviour. Reviews in
Fish Biology and Fisheries 2: 125-143.

Kiﬂawi, M. and A 1. Mazeroll. 2006. Female leadership during migration and the
potential for sex-speciﬁc beneﬁts of mass spawning in the brown surgeonﬁsh
Acanthurus nigroﬁtscus. Environmental Biology of Fishes 761: 19-23.

Kingsolver, J. G. and R B. Huey. 1998. Evolutionary analyses of morphological and
physiological plasticity in thermally variable environnements. American
Zoologist 383: 545-560.

Kinnison, M., M. Unwin, N. Boustead, and T. Quinn. 1998. Population-speciﬁc variation
in body dimensions of adult chinook salmon (Oncorhynchus tshawytscha) ﬁom

New Zealand and their source population, 90 years after introduction. Canadian
Journal of Fisheries and Aquatic Sciences 55:554-563.

Kitancharoen, N., K. Hatai, and A. Yamamoto. 1997. Aquatic fungi developing on eggs
of salmonids. Journal of Aquatic Animal Health 9:314-316.

Knapp, R. A. and H. K. Preisler. 1999. Is it possible to predict habitat use by spawning

163

sahnonids? A test using California golden trout (Oncorhynchus mykiss
aguabonita). Canadian Journal of Fisheries and Aquatic Sciences 56: 1576-1584.

Knapp, R. A. and V. T. Vredenburg. 1996. Spawning by California golden trout:
Characteristics of spawning ﬁsh, seasonal and daily timing, redd characteristics,
and rnicrohabitat preferences. Transactions of the American Fisheries Society
125:519-531.

Knotek, W. L., and D. J. Orth. 1998. Survival for speciﬁc life intervals of smallmouth
bass, Micropterus dolomieu, during parental care. Environmental Biology of
Fishes 51:285-296.

Knouft, J. H., L. M. Page, and M. J. Plewa. 2003. Antimicrobial egg cleaning by the
fringed darter (Perciformes : Percidae : Etheostoma crossopterum): implications
of a novel component of parental care in ﬁshes. Proceedings of the Royal Society
of London Series B-Biological Sciences 270:2405-2411.

Kondolf, G. M., J. C. Vick, and T. M. Ramirez. 1996. Salmon spawning habitat
rehabilitation on the Merced River, California: An evaluation of project planning
and performance. Transactions of the American Fisheries Society 125:899-912.

Kondou, T., N. Takeshita, A. Nakazono, and S. Kimura. 2001. Egg survival in ﬂuvial
population of Masu Salmon in relation to intragravel conditions in spawning
redds. Transactions of the American Fisheries Society 130:969-974.

Kirparinen, A., R. B. O'Hara, et al. 2009. Lunar periodicity and the timing of river entry
in Atlantic salmon Salmo salar. Journal of Fish Biology 7410: 2401-2408.

LaHaye, M., A. Branchaud, et al. 1992. Reproduction, early life history, and
characteristics of the spawning grounds of the lake sturgeon Acipenserﬁtlvescens
in Des Prairies and L'Assomption rivers, near Montreal, Quebec. Canadian
Journal of Zoology 70: 1681-1689.

Lapointe, M., B. Eaton, S. Driscoll, and C. Latulippe. 2000. Modeling the probability of
salmonid egg pocket scour due to ﬂoods. Canadian Journal of Fisheries and
Aquatic Sciences 57:1120-1130.

e Y , D., S. H. Park, A. Takemura, and K. Takano. 2002. Histological observations of
’ seasonal reproductive and lunar-related spawning cycles in the female honeycomb
grouper Epinephelus merra in Okinawan waters. Fisheries Science 68:872-8 77.

Leggett’ W. C. 197 7. The ecology of ﬁsh migrations. Annual Review of Ecology and
Systematics 8: 285-308.

Lessells, C. M. and P. T. Boag. 1987. Unrepeatable repeatabilities — a common mistake.
Auk 1041: 116-121.

164

Linhart, O. and S. Kudo. 1997. Surface ultrastructure of paddleﬁsh eggs before and after
fertilization Journal of Fish Biology 51:573-582.

Lohmann, K. J ., J. T. Hester, et al. 1999. Long-distance navigation in sea turtles.
Ethology Ecology & Evolution 111: 1-23.

Lubow, B. C. and B. L. Smith. 2004. Population dynamics of the Jackson elk herd.
Journal of Wildlife Management 68:810-829.

Lundberg, P. 1985. Dominance behavior, body-weight and fat variations, and partial
migration in European Blackbirds T urdus-Memla. Behavioral Ecology and

Sociobiology 17:185-189.

Luschi, P., S. Benhamou, et al. 2007. Marine turtles use geomagnetic cues during open-
sea homing. Current Biology 172: 126-133.

Magnin, E. 1966. Quelques donnees biologiques sur la reproduction des esturgeons
Acipenserﬁtlvescens raf de la riviere nottaway tributaire de la baie james.

Canadian Journal of Zoology 442: 257-270.

Malcolm, I. A., A. F. Youngson, and C. Soulsby. 2003. Survival of salmonid eggs in a
degraded gravel-bed stream: Effects of groundwater-surface water interactions.

River Research and Applications 19:303-313.

Malcolm, I. A, C. Soulsby, A. F. Youngson, and D. M. Hannah. 2005. Catchment-scale
controls on groundwater-surface water interactions in the hyporheic zone:
Implications for salmon embryo survival. River Research and Applications

2 1 :977-989.

Malcohn,1. A, C. Soulsby, A. F. Youngson, and D. Tetzlaff 2009. Fine scale variability
of hyporheic hydrochemistry in salmon spawning gravels with contrasting
groundwater-surface water interactions. Hydrogeology Journal 17:161-174.

Malcolm, I. A, C. Soulsby, A. F. Youngson, D. M. Hannah, 1. S. McLaren, and A.
Thorne. 2004. Hydrological inﬂuences on hyporheic water quality: implications
for salmon egg survival. Hydrological Processes 18:1543-1560.

Martin, T. E. 1995. Avian Life-History Evolution in Relation to Nest Sites, Nest
Predation, and Food. Ecological Monographs 65:101-127.

Mayﬁeld, R. B. and J. J. Cech. 2004. Temperature effects on green sturgeon
bioenergetics. Transactions of The American Fisheries Society 133:961-970.

McMichael, G. A. C. A. McKinstry, et al. 2005. Fall Chinook salmon spawning activity

165

versus daylight and ﬂow in the tailrace of a large hydroelectric Dam. North
American Journal of Fisheries Management 252: 573-580.

McNaught, M. K and 1. P. F. Owens. 2002. Interspeciﬁc variation in plumage colour
among birds: species recognition or light environment? Journal of Evolutionary

Biology 154: 505-514.

Melles, S. J., D. Badzinski, M. J. Fortin, F. Csillag, and K. Lindsay. 2009. Disentangling
habitat and social drivers of nesting patterns in songbirds. Landscape Ecology
24:519-531.

McKinley, S., G. Van Der Kraak, and G. Power. 1998. Seasonal migrations and
reproductive patterns in the lake sturgeon, Acipenserﬁtlvescens, in the vicinity of
hydroelectric stations in northern Ontario. Environmental Biology of Fishes

51 :245-256.

Miller, A. I. and L. G. Beckman. 1996. First record of predation on White Sturgeon eggs
by sympatric ﬁshes. Transactions of The American Fisheries Society 125:338-
340.

Mills, C. A. 1981. The Attachment of Dace, Leuciscus leuciscus L, Eggs to the Spawning
Substratum and the Inﬂuence of Changes in Water Current on Their Survival.
Journal of Fish Biology 19:129-134.

Mion, J ., R. Stein, and E. Marschall. 1998. River discharge drives survival of larval
walleye. Ecological Applications 8:88-103.

Moller, A P. 1994. Phenotype - dependent arrival time and its consequences in a
migratory bird. Behavioral Ecology and Sociobiology 352: 115-122.

Morbey, Y. E. 2003. Pair formation, pre-spawning waiting, and protandry in kokanee,
Oncorhynchus nerka. Behavioral Ecology and Sociobiology 542: 127-135.

Mulli, K E. and M. A. Wilzbach. 2007. Selection of spawning sites by coho salmon in a
northern California stream North American Journal of Fisheries Management
27: 1343-1354.

Mllths, E., P. S. Com, A P. Pessier, and D. E. Green. 2003. Evidence for disease-related
amphibian decline in Colorado. Biological Conservation 110:357-365.

Myers, R. A. 1998. When do environment-recnritrnent correlations work? Reviews in
Fish Biology and Fisheries 8:285-305.

Nichols, S. J ., G. Kennedy, E. Crawford, J. Allen, J. 1. French, G. Black, M. Blouin, J.

166

Kickey, S. Chemyak, R Haas, and M. Thomas. 2003. Assessment of lake
sturgeon (Acipenserﬁrlvescens) spawning efforts in the lower St. Clair River,
Michigan. Journal of Great Lakes Research 29:3 83-391.

Niemela, E., P. Orell, et al. 2006. Previously spawned Atlantic salmon ascend a large
subarctic river earlier than their maiden counterparts. Journal of Fish Biology 694:

1151-1163.

Norris, K. 1993. Seasonal-variation in the reproductive success of blue tits - an
experimental study. Journal of Animal Ecology 622: 287-294.

Northcote, T. G. and S. D. Gerking. 1978. Migratory strategies and production in
freshwater ﬁshes. Ecology of freshwater ﬁsh production: 326-359.

Olson, D.E., Shupp, DH. and Macins, V. 1978. A hypothesis of homing behavior of
walleyes as related to observed food patterns of passive and active movement.
American Fisheries Society of Special Publication 11, 52-57.

Opperman, J. J., K. A. Lohse, C. Brooks, N. M. Kelly, and A. M. Merenlender. 2005.
Inﬂuence of land use on ﬁne sediment in salmonid spawning gravels within the
Russian River Basin, California. Canadian Journal of Fisheries and Aquatic
Sciences 62:2740—275 1.

Orians, G. H. and J. F. Wittenberger. 1991. Spatial and temporal scales in habitat
selection. American Naturalist 137: S29-S49.

Osman, R. W. and R. B. Whitlatch. 1995. Predation on early ontogenic life stages and its
effect on recruitment into a marine epifaunal community. Marine Ecology
Progress Series 117:11 1-126.

Overton, A. and R. Rulifson. 2009. Annual variability in upstream migration of glass eels
in a southern USA coastal watershed. Environmental Biology of Fishes 841: 29-
37.

Paige, R. L., S. Sun, et a1. 2009. Variance Reduction in Smoothing Splines. Scandinavian
Journal of Statistics 361: 1 12-126.

Paitz. R. T., H. K Harms, et al. 2007. Experience pays: offspring survival increases with
female age. Biology Letters 31: 44-46.

Palm, D., M. Lindberg, E. Brannas, H. Lundqvist, J. Ostergren, and U. Carlsson. 2009.
Inﬂuence of European sculpin, Cottus gobio, on Atlantic salmon Salmo salar,
recruitment and the effect of gravel size on egg predation - implications for
spawning habitat restoration. Fisheries Management and Ecology 16:501-507.

Pankhurst, N. W. and H. R. King. 2010. Temperature and salmonid reproduction:

167

 

 

 

implications for aquaculture. Journal of Fish Biology 761: 69-85.

Papa, R., J. A. Israel, et al. 2007. Assessment of genetic variation between reproductive
ecotypes

Papa, R, J. A. Israel, et al. 2007. Assessment of genetic variation between reproductive
ecotypes of Klamath River steelhead reveals differentiation associated with
different run-timings. Journal of Applied Ichthyology 232: 142- 146.

Paragamian, V. L. and G. O. Kruse. 2001. Kootenai River White Sturgeon spawning
migration behavior and a predictive model. North American Journal of Fisheries
Management 21: 10-21.

Paragamian, V. L., R. McDonald, G. J. Nelson, and G. Barton. 2009. Kootenai River
velocities, depth, and white sturgeon spawning site selection - a mystery
unraveled? Journal of Applied Ichthyology 25:640-646.

Paragamian, V. L., V. D. Wakkinen, et al. 2002. Spawning locations and movement of
Kootenai River white sturgeon. Journal of Applied Ichthyology 184-6: 608-616.

Paragrnian, V. L., R C. P. Beamesderfer, et a1. 2005. Status, population dynamics, and
future prospects of the endangered Kootenai River White Sturgeon populations
with and without hatchery intervention. Transactions of the American Fisheries
Society 134: 518 532.

Parsley, M. J ., N. D. Popoff, et al. 2008. Seasonal and diel movements of white sturgeon
in the lower Columbia River. Transactions of the American Fisheries Society
1374: 1007-1017.

Parsley, M. J ., P. J. Anders, A. 1. Miller, L. G. Beckman, and G. T. J. McCabe. 2002.
Recovery of white sturgeon populations through natural production:
Understanding the inﬂuence of Abiotic and biotic factors on spawning and
subsequent recruitment. American Fisheries Society Symposium 28:55-66.

Paxton, C. G. M. and L. G. Willoughby. 2000. Resistance of perch eggs to attack by
aquatic fungi. Journal of Fish Biology 57:562-570.

Pcake, S., F. W. H. Beamish, R S. McKinley, D. A. Scruton, and C. Katopodis. 1997.
Relating swimming performance of lake sturgeon, Acipenserﬁdvescens, to
ﬁshway design. Canadian Journal of Fisheries And Aquatic Sciences 54:1361-

1366.

Petersen, C. W., R. R. Warner, S. Cohen, H. C. Hess, and A T. Sewell. 1992. Variable
pelagic fertilization success —implications for mate choice and spatial patterns of
mating. Ecology 73:391-401.

168

 

Peterson, D. L., P. Vecsei, and C. A. Jennings. 2007. Ecology and biology of the lake
sturgeon: a synthesis of current knowledge of a threatened North American
Acipenseridae. Reviews in Fish Biology and Fisheries 17:59-76.

Pike, D. A. 2008. Environmental correlates of nesting in loggerhead turtles, Caretta
caretta. Animal Behaviom' 76:603-610.

Portner, H. O. and A. P. Farrell. 2008. Ecology physiology and climate change.
Science 3225902: 690-692.

Priegel, G. R. and T. L. Wirth. 1977. The lake sturgeon: its life history, ecology and
management Wisconsin Department of Natural Resources, Publication 4-360077,
Madison, WI.

Psenicka, M., M. Rodina, and O. Linhart 2010. Ultrastructural study on the fertilization
process in sturgeon (A cipenser), frmction of acrosome and prevention of
polyspermy. Animal Reproduction Science 117:147-154.

Pulido, F., P. Berthold, et al. 2001. Heritability of the timing of autumn migration in a
natural bird population. Proceedings of the Royal Society of London Series B-
Biological Sciences 2681470: 953-959.

Pyper, B. J. and R. M. Peterman. 1998. Comparison of methods to account for
autocorrelation in correlation analyses of ﬁsh data. Canadian Journal Of Fisheries
And Aquatic Sciences 55:2127-2140.

Quinn, T. P. 1999. Variation in Paciﬁc salmon reproductive behaviour associated with
species, sex and levels of competition. Behaviour 136: 179-204.

Quinn, T. P. and D. J. Adams. 1996. Environmental changes affecting the migratory
timing of American shad and sockeye salmon. Ecology 77:1 151-1162.

Quinn, T. P., A. P. Hendry, and L. A Wetzel. 1995. The inﬂuence of life history trade-
offs and the size of incubation gravels on egg size variation in sockeye salmon
(Oncorhynchus nerka). Oikios 74:425-43 8.

Quinn, T. P., E. Graynoth, C. C. Wood, and C. J. Foote. 1998. Genotypic and phenotypic
divergence of sockeye salmon in New Zealand from their ancestral British
Columbia populations. Transactions of the American Fisheries Society 127:517-
534.

Qllinn, T. P., I. J. Stewart, et al. 2006. Experimental evidence of homing to site of
incubation by mature sockeye salmon, Oncorhynchus nerka. Animal Behaviour
72: 941-949.

QUinn, T. P., J. A. Peterson, et al. 2002. Artiﬁcial selection and environmental change:

169

 

Countervailing factors affecting the timing of spawning by coho and chinook
salmon. Transactions of the American Fisheries Society 1314: 591-598.

Quinn, T. P., M. J. Unwin, et al. 2000. Evolution of temporal isolation in the wild:
Genetic divergence in timing of migration and breeding by introduced chinook

salmon populations. Evolution 544: 1372-1385.

Quinn, T. P., S. Hodgson, et al. 2007. Directional selection by ﬁsheries and the timing of
sockeye salmon Oncorhynchus nerka migrations. Ecological Applications 17 3:
731-739.

Quinn, T.P., M.J. Unwin, and MT Kinnison. 2000. Evolution of temporal isolation in
the wild: genetic divergence in timing of migration and breeding by introduced
Chinook salmon populations. Evolution 54:1372-1385.

Quinton, C. D., L. R. McKay, et al. 2004. Strain and maturation effects on female
spawning time in diallel crosses of three strains of rainbow trout Oncorhynchus
mykiss. Aquaculture 2341-4: 99-110.

Quinn, T. P., S. Hodgson, and C. Peven. 1997. Temperature, ﬂow, and the migration of
adult sockeye salmon (Oncorhynchus nerka) in the Columbia River. Canadian
Journal Of Fisheries And Aquatic Sciences 54: 1349-1360.

Rakowitz, G., B. Berger, J. Kubecka, and H. Keckeis. 2008. Functional role of
environmental stimuli for the spawning migration in Danube nase Chondrostoma
nasus (L.). Ecology of Freshwater Fish 17:502-514.

Rach, J. J ., J. A. Marks, and V. K. Dawson. 1995. Effect of water-ﬂow rates in hatching
jars to control ﬁrngal infections of rainbow trout eggs. Progressive Fish-Culturist
57:226-230.

Rand, P. S. and S. G. Hinch. 1998. Swim speeds and energy use of upriver-migrating
sockeye salmon (Oncorhynchus nerka): simulating metabolic power and assessing
risk of energy depletion Canadian Journal of Fisheries And Aquatic Sciences

55:1832-1841.

Reznick, D. N. and C. K. Ghalambor. 2001. The population ecology of contemporary
adaptations: what empirical studies reveal about the conditions that promote

adaptive evolution. Genetica 112: 183-198.

Rideout, R. M., E. A. Trippel], and M. K. Litvak 2005. Effects of egg size, food supply
and spawning time on early life history success of haddock Melanogrammus
aegleﬁnus. Marine Ecology-Progress Series 285: 169-180.

Rochard, E., G. Castelnaud, and M. Lepage. 1990. Sturgeons (pisces, Acipenseridae) —
threats and prospects. Journal of Fish Biology 37:123-132.

170

 

‘mm: «W
.:

 

Rodewald, A. D. and R. H. Yahner. 2001. Avian nesting success in forested landscapes:
Inﬂuence of landscape composition, stand and nest-patch microhabitat, and biotic
interactions. Auk l 18: 1018-1028.

Roff, D. A. 2002. Life History Evolution, Sinauer Associates, Sunderland, MA.

Rogers, M. W., M. S. Allen, et al. 2006. Separating genetic and environmental inﬂuences
on temporal spawning distributions of largemouth bass Micropterus salmoides.
Canadian Journal of Fisheries and Aquatic Sciences 6311: 2391-2399.

Roussow, G. 1957. Some considerations concerning sturgeon spawning periodicity.
Journal of the Fisheries Research Board of Canada 144: 553.

Rupp, R. S. 1965. Shore spawning and survival of eggs of American smelt. Transactions
of the American Fisheries Society 94:160-165. ‘

Salinger, D. H. and J. J. Anderson. 2006. Eﬁ’ects of water temperature and ﬂow on adult
salmon migration swim speed and delay. Transactions of the American Fisheries
Society 1351: 188-199.

Savino, J. F., P. L. Hudson, M. C. Fabrizio, and C. A Bowen. 1999. Predation on lake
trout eggs and ﬂy: a modeling approach. Journal of Great Lakes Research 25:36-
44.

Scott, W.B. and El Crossman 1973. Freshwater ﬁshes of Canada. Bulletin 184,
Fisheries Research Board of Canada, Ottawa, 966 pp.

Schlacpfer, M. A, M. C. Runge, and P. W. Sherman. 2002. Ecological and evolutionary
traps. Trends in Ecology & Evolution 17:474-480.

Secor, D. H. and E. D. Houde. 1995. Temperature effects on the timing of striped bass
egg-production, larval viability, and recruitment potential in the Patuxent River
Chesapeake Bay. Estuaries 183: 527-544.

Sheldon, B. C., L. E. B. Kruuk, et al. 2003. Natural selection and inheritance of breeding
time and clutch size in the collared ﬂycatcher. Evolution 572: 406-420.

Sherrill-Mix, S. A, M. C. James, et al. 2008. Migration cues and timing in leatherback
sea turtles. Behavioral Ecology 192: 231-236.

Shine. R, M. J. Elphick, and P. s. Harlow. 1997. The inﬂuence of natural incubation
environments on the phenotypic traits of hatchling lizards. Ecology 78:2559-
2568.

Shmueli, M., I. llzhaki, A. Arieli, and Z. Arad. 2000. Energy requirements of migrating

171

 

T'Q—‘T"

Great White Pelicans Pelecanus onocrotalus. Ibis 142:208-216.

Siitonen, L. and G. A. E. Gall. 1989. Response to selection for early spawn date in
rainbow-trout, Salmo gairdneri. Aquaculture 782: 153-161.

Sillett, T. S. and R. T. Holmes. 2002. Variation in survivorship of a migratory songbird
throughout its annual cycle. Journal of Animal Ecology 71:296-308.

Singer, T. D., V. G. Mahadevappa, et al. 1990. Aspects of the energy-metabolism of lake
sturgeon, Acipenserﬁtlvescens, with special emphasis on lipid and ketone-body
metabolism. Canadian Journal of Fisheries and Aquatic Sciences 475: 873-881.

Smith, I. P. and G. W. Smith. 1997. Tidal and diel timing of river entry by adult Atlantic
salmon returning to the Aberdeenshire Dee, Scotland. Journal of Fish Biology
503: 463-474.

Smith, K. M. and D. K. King. 2005. Dynamics and extent of larval lake sturgeon
Acipenserﬁrlvescens drift in the Upper Black River, Michigan. Journal of Applied
Ichthyology 21: 161-168.

Smith, K. M. and D. K. King. 2005. Movement and habitat use of yearling and juvenile
lake sturgeon in Black Lake, Michigan. Transactions of the American Fisheries
Society 13421159-1172.

Smith, K. M. and E. A. Baker. 2005. Characteristics of spawning lake sturgeon in the
Upper Black River, Michigan. North American Journal of Fisheries Management
25: 301-307. '

Smith, R J. and F. R. Moore. 2005. Arrival timing and seasonal reproductive
performance in a long-distance migratory land bird. Behavioral Ecology and
Sociobiology 573: 231-239.

Smith, S. J. and J. E. Marsden. 2009. Factors Affecting Sea Lamprey Egg Survival. North
American Journal of Fisheries Management 29:859-868.

SOWden, T. K. and G. Power. 1985. Prediction of rainbow trout embryo survival in
relation to groundwater seepage and particle size of spawning substrates.
Transactions Of The American Fisheries Society 114:804-812.

SPiegelhalter, D. J ., N. G. Best, B. R Carlin, and A. van der Linde. 2002. Bayesian
measures of model complexity and ﬁt. Journal of the Royal Statistical Society
Series B-Statistical Methodology 64:583-616.

Standen, E. M., S. G. Hinch, and P. S. Rand. 2004. Inﬂuence of river speed on path

selection by migrating adult sockeye salmon (Oncorhynchus nerka). Canadian
Journal of Fisheries and Aquatic Sciences 61:905-912.

172

Stanford, J. A., J. V. Ward, W. J. Liss, C. A. Frissell, R. N. Williams, J. A. Lichatowich,
and C. C. Coutant. 1996. A general protocol for restoration of regulated rivers.
Regulated Rivers-Research & Management 12:391-413.

Stefan, H. G. and B. A. Sinokrot. 1993. Projected global clirnate-change impact on water
temperatures in 5 north central United States streams. Climatic Change 244: 353-
381.

Stewart, I. J., S. M. Carlson, C. P. Boatright, G. B. Buck, and T. P. Quinn. 2004. Site
ﬁdelity of spawning sockeye salmon (Oncorhynchus nerka W.) in the presence
and absence of olfactory cues. Ecology of Freshwater Fish 13: 104-1 10.

Stillwell, R. C. and C. W. Fox. 2005. Complex patterns of phenotypic plasticity:
Interactive effects of temperature during rearing and oviposition. Ecology 864:
924-934.

Stokes, D. L. and P. D. Boersma. 1998. Nest-site characteristics and reproductive success
in magellanic penguins (Spheniscus magellanicus). Auk 115:34-49.

Sugg, D. W., R. K Chesser, et a1. 1996. Population genetics meets behavioral ecology.
Trends in Ecology & Evolution 118: 338-342.

Sulak, K. J. and J. P. Clugston. 1998. Early life history stages of Gulf sturgeon in the
Suwannee River, Florida. Transactions of the American Fisheries Society 1275:
75 8-77 1 .

Svendsen, J. C., A. Koed, et al. 2004. Factors inﬂuencing the spawning migration of
female anadromous brown trout. Journal of Fish Biology 642: 528-540.

Taborsky, B. and K. Foerster. 2004. Female mouthbrooders adjust incubation duration to
perceived risk of predation. Animal Behaviour 68: 1275-128 1.

Takahashi, D., H. Asada, T. Takeyama, M. Takahata, R Katoh, S. Awata, and M. Kohda.
2004. Why egg-caring males of Isaza (Gymnogobius isaza, Gobiidae) reﬁrse
additional females: preliminary ﬁeld observations. Journal of Ethology 22:153-
159.

Taylor, M. H. 1986. Environmental and endocrine inﬂuences on reproduction of
Fundulus heteroclitus. American Zoologist 261: 159-171.

Taranger, G. L., C. Haux, S. O. Stefansson, B. T. Bjomsson, B. T. Walther, and T.
Hansen 1998. Abrupt changes in photoperiod affect age at maturity, timing of
ovulation and plasma testosterone and oestradiol-l7 beta proﬁles in Atlantic
salmon, Salmo salar. Aquaculture 162:85-98.

173

 

 

Tetzlaff, D., C. Soulsby, A. F. Youngson, C. Gibbins, P. J. Bacon, 1. A. Malcolm, and S.
Langan. 2005. Variability in stream discharge and temperature: a preliminary
assessment of the implications for juvenile and spawning Atlantic salmon.
Hydrology and Earth System Sciences 9:193-208.

Thorstad, E. B., F. Okland, et al. 2008. Factors affecting the within-river spawning
migration of Atlantic salmon, with emphasis on human impacts. Reviews in Fish
Biology and Fisheries 184: 345-371.

Thurow, R. F. and J. G. King. 1994. Attributes of Yellowstone cutthroat trout redds in a
tributary of the Snake River, Idaho. Transactions of the American Fisheries
Society 123:37-50.

Trepanier, S., M. A. Rodriguez, et al. 1996. Spawning migrations in landlocked Atlantic
salmon: Time series modeling of river discharge and water temperature effects.
Journal of Fish Biology 485: 925-936.

Tripp, S. J ., Q. E. Phelps, R. E. Colombo, J. E. Garvey, B. M. Burr, D. P. Herzog, and R.
A. Hrabik 2009. Maturation and Reproduction of Shovelnose Sturgeon in the
Middle Mississippi River. North American Journal of Fisheries Management
29:730-738.

Trippel, E. A. 1998. Egg size and viability and seasonal offspring production of young
Atlantic cod. Transactions of the American Fisheries Society 1273: 339-359.

 

Utne-Palrn, A C. 2001. Response of naive two-spotted gobies Gobiusculusﬂavescens to
visual and chemical stimuli of their natural predator, cod Gadus morhua. Marine
Ecology-Progress Series 218:267-274.

Vandenbos, R E., W. M. Tonn, and S. M. Boss. 2006. Cascading life-history
interactions: alternative density-dependent pathways drive recruitment dynamics
in a ﬁ'eshwater ﬁsh. Oecologia 148:573-582.

Van Eenennaam, J. P., J. Linares-Casenave, et al. 2005. Effect of incubation temperature
on green sturgeon embroys, Acipenser medirostris. Environmental Biology of
Fishes 72: 145-154.

Ventling-Schwank, A. R. and D. M. Livingstone. 1994. Transport and burial as a cause of
Whiteﬁsh (Coregonus sp.) egg mortality in a eutrophic lake. Canadian Journal of
Fisheries And Aquatic Sciences 51.

Verbyla, A. P., B. R. Cullis, et al. 1999. The analysis of designed experiments and
longitudinal data by using smoothing splines. Journal of the Royal Statistical
Society Series C-Applied Statistics 48: 269-300.

Wang, Y. D. 1998. Mixed effects smoothing spline analysis of variance. Journal of the

174

Royal Statistical Society Series B-Statistical Methodology 60: 159-174.

Waples, R. S. 2002. Effective size of ﬂuctuating sahnon populations. Genetics 161: 783-
791.

Webb, J. K, G. P. Brown, and R Shine. 2001. Body size, locomotor speed and
antipredator behaviour in a tropical snake (T ropidonophis mairii, Colubridae): the

inﬂuence of incubation environments and genetic factors. Functional Ecology
1 5 :561-568.

Webb, J. H. and H. A. McLay. 1996. Variation in the time of spawning of Atlantic
salmon (Salmo solar) and its relationship to temperature in the Aberdeenshire
Dee, Scotland Canadian Journal Of Fisheries And Aquatic Sciences 5322739-
2744.

Webb, M. A. H., J. P. Van Eenennaam, et al. 2001. Effects of thermal regime on ovarian
maturation and plasma sex steroids in farmed white sturgeon, Acipenser
transmontanus. Aquaculture 2011-2: 137-151.

Webb, P. W. 1986. Kinematics of lake sturgeon, Acrpenser-ﬁtlvescens, at cruising speeds.
Canadian Journal of Zoology-Revue Canadienne De Zoologie 6410: 2137-2141.

Wedekind, C. 2002. Induced hatching to avoid infectious egg disease in Whiteﬁsh.
Current Biology 12:69-71.

Wiegand, T., J. Naves, T. Stephan, and A. Fernandez. 1998. Assessing the risk of
extinction for the brown bear (Ursus arctos) in the Cordillera Cantabrica, Spain.
Ecological Monographs 68:539-570.

Williams, B.K. and J .D. Nichols. 1984. Optimal timing in biological processes. American
Naturalist 123: 1—19.

Williamson, CE, W. Dodds, T.K Kratz, and MA. Palmer. 2008. Lakes and streams as
sentinels of environmental change in terrestrial and atmospheric processes.
Frontiers in Ecology and Evolution 6: 247-254.

Willson, M. F. and K. C. Halupka. 1995. Anadromous ﬁsh as keystone species in
vertebrate communities.Conservation Biology 9:489-497.

Wilson, D. S. 1998. Nest-site selection: Microhabitat variation and its effects on the
survival of turtle embryos. Ecology 79: 1884-1892.

Winemiller, K. O. and K. A Rose. 1992. Patterns of life-history diversiﬁcation in North

American ﬁshes: Implications for population regulation. Canadian Journal of
Fisheries And Aquatic Sciences 49:2196-2218.

175

Wisdom, M. J., L. S. Mills, and D. F. Doak. 2000. Life stage simulation analysis:
Estimating vital-rate effects on population growth for conservation. Ecology
81:628-641.

Wollebaek, J ., R. Thue, and J. Heggenes. 2008. Redd site microhabitat utilization and
quantitative models for wild large brown trout in three contrasting boreal rivers.
North American Journal of Fisheries Management 28: 1249-1258.

Woody, C. A, J. Olsen, et al. 2000. Temporal Variation in Phenotypic and Genotypic
Traits in Two Sockeye Salmon Populations, Tustumena Lake, Alaska.
Transactions of the American Fisheries Society 1294: 1031-1043.

Workman, R. D., D. B. Hayes, et al. 2002. A model of steelhead movement in relation to
water temperature in two Lake Michigan tributaries. Transactions of the
American Fisheries Society 1313: 463-475.

Youngson, A. F ., I. A. Malcolm, J. L. Thorley, P. J. Bacon, and C. Soulsby. 2004. Long-
residence groundwater effects on incubating salmonid eggs: low hyporheic
oxygen impairs embryo development. Canadian Journal of Fisheries and Aquatic
Sciences 61 :2278-2287.

Zeh, M. and W. Donni. 1994. Restoration of spawning grounds for trout and grayling in
the river High Rhine. Aquatic Sciences 56:59-69.

Zhang, H., Q. W. Wei, and H. Du. 2009. A bedform morphology hypothesis for
spawning areas of Chinese sturgeon. Environmental Biology of Fishes 84: 199-
208.

176

111111

93 03063 5167

H
H
H
l
H
E“
II
N“
l
I
H

12

1111111

3