MOVEMENT ECOLOGY OF WHITE STURGEON IN THE REGULATED UPPER
COLUMBIA RIVER
by
Caleb Jetter
B.A., University of California, Santa Cruz, 2014

THESIS SUBMITTED IN PARTIAL FULFILLMENT OF
THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE
IN
NATURAL RESOURCES AND ENVIRONMENTAL SCIENCE

UNIVERSITY OF NORTHERN BRITISH COLUMBIA
April 2022
© Caleb Jetter, 2022

ii

ABSTRACT
Advancements in animal tracking technology and analytical techniques have expanded the
disciple of movement ecology. An improved understanding of where, when, and why an animal
moves can greatly assist in the management and recovery of endangered species. White sturgeon
(Acipenser transmontanus) in the Upper Columbia River (UCR) are listed as Endangered due to
their considerable decline over the last century. Habitat fragmentation, caused by the
construction of dams for hydroelectric generation and river regulation, has greatly impacted
white sturgeon populations by preventing access to critical habitat, impacting water quality, and
modifying natural flow regimes. The goal of this study was to apply more advanced analytical
techniques to white sturgeon movement datasets from a regulated and transboundary reach of the
UCR, so as to further understand white sturgeon movement and behaviour in relation to river
regulation. Using hidden Markov models (HMMs) and generalized linear mixed models
(GLMMs), this study investigated the fine-scale movements of wild adult white sturgeon
residing immediately downstream of a hydroelectric operation and the coarse-scale movements
of hatchery-origin white sturgeon residing throughout the Transboundary Reach of the UCR. In
my fine-scale dataset, two behaviour states were identified: a “residential” behaviour
representing short movements with persistent direction, and a “transitory” behaviour
representing longer movements with frequent turns. Results revealed that white sturgeon were
very likely to remain in a behaviour state, with environmental covariates influencing the
probability of a behaviour state only when there was less certainty of the state in the previous
timestep (i.e. probability of 50%). Overall, water temperature appeared to influence what
behaviour state was more predominantly expressed while d ischarge appeared to influence where
each behaviour state was expressed. Warming water temperatures related to decreased
probabilities of residential behaviour and changes in discharge shifted the location of residential
behaviour within eddy and tailrace habitats. In my coarse-scale dataset, two behaviour states
were also identified: a similar “residential” behaviour state representing localized movement at a
single receiver station, and a “transitory” state representing movements between receiver
stations. Overall, hatchery-origin white sturgeon moved very little and demonstrated strong site
fidelity. When fish did move, location (i.e. country, river zone), discharge, and water temperature
had the greatest influence on maximum displacement and behaviour probability. Increases in
discharge related to faster declines in maximum displacement in Canad a compared to the United
States, as well as fish in more channelized river zones having increased probabilities of
residential behaviour when previously in a transitory state. Warming water temperatures related
to increases in maximum displacement and decreases in the probability of residential behaviour
(when previously in a transitory state). In both fine-scale and coarse-scale datasets, trends could
be observed seasonally, with increased movement in the summer and fall compared to the winter
and spring. Results from this study better described white sturgeon interactions with dams, help
describe the effectiveness of recovery measures like conservation aquaculture, and inform
management decisions related to river regulation.

iii

TABLE OF CONTENTS

Abstract

ii

Table of Contents

iii

List of Tables

v

List of Figures

vi

Acknowledgements

xi

Chapter One

Introduction
References

1
5

Chapter Two

Fine-Scale Movement Ecology and Behaviour Patterns of
White Sturgeon in the Regulated Upper Columbia River
Introduction
Methods
Study Area
Fish Sampling and Tagging
Acoustic Array
Data Filtering
Movement Behaviour
Model Predictions
Model Selection
Results
Data Filtering
Movement Behaviour
Model Selection and Predictions
Discussion
Tables
Figures
References

8
9
12
12
13
14
15
15
17
19
20
20
20
21
23
32
34
55

iv

TABLE OF CONTENTS CONTINUED
Chapter Three

Chapter Four

Coarse-Scale Movement Ecology and Behaviour Patterns of
Hatchery-Origin White Sturgeon in the Regulated Upper
Columbia River
Introduction
Methods
Study Area
Fish Sampling and Tagging
Acoustic Array
Data Filtering
Movement Behaviour Analysis
Maximum Displacement Analysis
Model Selection
Results
Movement Behaviour Analysis
Maximum Displacement Analysis
Discussion
Tables
Figures
References

62

Conclusion
References

113
120

63
66
66
68
70
72
72
75
75
76
77
79
80
88
90
106

v

LIST OF TABLES
2.1 Models predicting the probability of residential behaviour given the probability of residential
behaviour at time t-1 ranked by increasing order of the bias-corrected Akaike Information
Criterion (AIC c). Differences in AIC c relative to the top model (ΔAIC c), weighted AIC c (wAIC c),
cumulative weighted AIC c (cwAIC c), log-likelihood (log(L)), and the number of parameters in a
model (K) are shown. The top two models were selected for averaged predictions. ………… 32
2.2 Models predicting the mean weekly probability of residential behaviour given the mean
weekly probability of residential behaviour at time t-1 ranked by increasing order of the biascorrected Akaike Information Criterion (AIC c). Differences in AIC c relative to the top model
(ΔAICc), weighted AIC c (wAICc), cumulative weighted AIC c (cwAIC c), log-likelihood (log(L)),
and the number of parameters in a model (K) are shown. The top four models were selected for
averaged predictions. ………………………………………………………………………… 33
3.1 Models predicting the probability of residential behaviour given the behaviour state at time t1 ranked by increasing order of the bias-corrected Akaike Information Criterion (AIC c).
Differences in AIC c relative to the top model (ΔAIC c), weighted AIC c (wAICc), cumulative
weighted AIC c (cwAIC c), log-likelihood (log(L)), and the number of parameters in a model (K)
are shown. The top six models were selected for averaged predictions. ……………………… 88
3.2 Models predicting maximum displacement ranked by increasing order of the bias-corrected
Akaike Information Criterion (AIC c). Differences in AIC c relative to the top model (ΔAIC c),
weighted AIC c (wAICc), cumulative weighted AIC c (cwAIC c), log-likelihood (log(L)), and the
number of parameters in a model (K) are shown. The top 13 models were selected for averaged
predictions. …………………………………………………………………………………….. 89

vi

LIST OF FIGURES
2.1 Map of study area below Hugh L. Keenleyside Dam (HLK) and Arrow Lakes Generating
Station (ALH) (black dots), British Columbia, Canada. Habitat zones are highlighted by colour,
acoustic receivers are represented by white circles, and anchored sentinel tags are represented by
yellow circles. …………………………………………………………………………………. 34
2.2 State-dependent distributions for step length (meters) and turning angle (radians). ……… 35
2.3 Individual bouts from two white sturgeon in the study area with behaviour state estimated by
the hidden Markov model represented by colour. Ovals represent the error ellipse associated with
a position. ……………………………………………………………………………………… 36
2.4 Proportion of time white sturgeon spent having an upper (>75%), middle (25-75%), or lower
(<25%) probability of being in residential behaviour. Results are separated by season and habitat
zone. …………………………………………………………………………………………… 37
2.5 Proportion of time individual white sturgeon spent having an upper (>75%), middle (2575%), or lower (<25%) probability of being in residential behaviour. Results are separated by
season and empty columns indicate a fish was not detected in a season. ……………………… 38
2.6 White sturgeon positions having greater than 75% probability of being in a residential
behaviour state. Positions are shown by season with black lines distinguishing habitat zones. .. 39
2.7 White sturgeon positions having greater than 75% probability of being in a transitory
behaviour state. Positions are shown by season with black lines distinguishing habitat zones. .. 40
2.8 Model averaged coefficients and standard errors (SE) of the beta generalized linear mixed
model parameters predicting the probability of residential behaviour given the probability of
residential behaviour at time t-1. Model averaged intercept estimate of -3.76 (± 0.06 SE). …... 41
2.9 Model averaged predictions of the probability of being in the residential behaviour state by
season, habitat zone, and probability of residential behaviour at time t-1. Whiskers denote 95%
confidence interval. Categorical covariates sex and photoperiod were set to Female and Day
respectively. ……………………………………………………………………………………. 42
2.10 Model averaged predictions of the probability of being in the residential behaviour state by
photoperiod and probability of residential behaviour at time t-1. Whiskers denote 95%
confidence interval. Categorical covariates sex, season, and habitat zone were set to Female,
Summer, and HLK Eddy respectively.………………………………………………………… 43

vii

LIST OF FIGURES CONTINUED
2.11 Model averaged predictions of the probability of being in the residential behaviour state by
sex and probability of residential behaviour at time t-1. Whiskers denote 95% confidence
interval. Categorical covariates photoperiod, season, and habitat zone were set to Day, Summer,
and HLK Eddy respectively. …………………………………………………………………… 44
2.12 Model averaged coefficients and standard errors (SE) of the beta generalized linear mixed
model parameters predicting the mean weekly probability of residential behaviour given the
mean weekly probability of residential behaviour at time t-1. Model averaged intercept estimate
of -3.30 (± 0.07 SE). …………………………………………………………………………… 45
2.13 Model averaged predictions of the mean weekly probability of being in the residential
behaviour state by habitat zone and mean weekly probability of residential behaviour at time t-1.
Whiskers denote 95% confidence interval. To isolate effects, other numeric model covariates
were fixed at their mean and the categorical covariate sex was set to Female. ………………... 46
2.14 Model averaged predictions of the mean weekly probability of being in the residential
behaviour state by water temperature and mean weekly probability of residential behaviour at
time t-1. Shaded areas denote 95% confidence interval. To isolate effects, other numeric model
covariates were fixed at their mean and the categorical covariates sex and habitat zone were set
to Female and HLK Eddy respectively. …...…………………………………………………… 47
2.15 Model averaged predictions of the mean weekly probability of being in the residential
behaviour state by ALH discharge and mean weekly probability of residential behaviour at time
t-1. Shaded areas denote 95% confidence interval. To isolate effects, other numeric model
covariates were fixed at their mean and the categorical covariates sex and habitat zone were set
to Female and HLK Eddy respectively.………………………………………………………… 48

2.16 Model averaged predictions of the mean weekly probability of being in the residential
behaviour state by HLK discharge and mean weekly probability of residential behaviour at time
t-1. Shaded areas denote 95% confidence interval. To isolate effects, other numeric model
covariates were fixed at their mean and the categorical covariates sex and habitat zone were set
to Female and HLK Eddy respectively. ……..………………………………………………… 49
2.17 Model averaged predictions of the mean weekly probability of being in the residential
behaviour state by sex and mean weekly probability of residential behaviour at time t-1.
Whiskers denote 95% confidence interval. To isolate effects, other numeric model covariates
were fixed at their mean and the categorical covariate habitat zone was set to HLK Eddy. ….. 50

viii

LIST OF FIGURES CONTINUED
2.18 Water temperature in the study area from December 2013 – November 2014. Blue line
represents LOESS smoothed data and grey points represent raw data. ………………………... 51
2.19 Discharge from Hugh L. Keenleyside Dam (HLK) and Arrow Lakes Generating Station
(ALH) from December 2013 – November 2014. Total discharge is the combination of HLK and
ALH. Blue lines represent LOESS smoothed data and grey points represent raw data. ……… 52
2.20 Modeled velocity conditions during peak summer discharge below Hugh L. Keenleyside
Dam (HLK) and Arrow Lakes Generating Station (ALH), British Columbia, Canada. Velocity
and direction of discharge represented by coloured arrows. Figure pulled from West et al. (2020)
unpublished data. ……………………………………………………………………………… 53
2.21 Modeled velocity conditions during low fall discharge below Hugh L. Keenleyside Dam
(HLK) and Arrow Lakes Generating Station (ALH), British Columbia, Canada. Velocity and
direction of discharge represented by coloured arrows. Figure pulled from West et al. (2020)
unpublished data. ………………………………………………………………………………. 54
3.1 Transboundary Reach of the Upper Columbia River from HLK in British Columbia, Canada,
through Gifford in Washington, USA. River kilometers of interest are labeled as whit e dots. .. 90
3.2 Canadian section of the Transboundary Reach in British Columbia, Canada, (left) and United
States section of the Transboundary Reach in Washington, USA (right). River zones are
distinguished by grey and white sections of river. Dashed lines represent the extent of the six
environmental monitoring zones. White dots represent aggregated acoustic receiver locations
(some locations may contain more than one receiver). Yellow dots represent capture locations of
tagged fish. Shaded blue circles represent known white sturgeon spawning areas. …………… 91
3.3 Detection efficiency (DE) from May 2019 – August 2020. A) Mean daily detection
efficiency by season and year. B) Mean daily detection efficiency by river zone. Solid blue lines
represent river zone mean and dashed blue lines represent river zone median for the study period.
…………………………………………………………………………………………………... 92
3.4 Mean daily water temperature (A) and discharge (B) from May 2019 – August 2020. Line
type represents gauges at Hugh L. Keenleyside Dam (rkm 1261.2) and Border (rkm 1204.2).
…………………………………………………………………………………………………... 93

ix

LIST OF FIGURES CONTINUED
3.5 Tagged hatchery-origin white sturgeon used in study. A) Fork length and weight of tagged
fish, with shades of gray describing sex, shape describing country of capture, and line type
representing expected values based on country of capture (relative condition of 0). B) Number of
fish tagged by age and sex. C) Number of fish tagged by habitat zone of capture and sex. D)
Boxplot of fish tagged by age and habitat zone of capture. Dashed line represents the border
between countries. ……………………………………………………………………………… 94
3.6 Distribution of tagged hatchery-origin white sturgeon by month and river zone between May
2019 – August 2020. Fish were counted for the river zone in which they spent the longest
duration of time in a month. Last box represents the number of fish not detected during a month.
………………………………………………………………………………………………….. 95
3.7 Daily number of tagged hatchery-origin white sturgeon in each river zone between May 2019
– August 2020. Dashed line represents the border between countries. ………………………... 96

3.8 Presence of tagged hatchery-origin white sturgeon between May 2019 – August 2020.
Presence indicated by black lines with instances of movement across zones highlighted by
points. Point colour represents river zone fish are crossing into and shape represents country
where the river zone is located. Crosses represent time and river zone of fish release after
tagging. ………………………………………………………………………………………… 97
3.9 Model averaged coefficients and standard errors (SE) of the binomial generalized linear
mixed model parameters predicting the probability of residential behaviour given the behaviour
state at time t-1. Model averaged intercept estimate of -0.50 (± 0.34 SE). ……………………. 98
3.10 Model averaged predictions of the probability of being in the residential behaviour state by
river zone, discharge, and the behaviour state at time t-1. Shaded areas denote 95% confidence
interval. To isolate the effect, other numeric model covariates were fixed at their mean and the
categorical covariate sex was set to Female. …………………………………………………... 99
3.11 Model averaged predictions of the effect of water temperature (A), sex (B), age (C), and
relative condition (D) on the probability of being in the residential behaviour state. Behaviour
state at time t-1 represented by colour with 95% confidence interval (shaded area and whiskers).
To isolate the effect, other numeric model covariates were fixed at their mean and the categorical
covariates sex and river zone were set to Female and Zone 1 respectively. ………………….. 100
3.12 Maximum displacement (rkm) of tagged hatchery-origin white sturgeon by season and
country where movement took place between May 2019 – August 2020. …………………… 101

x

LIST OF FIGURES CONTINUED
3.13 Model averaged coefficients and standard errors (SE) of the gamma hurdle generalized
linear mixed model parameters predicting maximum displacement. Results are separated by
binomial and gamma components of the gamma hurdle model. Model averaged intercept
estimates are -0.75 (± 0.52 SE) and 1.58 (± 0.42 SE) respectively. ………………………….. 102

3.14 Model averaged predictions of the effect of country (A) and discharge and country (B) on
maximum displacement (rkm) with 95% confidence interval (shaded area and whiskers). To
isolate effects, other model covariates were fixed at their mean. Categorical covariates of sex and
season were set to Female and Summer respectively. ……………………………………….. 103

3.15 Model averaged predictions of the effect of water temperature on maximum displacement
(rkm) with 95% confidence interval (shaded area). To isolate effects, other model covariates
were fixed at their mean. Categorical covariates of sex, season, and country were set to Female,
Summer, and USA respectively. ……………………………………………………………… 104

3.16 Model averaged predictions of the effect of season (A), sex (B), age (C), and relative
condition (D) on maximum displacement (rkm) with 95% confidence interval (shaded area and
whiskers). To isolate effects, other model covariates were fixed at their mean. Categorical
covariates of sex, season, and country were set to Female, Summer, and USA respectively. .. 105

xi

ACKNOWLEDGEMENTS

I would like to thank BC Hydro for funding this project and I am grateful for their
commitment to supporting white sturgeon conservation and recovery research. I extend huge
gratitude to my committee: Dr. Eduardo G. Martins, Dr. James Crossman, and Dr. Chris
Johnson. Your guidance and advice has made this project possible and greatly improved my
skills as a scientist. I’d like to thank my fellow members of the Freshwater Fish Ecology Lab at
the University of Northern British Columbia. The support and community of other fish-minded
students was invaluable. This work would have also not been possible without the contributions
of Marco Marrello, Dean den Biesen, and Mike Keehn. I am grateful for the opportunity to have
worked with you all in the field and you made my experience on the Columbia River memorable.
I am also grateful to Jason McLellan, Andy Miller, and the Colville Confederated Tribes and
Spokane Tribal Fisheries. This project would not be possible without their contributions and
work on Columbia River white sturgeon in the United States. Lastly, I would like to thank my
parents for their continuous support and my friends for always making me laugh through this
whole process.

1

CHAPTER ONE

INTRODUCTION

White sturgeon (Acipenser transmontanus) are the largest and longest lived freshwater
species in North America (Scott & Crossman 1973). They occupy large river systems and have
been identified in the Sacramento, San Joaquin, Columbia, and Fraser river basins (Hildebrand et
al. 2016). The unique life history characteristics of white sturgeon have allowed the species to
persist over millions of years through major geologic and climatic events. They are slow growing and reach sexual maturity later in life, spawn intermittently with high fecundity, and can
migrate great distances to take advantage of favorable feeding and spawning conditions
(Hildebrand et al. 2016; Haxton & Cano 2016). While these life history traits allowed white
sturgeon to adapt to environmental stochasticity over thousands of years, these same traits now
make the species highly susceptible to anthropogenic impacts which have occurred within a
single generation (Bemis et al. 1997). Over the last century, historical overharvesting, alterations
in habitat quality and connectivity from damming and river regulation, and other industrial uses
of rivers have led to the decline of white sturgeon across much of their range (Hildebrand et al.
2016). White sturgeon, as well as many other sturgeon and paddlefish species in the order of
Acipenseriformes, are now among the most threatened species globally (Cooke et al. 2012).
In the Upper Columbia River (UCR), the construction of dams for river regulation and
power generation has resulted in the reduction and fragmentation of white sturgeon populations.
White sturgeon in the Transboundary Reach of the UCR, located between the Grand Coulee
Dam (1941) in Washington and Hugh L. Keenleyside Dam (1968) in British Columbia, have

2

demonstrated population declines and demographic shifts since post-dam assessments began in
the 1980s (Hildebrand et al. 1999). A population once dominated by juveniles to one presently
dominated by adults has been attributed to approximately 50 years of low recruitment (i.e.
survival of juveniles into adulthood). In 2006, the population was listed as Endangered under the
Canadian Species at Risk Act (Fisheries and Oceans 2014). Dams can prevent access to critical
habitats required to complete life processes (e.g. spawning or feeding), impact water quality (e.g.
temperature), and modify natural flow regimes (Cooke et al. 2016). Due to a combination of
river regulation and climate change, the Transboundary Reach currently experiences spring and
summer water temperatures 2 to 3 °C warmer than historical averages (McAdam 2001) and
significant reductions in flooding and spring freshet flows (Hildebrand & Parsley 2013).
Changes in water temperature and flow have been thought to influence the biological and
ecological processes of white sturgeon, such as metabolic rates (Volkoff et al. 2020), migration
cues and habitat suitability related to spawning (Hildebrand & Parsley 2013; Beamesderfer &
Farr 1997), and resource demands and distributions related to feeding (Taylor et al. 2012;
Turschwell et al. 2019). Better understanding how white sturgeon behave in response to
changing conditions in this regulated and transboundary river reach could assist in species
recovery through the development of improved mitigative measures related to river regulation
and hydroelectric operations.
The study of animal movement has become a critical component of fisheries management
and recovery (Crossin et al. 2017; Allen et al. 2016). Movement is fundamental to animal
ecology, influencing growth, survival, and reproductive success (Shaw 2020). Advancements in
acoustic telemetry over the past few decades, such as tag miniaturization, improved battery life,
and software development, now allow for greater tracking of fish movement at different spatial

3

and temporal scales (Hussey et al. 2015). Additionally, the development of advanced analytical
techniques (Joo et al. 2019; Whoriskey et al. 2019) has improved the ability of researchers to
answer fundamental questions in movement ecology, such as ‘why move?’, ‘how to move?’,
‘when to move?’, and ‘where to move?’ (Mueller et al. 2008; Nathan et al. 2008). Movement
behaviour analyses, for example, deconstruct animal movement into distinct underlying
behaviour states (e.g. assumed to represent resting, foraging, spawning) and offer researchers a
new way of understanding how biological and environmental factors influence the movements of
individuals and populations (Patterson et al. 2017). Such analyses have been utilized in fine-scale
and coarse-scale acoustic telemetry studies of sturgeon (McLean et al. 2014; Dorazio & Price
2018) as well as other fish species (Bacheler et al. 2019; Cote et al. 2020) to identify interactions
and responses to human activities and environmental conditions. Applying similar methods to
threatened populations, such as white sturgeon affected by river regulation and hydroelectric
operations, could provide important insight into the management and recovery of a species.
The goal of this study was to describe and quantify white sturgeon movement behaviour
in relation to river regulation. To accomplish this, acoustic telemetry was utilized to collect data
on white sturgeon movement at two different spatial scales. Chapter 2 used an overlapping
acoustic telemetry array covering 2 river kilometers (rkm) to track the fine-scale movements of
wild adult white sturgeon occupying critical habitat immediately downstream of a hydroelectric
operation. In Chapter 3, a large acoustic telemetry array covering around 300 rkm was used to
track the coarse-scale movements of hatchery-origin white sturgeon residing throughout the
Transboundary Reach of the UCR. Hidden Markov models (HMMs) and generalized linear
mixed models (GLMMs) were then applied to my fine-scale and coarse-scale movement datasets
to identify ecologically meaningful movement behaviours and determine how movement

4

behaviour is influenced by biological factors and environmental conditions (e.g. water
temperature and discharge) that are frequently altered by river regulation. The results from this
study elucidate how white sturgeon interact with dams and river regulation and will inform
recovery measures and management decisions for the threatened population in the UCR.

5

References
Allen AM, Singh NJ (2016) Linking movement ecology with wildlife management and
conservation. Frontiers Ecol Evol 3:155. https://doi.org/10.3389/fevo.2015.00155
Bacheler NM, Michelot T, Cheshire RT, Shertzer KW (2019) Fine-scale movement patterns and
behavioural states of gray triggerfish Balistes capriscus determined from acoustic telemetry and
hidden Markov models. Fish Res 215:76-89. https://doi.org/10.1016/j.fishres.2019.02.014
Beamesderfer RCP, Farr RA (1997) Alternatives for the protection and restoration of sturgeons
and their habitat. In: Birstein VJ, Waldman JR, Bemis WE (ed) Sturgeon Biodiversity and
Conservation. Kluwer Academic Publishers, Dordrecht, pp 407-417
Bemis WE, Findeis EK, Grande L (1997) An overview of Acipenseriformes. Environ Biol Fish
48:25-71. https://doi.org/10.1023/A:1007370213924
Cooke S, Paukert C, Hogan Z (2012) Endangered river fish: factors hindering conservation and
restoration. Endanger Species Res 17:179-191. https://doi.org/10.3354/esr00426
Cooke SJ, Martins EG, Struthers DP, Gutowsky LF, Power M, Doka SE, Dettmers JM, Crook
DA, Lucas MC, Holbrook CM, Krueger CC (2016) A moving target—incorporating knowledge
of the spatial ecology of fish into the assessment and management of freshwater fish populations.
Environ Monit Assess 188:239. https://doi.org/10.1007/s10661-016-5228-0
Cote D, Morris CJ, Regular PM, Piersiak MG (2020) Effects of 2D seismic on snow crab
movement behaviour. Fish Res 230:105661. https://doi.org/10.1016/j.fishres.2020.105661
Crossin GT, Heupel MR, Holbrook CM, Hussey NE, Lowerre-Barbieri SK, Nguyen VM, Raby
GD, Cooke SJ (2017) Acoustic telemetry and fisheries management. Ecol Appl 27:1031-1049.
https://doi.org/10.1002/eap.1533
Dorazio RM, Price M (2019) State-space models to infer movements and behaviour of fish
detected in a spatial array of acoustic receivers. Can J Fish Aquat Sci 76:543-550.
https://doi.org/10.1139/cjfas-2018-0067
Fisheries and Oceans Canada (2014) Recovery strategy for white sturgeon (Acipenser
transmontanus) in Canada. In Species at Risk Act Recovery Strategy Series. Fisheries and
Oceans Canada, Ottawa, ON.
Haxton T, Cano T (2016) A global perspective of fragmentation on a declining taxon—the
sturgeon (Acipenseriformes). Endanger Species Res 31:203-210.
https://doi.org/10.3354/esr00767
Hildebrand L, McLeod C, McKenzie S (1999) Status and management of White Sturgeon in the
Columbia River in British Columbia, Canada: An Overview. J Appl Ichthyol 15:164-172.
https://doi.org/10.1111/j.1439-0426.1999.tb00227.x

6

Hildebrand LR, Parsley M (2013) Upper Columbia White Sturgeon Recovery Plan - 2012
Revision. Prepared for the Upper Columbia White Sturgeon Recovery Initiative.
Hildebrand LR, Schreier AD, Lepla K, McAdam SO, McLellan J, Parsley MJ, Paragamian VL,
Young SP (2016) Status of white sturgeon (Acipenser transmontanus Richardson 1863)
throughout the species range, threats to survival, and prognosis for the future. J Appl Ichthyol
32:261-312. https://doi.org/10.1111/jai.13243
Hussey NE, Kessel ST, Aarestrup K, Cooke SJ, Cowley PD, Fisk AT, Harcourt RG, Holland
KN, Iverson SJ, Kocik JF, Flemming JEM, Whoriskey FG (2015) Aquatic animal telemetry: a
panoramic window into the underwater world. Science 348:1255642.
https://doi.org/10.1126/science.1255642
Joo R, Boone ME, Clay TA, Patrick SC, Clusella‐Trullas S, Basille M (2019) Navigating
through the r packages for movement. J Anim Ecol 89:248-267. https://doi.org/10.1111/13652656.13116
McAdam S (2001) Summary of historic and contemporary water temperatures for the Columbia
River and the potential effects of impoundment. Fisheries Technical Circular No 103. BC
Ministry of Fisheries, Victoria, BC.
McLean M, Simpfendorfer C, Heupel M, Dadswell M, Stokesbury M (2014). Diversity of
behavioural patterns displayed by a summer feeding aggregation of Atlantic sturgeon in the
intertidal region of Minas Basin, Bay of Fundy, Canada. Mar Ecol Prog Ser 496:59-69.
https://doi.org/10.3354/meps10555
Mueller T, Fagan WF (2008) Search and navigation in dynamic environments – from individual
behaviours to population distributions. Oikos 117:654-664. https://doi.org/10.1111/j.00301299.2008.16291.x
Nathan R, Getz WM, Revilla E, Holyoak M, Kadmon R, Saltz D, Smouse PE (2008) A
movement ecology paradigm for unifying organismal movement research. Proc Natl Acad Sci
USA 105:19052-9. https://doi.org/10.1073/pnas.0800375105
Patterson TA, Parton A, Langrock R, Blackwell PG, Thomas L, King R (2017) Statistical
modelling of individual animal movement: an overview of key methods and a discussion of
practical challenges. AStA Adv Stat Anal 101:399-438. https://doi.org/10.1007/s10182-0170302-7
Scott WB, Crossman EJ (1973) Freshwater fishes of Canada, Bulletin 184. Fisheries Research
Board of Canada, Department of Fisheries and Oceans, Scientific Information and Publications
Branch, Ottawa, ON.
Shaw AK (2020) Causes and consequences of individual variation in animal movement. Mov
Ecol 8:12. https://doi.org/10.1186/s40462-020-0197-x
Taylor MK, Cooke SJ (2012) Meta-analyses of the effects of river flow on fish movement and
activity. Environ Rev 20:211-219. https://doi.org/10.1139/a2012-009

7

Turschwell MP, Stewart‐Koster B, King AJ, Pusey B, Crook D, Boone E, Douglas M, Allsop Q,
Jackson S, Kennard MJ (2019) Flow‐mediated predator-prey dynamics influence fish
populations in a tropical river. Freshwater Biol 64:1453-1466. https://doi.org/10.1111/fwb.13318
Volkoff H, Rønnestad I (2020) Effects of temperature on feeding and digestive processes in fish.
Temp 7:1-14. https://doi.org/10.1080/23328940.2020.1765950
Whoriskey K, Martins EG, Auger‐Méthé M, Gutowsky LFG, Lennox RJ, Cooke SJ, Power M,
Flemming JM (2019) Current and emerging statistical techniques for aquatic telemetry data: a
guide to analyzing spatially discrete animal detections. Methods Ecol Evol 10:935-948.
https://doi.org/10.1111/2041-210X.13188

8

CHAPTER TWO

FINE-SCALE MOVEMENT ECOLOGY AND BEHAVIOUR PATTERNS OF WHITE
STURGEON ON THE REGULATED UPPER COLUMBIA RIVER

9

Introduction
Movement ecology has become a rapidly expanding discipline in recent decades, with
large strides towards investigating the relationship between movement patterns and behaviour.
This growth of the discipline can be attributed to significant advancements in animal tracking
technology (Hussey et al. 2015; Lennox et al. 2017), analytical tools (Joo et al. 2019; Whoriskey
et al. 2019), and movement theory (Mueller et al. 2008; Nathan et al. 2008). With a majority of
studies in the field being empirical and focused on different components of movement ecology
(Holyoak et al. 2008), there has been a push to implement a unifying conceptual framework
which compels researchers to evaluate movement in a more comprehensive and broader context.
Such frameworks seek to integrate the basic questions of movement ecology: ‘why move?’, ‘how
to move?’, ‘when to move?’, and ‘where to move?’ (Mueller et al. 2008; Nathan et al. 2008).
Novel analytical techniques have begun to allow for the integration of these questions in
movement ecology, with a focus on investigating the intrinsic (e.g. biological) and extrinsic (e.g.
environmental) factors that influence how animals move across space and time (Patterson et al.
2017). Methods increasingly applied are those popularized by Morales et al. (2004) and Jonsen et
al. (2005), which deconstruct animal movement track data into underlying behaviours. These
methods center on the idea that animal movement is driven by switches in behaviour, and these
switches can be influenced by biological and environmental variables (Nathan et al. 2008).
Assessing movement in terms of behaviour (e.g. resting, foraging, migration) and the factors that
influence switches in behaviour can help conceptualize and infer what drives growth, survival,
and reproductive success of an animal (Shaw 2020). Such inferences are important when it
comes to recovery and conservation of threatened species, particularly for fish populations that
reside in regulated rivers.

10

Riverine fishes are some of the most threatened species globally (Ricciardi et al. 1999;
Darwall et al. 2016; Cooke et al. 2012). This is in part a result of anthropogenic impacts that
include habitat fragmentation from the construction of dams for hydroelectric generation and
river regulation (Dudgeon et al. 2006; Nilsson et al. 2005; Grill et al. 2015, 2019). Dams have
modified natural flow regimes, impacted water quality (e.g. temperature), and can prevent access
to critical habitats required to complete life processes (e.g. spawning or feeding) (Cooke et al.
2016). These changes can impact fish movement and behaviour. Water temperature, as a main
driver of fish metabolism and physiology, can influence movement by altering metabolic
demands for resource (e.g. prey) consumption (Volkoff et al. 2020). Flow influences movement
by altering the energetic demands of fish to maintain position and move (Shepard et al. 2013).
Flow can also change the resource abundance and distribution within a habitat (Taylor et al.
2012; Turschwell et al. 2019). Switches in foraging behaviour in response to changing prey
availability has been observed in fish. Multiple species appear to adopt low-energy sit-and-wait
strategies when prey is abundant and high-energy search strategies when prey is scarce (Killen et
al. 2007; Higginson et al. 2015; Grant et al. 1987; Fausch et al. 1997; O’Brien et al. 1989).
Threatened fish adjusting their movements to changing conditions, such as those caused by river
regulation, could have serious implications for a species (Tuomainen et al. 2011). This is because
fish must manage finite energy budgets, and there will be trade-offs between expending energy
on movement or other physiological functions such as growth and reproduction (Rennie et al.
2005).
In this study, I examined the movement behaviours of white sturgeon residing below a
dam on the regulated Upper Columbia River (UCR). White sturgeon are a species at risk almost
everywhere they occur (Hildebrand et al. 2016), with a majority of their historical range being

11

altered by river regulation and fragmented by dams (Haxton & Cano 2016; Dynesius et al. 1994).
In the UCR, white sturgeon were listed as Endangered under the Canadian Species at Risk Act
after continuous recruitment failure and population declines over the last century (Fisheries and
Oceans 2014). The population has critical habitat associated with feeding, overwintering, and
spawning located below, or influenced by, multiple hydroelectric operations (Hildebrand et al.
1999). This is not uncommon for sturgeon (Haxton & Cano 2016) and has been observed in
other populations (Gillespie et al. 2020; Cooke et al. 2012). The release of fast-flowing waters
over coarse substrate can create habitat typically selected by sturgeon for spawning, and these
fast-flowing waters can also form adjacent eddies which create both productive feeding and
overwintering areas (Hildebrand et al. 1999; Hildebrand et al. 2016). This occupance of white
sturgeon in critical habitat below hydroelectric facilities makes understanding their interactions
with operations and river regulation imperative.
The objective of this study was to describe and quantify fine-scale movement behaviour
of white sturgeon occupying critical habitat immediately downstream of a dam on the UCR in
relation to hydroelectric operations and river regulation. An overlapping array of acoustic
telemetry receivers was used to collect fine-scale positions of acoustically tagged wild adult
white sturgeon. I then applied hidden Markov models (HMMs) and generalized linear mixed
models (GLMMs) to investigate the relationship between white sturgeon movement behaviour
and its possible driving factors. This was accomplished by 1) identifying ecologically
meaningful movement behaviours within the fine-scale dataset and 2) assessing the influence of
biological (e.g. sex) and environmental (e.g. water temperature, discharge, season, photoperiod)
factors on movement behaviour. The results reveal how white sturgeon interact with dams and

12

will inform recovery measures, such as discharge management and optimization of operations to
reduce the possibility of negative impacts to white sturgeon.

Methods
Study area
This study was conducted in the Transboundary Reach of the UCR, an impounded
section of river between the Hugh L. Keenleyside Dam in British Columbia and the Grand
Coulee Dam in Washington. Movements were tracked in the 2 rkm section immediately
downstream of the Hugh L. Keenleyside Dam (HLK) and Arrow Lakes Generating Station
(ALH). HLK is a storage-only facility (i.e. no power generation) and was constructed in 1968
under the Columbia River Treaty. ALH, located adjacent to HLK, is a power-generation facility
constructed in 2001 and operated by Columbia Power Corporation. Water supply for both
facilities comes from the upstream Arrow Lakes Reservoir. Combined HLK and ALH discharge
is coordinated weekly with the United States under Columbia River Treaty agreements (BC
Hydro 2021). While discharge volume is held consistent through the week, hourly or daily
adjustments are sometimes required for maintenance activities or other demands around power
generation. Fluctuations in discharge marginally alter water surface elevation and do not affect
the habitat availability of white sturgeon anywhere in the study area.
The study area was divided into four habitat zones based on river structure and
environmental parameters: HLK Eddy, HLK Tailrace, ALH Tailrace, and Downstream (Figure
2.1). HLK Eddy is a large circulating eddy caused by the inflows from both tailraces. High use of
the HLK Eddy for feeding and overwintering has been identified for white sturgeon (Fisheries
and Oceans 2014; BC Hydro 2016). The HLK Tailrace consists of inflows from HLK and is a

13

deep pool that shallows where it joins the ALH Tailrace. Use of the HLK Tailrace for feeding
has been observed seasonally. The ALH Tailrace consists of inf lows from ALH and is a shallow
section of the study area. Prior research has identified spawning in the ALH Tailrace
(Terraquatic Resources 2011). Lastly, Downstream consists of more riverine (i.e. channelized)
habitat from combined HLK and ALH inflows. It is a transitional area in which fish move in and
out of the study area.

Fish sampling and tagging
Wild adult white sturgeon were captured and tagged with acoustic tags in the Canadian
portion of the Transboundary Reach (rkm 1261 – 1204) between the years of 2007 and 2013 for
projects outside of this study related to movement and habitat use research and broodstock
collection for conservation aquaculture. Fish capture abided by capture protocols and set line
configurations commonly used for white sturgeon throughout the Columbia River (Golder
Associates 2006; Nigro et al. 1988). Set lines were deployed at both random and preselected
sampling locations determined by the physical parameters (e.g. depth and velocity) of the site in
relation to efficiency and safety of fish capture (BC Hydro 2015).
A total of 132 wild adult white sturgeon captured over this period were implanted with
VEMCO V16-4x acoustic tags (69 kHz, 120-180 sec ping rate). Surgical procedures were used
to tag and determine the sex of fish. A 1.5 – 2.0 cm long incision was made through the ventral
body wall and adjacent to the sturgeon’s mid-line. Using an otoscope, the gonad tissue was
visually assessed for sex (Webb et al. 2019). Afterwards, fish were implanted with a uniquelycoded VEMCO V16-4x acoustic tag through the surgical incision. Incisions were then closed
using half circle CP-2 reverse cutting-edge needles wedged to 2-0 polydioxanone violet

14

monofilament sutures (Ethicon, Z969H). Fish selected for broodstock (n = 106) experienced
transportation and holding at the Kootenay Sturgeon Hatchery in Wardner, BC for induced
spawning prior to being acoustically tagged (FFSBC 2014). After processing, sturgeon were
returned to the water near their location of capture (BC Hydro 2015).

Acoustic array
The fine-scale movement dataset of tagged white sturgeon was acquired from a VEMCO
Positioning System (VPS) study conducted between December 1, 2013 and November 15, 2014.
VPS studies consist of an array of fixed submersible acoustic receivers that record a timestamp
of the arrival of transmissions from acoustic tags. The collected time-of-arrival data of at least
three separate receivers are then processed using a hyperbolic positioning technique to achieve
fine-scale tracking of acoustically tagged fish (Smith 2013). This technique has been widely used
to investigate the movement of species within marine and freshwater environments (Espinoza et
al. 2011a,b; Farrugia et al. 2011; Coates et al. 2013; McLean et al. 2014; Bacheler et al. 2019;
Cote et al. 2020). The VPS study consisted of an array of 29 acoustic receivers (VEMCO
VR2W) in the 2 rkm study area (Figure 2.1). The number of receivers deployed in the VPS array
did fluctuate through time (range: 18 – 29). Fluctuations were due to dislocation of receivers by
seasonal discharge or entanglement with commercial tugboat operations. Maintaining these types
of arrays in fast-flowing and working rivers can be challenging (Bergé et al. 2012). To account
for receiver clock drift, horizontal position error (HPE), and detection efficiency over time, each
receiver was equipped with a synchronization tag and four separately anchored sentinel tags
were placed within the array (Golder Associates 2015).

15

Data filtering
Fine-scale acoustic telemetry can result in a significant amount of location data that are
not accurate and requires filtering before analysis. Factors such as the spatial configuration of
acoustic receivers (e.g. density) and surrounding environment (e.g. water velocity) can affect
both detection efficiency and positional error in an array (Bergé et al. 2012; Roy et al. 2014).
Prior analysis of the VPS found that the number of receivers in the array did influence overall
detection efficiency and the number of tags positioned, resulting in fewer positions being
resolved with a reduction in array size (Golder Associates 2015). For positional error,
assessments were performed using HPE. HPE is a relative and unitless estimate of how sensitive
a calculated position is to error. This estimate is derived by applying known measurement error
(e.g. synchronization and sentinel tags) to the sensitivity of the system (Smith 2013). Higher
HPE values represent less accurate positioning than lower HPE values. Prior analysis of the VPS
determined an HPE value of 16 to be the optimal threshold for positional accuracy without
reducing too much of the dataset via filtering (Golder Associates 2015). This value lies within
the typical 10 to 20 HPE threshold range of other VPS studies (Roy et al. 2014). Positions with
HPE >16 were considered inaccurate and excluded from the final dataset I used in the analysis.

Movement behaviour
I used hidden Markov models (HMMs; R package momentuHMM) to conduct the
analyses of movement behaviour (McClintock et al. 2018). HMMs are statistical models which
describe sequences of observable events that depend on unobservable or “hidden” discrete states
(McClintock et al. 2020). HMMs use the probability distribution of movement metrics,
computed from sequential animal locations, to decompose movement tracks into distinct hidden

16

states (assumed to represent different behaviour states of an animal) (Jonsen et al. 2005). For this
study, I used step length (meters; Gamma distribution with zero-inflation) and turning angle
([−π, π]; Von Mises distribution) as the movement metrics for a two-state HMM. Selecting the
appropriate number of states for a HMM can be challenging (Pohle et al. 2017). Based on the
movement tendencies described for white sturgeon as a species and in the Transboundary Reach
(Hildebrand et al. 2016; BC Hydro 2018a), a two-state model was determined to be the best fit
for the dataset. The two behaviour states identified by the HMM fit to the white sturgeon dataset
were characterized by 1) short step length with infrequent changes in direction (hereafter,
residential behaviour) and 2) longer step lengths with frequent changes in direction (hereafter,
transitory behaviour).
HMMs are well suited to analyze movement datasets that are 1) recorded at regular time
intervals and 2) recorded with negligible positional error (Patterson et al. 2017). Fish detections
in my study often occurred on an irregular time interval due to fluctuations in imperfect detection
efficiency. To regularize the data, fish movement tracks were first split into unique bouts if time
between consecutive positions was greater than 2 hours. Utilizing bouts in this way helps define
large gaps in detection as periods in which fish location cannot be determined or fish temporarily
left the array (Cote et al. 2020). Furthermore, bouts with less than 100 positions were removed to
avoid problems with model convergence (Bacheler et al. 2019). Remaining bouts were then
interpolated at a regular 15-minute time interval using the crawlWrap function in
momentuHMM, which implements a continuous-time correlated random walk model to predict
locations at the selected time interval (McClintock et al. 2018). Positional error associated with
intervals of HPE (e.g. (14,15]) was accounted for in the crawlWrap function by including an
error ellipse model. Parameters for the error ellipse model (log standard deviation, correlation)

17

were calculated using the argosDiag2Cov function in the R package crawl (Johnson et al. 2008)
from positional error data acquired by the study’s sentinel tags (i.e. difference between true and
estimated sentinel tag positions).
After movement interpolation, HMMs were fit in momentuHMM using multiple
imputation. Multiple imputation is a method which allows pooled parameter estimates to reflect
uncertainty from factors such as positional error (McClintock et al. 2018). Instead of a definite
behaviour state estimated for each observation, the probability of a behaviour state was
determined for each interpolated location by calculating the proportion of times the state was
estimated over 100 imputations. Before fitting any imputation, HMMs must be provided starting
parameters to estimate behaviour states. These starting parameters describe the distribution of
movement metrics in each state, which in my case would be represented by mean step length
with a zero mass parameter (i.e. proportion of step length values at 0) and turning angle
concentration (Michelot et al. 2016). Turning angle concentration is a positive number which
measures how concentrated turning angles are around their mean, with larger values representing
smaller variance (Michelot et al. 2019). Mean turning angle is usually 0, as is the case for my
dataset, and represents a persistence in direction of movement. Starting parameters for this study
were determined using the getPar function in momentuHMM. This function helps estimate the
best starting parameters for different behaviour states when fitting HMMs to a dataset.

Model predictions
Once the probability of a behaviour state was estimated for each position using hidden
Markov modelling, the influence of biological and environmental covariates on movement
behaviour was assessed. As some environmental covariates changed on a coarser timescale than

18

the 15-minute interval in my HMM (e.g. HLK and ALH discharge adjusted on a weekly basis), I
assessed movement behaviour using two methods: 1) the probability of fish expressing
residential behaviour given the probability of residential behaviour at time t-1 (i.e. 15-min
before), and 2) the mean weekly probability of residential behaviour given the mean weekly
probability of residential behaviour at time t-1 (i.e. 1 week before). Both methods applied
generalized linear mixed models (GLMMs) with a beta distribution and logit link function.
Template Model Builder (TMB, Kristensen et al. 2016) via the package glmmTMB (Brooks et
al. 2017) was used to perform all GLMMs. Analyses of movement behaviour in this study will
focus on reporting the probability of residential behaviour, however it is important to note that
the probability of transitory behaviour would be the complement of these results.
For GLMMs assessing the probability of residential behaviour given the probability at
time t-1, candidate models included combinations of one individual (sex) and seven
environmental (water temperature, combined discharge, ALH discharge, HLK discharge,
photoperiod, habitat zone, season) covariates. Discharge was assessed in two formats: a single
parameter representing combined ALH and HLK discharge, and two separate parameters each
representing ALH or HLK discharge. Due to correlation, combined discharge was not included
in any candidate models with ALH or HLK discharge, and season was not included in any
candidate models with water temperature or discharge covariates. A random intercept by bout
nested within individual was included in all models. For GLMMs assessing mean weekly
probability of residential behaviour given the mean weekly probability at time t-1, candidate
models included the same covariates with the exclusion of photoperiod. Mean weekly
probabilities were calculated for each fish and habitat zone in which a fish was positioned. Water

19

temperature and discharge covariates represented mean values for a given week. A random
intercept for individual was included in all models.
Temperature and discharge data were measured at 1-hour increments and values were
interpolated across 15-minute time intervals to match the locations estimated at a 15-minute
interval. Temperature data were acquired from BC Hydro’s Birchbank gauging station (rkm
1,290) and discharge data were obtained from the ALH and HLK facilities. A 6-hour lag effect
was applied to the discharge data to ensure that all flow changes had stabilized in the
downstream environment, given that flows were only changed on a weekly basis. Photoperiod
was categorized as day, night, or twilight. Sunrise and sunset times for each day of the study
period were obtained from the National Research Council Canada website (http://www.nrccnrc.gc.ca). Season was categorized as winter (December – February), spring (March – May),
summer (June – August), and fall (September – November).

Model selection
I used bias-corrected Akaike Information Criterion (AIC c) scores to rank and select
candidate models. AIC is useful in model selection in that it estimates the relative quality of a
model within the candidate set for explaining the given data (McElreath 2020), with the lowest
AIC representing the best or top model. Top models were compared against the null model
(intercept and state probability at time t-1 variable) for validation. Due to model selection
uncertainty (Burnham et al. 2004), models included in a set defined by a cumulative weight AIC c
residing just at or below 0.95 were used for model averaged predictions. Model averaging was
performed using the R package MuMIn (Bartoń et al. 2020), and all analyses in this study were
performed in R (R Core Team 2021).

20

Results
Data filtering
A total of 818,577 positions from 51 acoustically tagged white sturgeon (16,050 ± 14,777
positions per fish; range: 14 – 67,176) were determined from the VPS. HPE filtering resulted in a
total of 459,569 positions for the 51 fish (9,011 ± 8,223 positions per fish; range: 11 – 28,432).
Six fish did not meet the minimum requirements I identified for the HMM (bouts of at least 100
positions), resulting in a sample of 45 fish. The final dataset resulted in a total of 596 bouts, with
an average of 13 bouts per fish (range: 1 – 53) and five days per bout (range: 0.4 - 89.6 days).

Movement behaviour
Movement parameters of the residential and transitory behaviour states resulted in mean
step lengths of 1.9 (± 2.0) and 27.4 (± 28.5) meters respectively (Figure 2.2). Both behaviour
states had zero mass parameters close to 0, representing few instances of no movement between
consecutive positions. For turning angle, residential behaviour had an angle concentration of 0.8
and transitory behaviour had a concentration of 0.3. These concentrations describe residential
behaviour as having more directed movement than transitory behaviour. This is seen in turning
angle distributions, with residential behaviour having 46% of turning angles concentrated
between -0.5 and 0.5 radians and transitory behaviour being more homogeneously distributed
between -π and π radians (Figure 2.2). Residential behaviour was therefore characterized as
movements with short step lengths and more persistent direction, and transitory behaviour by
movements with longer step lengths and more frequent turns (Figure 2.3).
White sturgeon were detected for an average of 65 days (range: 1 – 193 days), or 19%
(range: 0.3% – 55.8%) of the study period. During that time, sturgeon spent 50% of it with

21

residential probabilities greater than 75% (likely residential) and 36% with residential
probabilities less than 25% (likely transitory). A total of 14% of time was spent with residential
probabilities between 25% and 75%, representing instances with less certainty of behaviour state.
Using these intervals of certainty, patterns in what behaviour state fish exhibited changed across
season and habitat zone (Figure 2.4). When broken down by season, time in residential
behaviour was highest in the winter (48%) and spring (68%), and lowest in the summer (24%)
and fall (12%). When broken down by habitat zone, the HLK eddy had the greatest proportion of
time in residential behaviour (58%), followed by the ALH tailrace (32%), Downstream (23%),
and HLK tailrace (15%). Seasonal trends outlined above were observed in all habitat zones
except the HLK tailrace, which had a larger proportion of time as residential during the fall
(36%) compared to other seasons. Among individual fish, variance for time spent in a behaviour
state was observed within seasons (Figure 2.5). Around 18% of fish had bouts in all four seasons.
When examined spatially, residential behaviour and movement during the winter and
spring was concentrated in large patches (>100 m diameter) of the ALH tailrace and HLK eddy
(Figure 2.6 and 2.7). In the summer and fall, however, movement was more dispersed, with
higher rates of transitory behaviour exhibited across all habitat zones and residential behaviour
concentrated in smaller patches of the ALH tailrace, HLK tailrace, and HLK eddy. Downstream
had the least fish presence and demonstrated seasonal shifts from residential behaviour in the
winter and spring to transitory behaviour in the summer and fall.

Model selection and predictions
The top model ranked by AIC c predicting the probability of fish expressing residential
behaviour given the probability at time t-1 (i.e. 15-min before) included photoperiod and a

22

season and habitat zone interaction as covariates (Table 2.1). One other candidate model
including sex as a covariate also closely explained the dataset and met criteria for inclusion in a
95% confidence set for the best model. Consequently, this candidate model in addition to the top
model was included in model averaged predictions. Overall, white sturgeon exhibited a strong
persistence in expressing a behaviour state. When a fish expressed a given state at time t-1 (e.g.
residential probability of 0% or 100%), it was very likely to remain in that state at time t. When
fluctuations did occur, it was when there was less certainty about residential behaviour in the
previous timestep (e.g. residential probability of 50%). The season and habitat zone interaction
was the most influential predictor in these instances (Figures 2.8 and 2.9). Fish in the HLK Eddy
demonstrated increased probabilities of residential behaviour in the winter and spring when
probabilities at time t-1 were 50%. In the Downstream habitat zone, fish demonstrated decreased
residential probabilities during the winter and increased residential probabilities in the spring.
For the ALH and HLK tailraces, season had less of an influence on residential behaviour when
probabilities at time t-1 were 50%. Photoperiod and sex were the least influential covariates.
There were minimal differences between photoperiods and sexes when probabilities at time t-1
were 50% (Figures 2.10 and 2.11).
The top model ranked by AIC c predicting the mean weekly probability of residential
behaviour given the probability at time t-1 (i.e. 1 week before) included the covariates of water
temperature, ALH discharge, and habitat zone (Table 2.2). Three other candidate models
including HLK discharge and sex as covariates also closely explained the dataset and met criteria
for inclusion in a 95% confidence set for the best model. Consequently, these three candidate
models in addition to the top model were included in model averaged predictions. Averaged
model predictions again demonstrated that white sturgeon were very likely to remain in the same

23

state at time t-1 that they expressed at time t. Fluctuations in mean weekly residential probability
did occur when probabilities at time t-1 were less certain (e.g. residential probability of 50%).
Habitat zone and water temperature were the most influential predictors in these instances
(Figure 2.12). Fish in the HLK Tailrace and Downstream habitat zones demonstrated decreased
mean weekly probabilities of residential behaviour when probabilities at t-1 were 50% (Figure
2.13), while the ALH Tailrace and HLK Eddy had less of an influence on mean weekly fish
behaviour. Warming water temperature resulted in decreased mean weekly probabilities of
residential behaviour when probabilities at t-1 were 50% (Figure 2.14). Discharge from the ALH
facility was the next strongest predictor. Increased flow from ALH resulted in increased mean
weekly residential probabilities given probabilities of 50% at time t-1 (Figure 2.15). The
covariates of sex and HLK discharge had the least influence on mean weekly residential
probability. There were minimal differences in mean weekly residential probability between
sexes and across HLK discharge when probabilities at time t-1 were 50% (Figures 2.16 and
2.17).

Discussion
Analyzing movement in terms of behaviour, compared to other conventional approaches,
enables researchers to identify emergent patterns in movement, evaluate relationships with
internal and external mechanisms, and conceptualize results in terms of ecologically meaningful
processes (Nathan et al 2008). I used fine-scale positioning of wild adult white sturgeon located
below a hydroelectric facility to identify distinct residential and transitory patterns in movement
behaviour, with the former representing short movements with more persistent direction and the
latter representing longer movements with more frequent turns. In my analyses, sturgeon

24

demonstrated strong persistence in a behaviour state (i.e. high probability of expressing the same
behaviour state at time t-1). Environmental covariates influenced behaviour only when there was
less certainty of a behaviour state in the previous timestep (i.e. behaviour probability of 50%). At
a 15-minute timescale, the interaction of season and habitat zone was the strongest predictor of
behaviour probability given probabilities at time t-1 were 50%. The HLK Eddy and Downstream
habitat zones were the most sensitive to change, with probabilities of residential behaviour
fluctuating differently in each habitat zone depending on season. At a coarser timescale, the
environmental covariates of habitat zone, water temperature, and discharge were strong
predictors of mean weekly behaviour probability given probabilities at time t-1 were 50%. Fish
located in the HLK Tailrace and Downstream habitat zones, as well as warming water
temperatures, related to decreased mean weekly probabilities of residential behaviour. For
discharge, increased flows from the ALH facility were related to increased mean weekly
probabilities of residential behaviour.
While analyses of movement behaviour can broaden our understanding of an animal’s
movement ecology, interpreting behaviour states in terms of ecological functions (e.g. resting,
foraging, migrating) can be challenging (McClintock et al. 2020). In fact, recognizing
uncertainty when interpreting behaviour states is critical for ecologists using HMMs. These
models simply decipher the strongest patterns in movement data regardless of the ecological
meaningfulness behind those patterns (Patterson et al. 2017). Identified patterns are also
inherently guided by the number of states and movement metrics specified by the researcher
(Buderman et al. 2021). To highlight this, Patterson et al (2017) offers the hypothetical example
of identifying resting, feeding, and travelling behaviour from a movement dataset. That threestate model could be unsuccessful, especially when there are subtle differences in movement

25

metric distributions. Patterson et al (2017) explains how the model may lump resting and feeding
into one behaviour state due to distance and turn rates being of similar magnitude, and then split
travelling into two behaviour states based on distances traveled. While the incorporation of more
states into a model may be tempting to remedy such an issue, often the temporal resolution of
data (e.g. 15-min time steps) limits our ability to appropriately capture subtle differences in
movement (Patterson et al. 2017). Additionally, incorporating a larger number of states can lead
to more difficulty distinguishing ecological meaningfulness between behaviours (Pohle et al.
2017). Understanding the challenges and limitations of HMMs should not d eter researchers from
their use. Rather, researchers should recognize the uncertainty in the data and modelling process,
and use their own intuition and knowledge when interpreting the ecological meaningfulness of
movement behaviours.
While the exact biological and ecological processes that dictate behaviour cannot be
known for certain, an understanding of a species’ life functions can offer relevant insight. In the
case of sturgeon, movement and habitat use are often driven by the biological functions of
feeding, overwintering, and spawning (Hildebrand et al. 2016; Kynard et al. 2016). Interpreting
foraging behaviours from fine-scale telemetry is not new to sturgeon research. A VPS study of
Atlantic sturgeon within an intertidal summer feeding area identified three types of movement
which they related to foraging strategies and prey availability (McLean et al. 2014). These
movements were described as Type 1) slow and winding (i.e. less intense turn angles), Type 2)
fast and tortuous (i.e. more intense turn angles), and Type 3) fast and linear. While Type 3 was
more indicative of larger migratory movements in and out of the feeding area, McLean et al
(2014) interpreted the decreased distance and turn angles of Type 1 as representing active
feeding within a suitable prey patch; and the increased distance and turn angles of Type 2 as

26

representing a searching strategy which can cover large areas. These two types of foraging
strategies have also been identified in lab studies of sturgeon (Kasumyan 1999) and other fish
species (Killen et al. 2007; Grant et al. 1987; Fausch et al. 1997; O’Brien et al. 1989), and
changes in strategy are often explained by changes in prey abundance and distribution
(Higginson et al. 2015). With descriptions of Type 1 and Type 2 movements from McLean et al
(2014) aligning closely with the residential and transitory behaviours identified in my study, the
mechanisms for these behaviour states might reflect the feeding ecology of white sturgeon and
offer insight into how sturgeon are influenced by hydroelectric operations.
Overall, my results reflected that environmental covariates did not have a strong effect on
fish changing from one behaviour state to another (e.g. 100% residential probability to 100%
transitory probability). Analyses at both a fine and coarse timescale revealed that white sturgeon
often remain in the same state, with the probability at time t-1 having high autocorrelation to the
probability at time t (Figures 2.8 and 2.12). However, trends were observed seasonally and
spatially (Figure 2.4). The proportion of time fish expressed residential behaviour (>75%
probability) was greatest in the winter and spring, and lowest in the summer and fall. Across
habitat zones, the HLK Eddy followed by the ALH Tailrace had the greatest proportion of time
as residential. Fish spent less time overall in the HLK Tailrace and Downstream, as well as
having a lower proportion of time expressing residential behaviour. It is possible the habitat
zones in my analyses do not appropriately capture the more subtle spatial distribution of
behaviour across time. Patches of residential behaviour (>75% probability), for example, were
observed to shift seasonally even within a single habitat zone (Figure 2.6). These seasonal and
spatial trends are likely influenced by environmental variables related to hydroelectric operations
(e.g. water temperature, discharge). In cases where there was less certainty of a behaviour state

27

in the previous timestep (50% probability), the influence of environmental covariates on
behaviour was reflected in my results.
My results demonstrated that water temperature had a strong influence on behaviour state
when there was uncertainty of the state in the prior time step. Warming temperature led to
decreased mean weekly probabilities of residential behaviour given probabilities at time t-1 were
50% (Figure 2.14). These results align with seasonal trends described earlier, with sturgeon
during warmer seasons (i.e. summer and fall; Figure 2.18) spending the lowest proportion of
time as residential (Figure 2.4). Observing changes in movement activity in response to water
temperature is not uncommon for white sturgeon (Hildebrand et al. 2016; Robichaud & English
2017), as water temperature is a driver of fish metabolism and alters demands for prey
consumption (Volkoff et al. 2020). In terms of feeding ecology, it would be reasonable to assume
that sturgeon are adopting transitory behaviour, a more high-energy search strategy, as metabolic
demands for energy increase with warming water temperatures. Transitory behaviour also allows
for the coverage of larger areas in the pursuit of prey, and in the summer and fall I observed not
only more frequent but also more dispersed transitory behaviour across the study area (Figure
2.7). While water temperature appeared to influence what behaviour state was more
predominantly expressed, discharge appeared to influence where each behaviour state was
expressed.
The influence of discharge on behaviour was observed in relation to individual facilities
and the spatial distribution of behaviour across seasons. Discharge has been shown to influence
fish movement by altering the energetic demands of fish to maintain position and move, with
greater energy required in higher flows (Shepard et al. 2013). In my mean weekly analysis of
behaviour, increased discharge from the ALH facility was related to increased mean weekly

28

probabilities of residential behaviour given probabilities at time t-1 were 50% (Figure 2.15). This
result may reflect the increased likelihood of sturgeon to adopt low-energy residential behaviour
in high flows, as well as the importance of ALH discharge on the flow environment (e.g. most
consistent source of inflow throughout a given year). The spatial distribution of where fish
expressed residential behaviour also appeared to be influenced by discharge. For example, in the
summer and winter when ALH and HLK are both discharging water (Figure 2.19), patches of
residential behaviour are observed in the center of the HLK Eddy where circulation is
strengthened by discharge from both facilities (Figure 2.6 and 2.20). Patches of residential
behaviour can also be seen in areas adjacent to inflows and behind environmental structures (e.g.
peninsulas) providing relief from water velocities. In the fall and spring however, when
discharge from the HLK facility is greatly reduced (Figure 2.19), residential patches either
disappear or shift away from the eddy’s center as flow dynamics change and circulation is
reduced (Figures 2.6 and 2.21). Residential behaviour, in terms of feeding ecology, represents a
movement strategy adopted when an area of suitable prey abundance has been identified.
Discharge has been shown to alter prey abundance and distribution within habitats (Taylor et al.
2012; Turschwell et al. 2019). For sturgeon, research has suggested that low velocity habitats
adjacent to fast-flowing waters can foster productive feeding areas due to the continuous
deposition of resources from upstream inflows and minimal energy required for movement
(Hildebrand & Parsley 2013). Adjustments in facility discharge, therefore, may be causing the
relocation of these productive feeding areas and changing spatial distribution of residential
behaviour through time.
While feeding ecology is a relevant and insightful lens in which to interpret the behaviour
states identified in this study, other biological functions and activities may be important. The

29

movement metrics of residential behaviour, specifically, could possibly reflect functions such as
overwintering or spawning. Overwintering is a state in which fish activity levels and metabolic
rates decline in response to cooling water temperatures. In sturgeon, feeding does still occur in
river systems that do not freeze-over, however, high-energy movements and foraging in general
is greatly reduced (Hildebrand et al. 2016). The decreased probability of transitory behaviour and
low proportion of time as transitory observed in the cooler winter and spring seasons is likely a
reflection of the overwintering biology of these fish. Visually, one can even observe transitory
behaviour being less dispersed across habitat zones when it does occur in cooler water
temperatures (Figure 2.7). For residential behaviour, feeding and overwintering biology may be
overlapping. Metabolic demands and movement are greatly reduced due to overwintering,
however, seasonal shifts in the location of residential patches between the winter and spring
suggest white sturgeon are still making spatial selections in response to discharge dynamics and
possible changing prey availability.
In this study, I could not differentiate spawning activity from residential or transitory
behaviour. White sturgeon in the UCR typically spawn during the summer from June to July, and
spawning habitat in the ALH tailrace has been confirmed via passive embryo and larvae
collection (Terraquatic 2011). While sturgeon did express residential and transitory behaviour in
the known spawning area of the ALH tailrace during the spawning period, there is little
supporting knowledge of fine-scale spawning movements that would relate either behaviour state
to that activity. Sturgeon spawning is rarely observed because it often takes place in deep and
fast-flowing waters (Hildebrand et al. 2016). The few descriptions of white sturgeon movement
in known spawning areas include surface breaching and rolling (Parsley et al. 1993), as well as
video of a presumed gamete release from a spawning pair after rapid body undulations and

30

unison upstream movement (Liebe et al. 2004). Even with more easily observable species, mateinteraction can be hard to identify in HMMs. It is suggested models incorporate more
independent data to reduce assumptions about spawning behaviour inferred from location data
(Buderman et al. 2021). In my case, this could include timestamped embryo collection,
reproductive stage, or mate availability. An embryo and larvae collection program in the ALH
Tailrace did take place concurrently with this study, however no embryos or larvae were
collected and spawning activity could therefore not be confirmed (BC Hydro 2016).
While the interpretation of movement behaviour is not without uncertainty, analyzing
telemetry data in this way is an insightful and effective method for understanding fish
movement. Due to their conservation status, telemetry arrays tracking the movements of white
sturgeon, as well as other riverine species, are plentiful. However, very little of this research
becomes published (Fraser et al. 2018) or assesses movement beyond basic detection summaries
(Holyoak et al. 2008; Katzner & Arlettaz 2020). While detection summaries do provide useful
information such as habitat occupation and population densities, they lack the ability to
deconstruct the complexities of movement and analyze movement in a more ecologically
meaningful way. Using movement behaviour analyses, I was able to identify and relate states of
movement to critical ecological functions such as feeding, and determine the effects of
environmental variables on such states. Discharge, an environmental variable controlled directly
by hydroelectric operations, demonstrated a strong spatial correlation to movement patterns
suggestive of a low-energy feeding strategy and likely changing prey availability. Knowledge
like this can help inform recovery measures for species impacted by dam operations, such as the
optimization of discharge management to reduce the impacts of river regulation during periods
of critical feeding activity. Overall, analyses of movement behaviour proved a useful tool in

31

understanding white sturgeon movement below a hydroelectric facility and I would recommend
the application of similar comprehensive frameworks in fish movement ecology and telemetry.

32

Table 2.1. Models predicting the probability of residential behaviour given the probability of
residential behaviour at time t-1 ranked by increasing order of the bias-corrected Akaike
Information Criterion (AIC c). Differences in AIC c relative to the top model (ΔAIC c), weighted
AIC c (wAIC c), cumulative weighted AIC c (cwAIC c), log-likelihood (log(L)), and the number of
parameters in a model (K) are shown. The top two models were selected for averaged
predictions.

zdisc, Combined discharge from ALH and HLK facilities; zalh, Discharge from ALH facility;
zhlk, Discharge from HLK facility; zone, Habitat zone; tm1, Probability of residential behaviour
at time t-1; seas, Season; sex, Sex of fish; ztemp, Water temperature. All models include bout
nested within individual as a random effect.

33

Table 2.2. Models predicting the mean weekly probability of residential behaviour given the
mean probability of residential behaviour at time t-1 ranked by increasing order of the biascorrected Akaike Information Criterion (AIC c). Differences in AIC c relative to the top model
(ΔAICc), weighted AIC c (wAICc), cumulative weighted AIC c (cwAIC c), log-likelihood (log(L)),
and the number of parameters in a model (K) are shown. The top four models were selected for
averaged predictions.

zdisc, Combined discharge from ALH and HLK facilities; zalh, Discharge from ALH facility;
zhlk, Discharge from HLK facility; zone, Habitat zone; tm1, Mean weekly probability of
residential behaviour at time t-1; seas, Season; sex, Sex of fish; ztemp, Water temperature. All
models include individual as a random effect.

34

Figure 2.1. Map of study area below Hugh L. Keenleyside Dam (HLK) and Arrow Lakes
Generating Station (ALH) (black dots), British Columbia, Canada. Habitat zones are highlighted
by colour, acoustic receivers are represented by white circles, and anchored sentinel tags are
represented by yellow circles.

35

Figure 2.2. State-dependent distributions for step length (meters) and turning angle (radians).

36

Figure 2.3. Individual bouts from two white sturgeon in the study area with behaviour state
estimated by the hidden Markov model represented by colour. Ovals represent the error ellipse
associated with a position.

37

Figure 2.4. Proportion of time white sturgeon spent having an upper (>75%), middle (25-75%),
or lower (<25%) probability of being in residential behaviour. Results are separated by season
and habitat zone.

38

Figure 2.5. Proportion of time individual white sturgeon spent having an upper (>75%), middle
(25-75%), or lower (<25%) probability of being in residential behaviour. Results are separated
by season and empty columns indicate a fish was not detected in a season.

39

Figure 2.6. White sturgeon positions having greater than 75% probability of being in a residential
behaviour state. Positions are shown by season with black lines distinguishing habitat zones.

40

Figure 2.7. White sturgeon positions having greater than 75% probability of being in a transitory
behaviour state. Positions are shown by season with black lines distinguishing habitat zones.

41

Figure 2.8. Model averaged coefficients and standard errors (SE) of the beta generalized linear
mixed model parameters predicting the probability of residential behaviour given the probability
of residential behaviour at time t-1. Model averaged intercept estimate of -3.76 (± 0.06 SE).

42

Figure 2.9. Model averaged predictions of the probability of being in the residential behaviour
state by season, habitat zone, and probability of residential behaviour at time t-1. Whiskers
denote 95% confidence interval. Categorical covariates sex and photoperiod were set to Female
and Day respectively.

43

Figure 2.10. Model averaged predictions of the probability of being in the residential behaviour
state by photoperiod and probability of residential behaviour at time t-1. Whiskers denote 95%
confidence interval. Categorical covariates sex, season, and habitat zone were set to Female,
Summer, and HLK Eddy respectively.

44

Figure 2.11. Model averaged predictions of the probability of being in the residential behaviour
state by sex and probability of residential behaviour at time t-1. Whiskers denote 95%
confidence interval. Categorical covariates photoperiod, season, and habitat zone were set to
Day, Summer, and HLK Eddy respectively.

45

Figure 2.12. Model averaged coefficients and standard errors (SE) of the beta generalized linear
mixed model parameters predicting the mean weekly probability of residential behaviour given
the mean weekly probability of residential behaviour at time t-1. Model averaged intercept
estimate of -3.30 (± 0.07 SE).

46

Figure 2.13. Model averaged predictions of the mean weekly probability of being in the
residential behaviour state by habitat zone and mean weekly probability of residential behaviour
at time t-1. Whiskers denote 95% confidence interval. To isolate effects, other numeric model
covariates were fixed at their mean and the categorical covariate sex was set to Female.

47

Figure 2.14. Model averaged predictions of the mean weekly probability of being in the
residential behaviour state by water temperature and mean weekly probability of residential
behaviour at time t-1. Shaded areas denote 95% confidence interval. To isolate effects, other
numeric model covariates were fixed at their mean and the categorical covariates sex and habitat
zone were set to Female and HLK Eddy respectively.

48

Figure 2.15. Model averaged predictions of the mean weekly probability of being in the
residential behaviour state by ALH discharge and mean weekly probability of residential
behaviour at time t-1. Shaded areas denote 95% confidence interval. To isolate effects, other
numeric model covariates were fixed at their mean and the categorical covariates sex and habitat
zone were set to Female and HLK Eddy respectively.

49

Figure 2.16. Model averaged predictions of the mean weekly probability of being in the
residential behaviour state by HLK discharge and mean weekly probability of residential
behaviour at time t-1. Shaded areas denote 95% confidence interval. To isolate effects, other
numeric model covariates were fixed at their mean and the categorical covariates sex and habitat
zone were set to Female and HLK Eddy respectively.

50

Figure 2.17. Model averaged predictions of the mean weekly probability of being in the
residential behaviour state by sex and mean weekly probability of residential behaviour at time t1. Whiskers denote 95% confidence interval. To isolate effects, other numeric model covariates
were fixed at their mean and the categorical covariate habitat zone was set to HLK Eddy.

51

Figure 2.18. Water temperature in the study area from December 2013 – November 2014. Blue
line represents LOESS smoothed data and grey points represent raw data.

52

Figure 2.19. Discharge from Hugh L. Keenleyside Dam (HLK) and Arrow Lakes Generating
Station (ALH) from December 2013 – November 2014. Total discharge is the combination of
HLK and ALH. Blue lines represent LOESS smoothed data and grey points represent raw data.

Figure 2.20. Modeled velocity conditions during peak summer discharge below Hugh L. Keenleyside Dam (HLK) and Arrow Lakes
Generating Station (ALH), British Columbia, Canada. Velocity and direction of discharge represented by coloured arrows. Figure
pulled from West et al. (2020) unpublished data.

53

Figure 2.21. Modeled velocity conditions during low fall discharge below Hugh L. Keenleyside Dam (HLK) and Arrow Lakes
Generating Station (ALH), British Columbia, Canada. Velocity and direction of discharge represented by coloured arrows. Figure
pulled from West et al. (2020) unpublished data.

54

55

References
Bacheler NM, Michelot T, Cheshire RT, Shertzer KW (2019) Fine-scale movement patterns and
behavioural states of gray triggerfish Balistes capriscus determined from acoustic telemetry and
hidden Markov models. Fish Res 215:76-89. https://doi.org/10.1016/j.fishres.2019.02.014
Bartoń K (2020) MuMIn: multi-model inference. R package version 1.43.17. https://CRAN.Rproject.org/package=MuMIn
BC Hydro (2015) Lower Columbia River Adult White Sturgeon Monitoring Program
(CLBMON-28). Years 5 and 6 Data Report. Report by BC Hydro, Castlegar, BC.
BC Hydro (2016) Lower Columbia River Juvenile Detection Program (CLBMON-29). Years 6
and 7 Data Report. Report by BC Hydro, Castlegar, BC.
BC Hydro (2018a) Lower Columbia River Adult White Sturgeon Monitoring Program
(CLBMON-28). Year 10 Data Report. Report by BC Hydro, Castlegar, BC.
BC Hydro (2021) Arrow Operations Summary.
Bergé J, Capra H, Pella H, Steig T, Ovidio M, Bultel E, Lamouroux N (2012) Probability of
detection and positioning error of a hydro acoustic telemetry system in a fast-flowing river:
Intrinsic and environmental determinants. Fish Res 125:1-13.
https://doi.org/10.1016/j.fishres.2012.02.008
Buderman FE, Gingery TM, Diefenbach DR, Gigliotti LC, Begley-Miller D, McDill MM,
Wallingford BD, Rosenberry CS, Drohan PJ (2021) Caution is warranted when using animal
space-use and movement to infer behavioural states. Mov Ecol 9:30.
https://doi.org/10.1186/s40462-021-00264-8
Burnham KP, Anderson DR (2004) Multimodel inference: understanding AIC and BIC in model
selection. Sociol Methods Res 33:261-304. https://doi.org/10.1177/0049124104268644
Brooks ME, Kristensen K, Van Benthem KJ, Magnusson A, Berg CW, Nielsen A, Skaug HJ,
Machler M, Bolker BM (2017) glmmTMB balances speed and flexibility among packages for
zero-inflated generalized linear mixed modeling. R J 9:378-400. https://doi.org/10.3929/ethz-b000240890
Coates JH, Hovel KA, Butler JL, Klimley AP, Morgan SG (2013) Movement and home range of
pink abalone Haliotis corrugata: implications for restoration and population recovery. Mar Ecol
Prog Ser 486:189-201. https://doi.org/10.3354/meps10365
Cooke S, Paukert C, Hogan Z (2012) Endangered river fish: factors hindering conservation and
restoration. Endanger Species Res 17:179-191. https://doi.org/10.3354/esr00426
Cooke SJ, Martins EG, Struthers DP, Gutowsky LF, Power M, Doka SE, Dettmers JM, Crook
DA, Lucas MC, Holbrook CM, Krueger CC (2016) A moving target—incorporating knowledge

56

of the spatial ecology of fish into the assessment and management of freshwater fish populations.
Environ Monit Assess 188:239. https://doi.org/10.1007/s10661-016-5228-0
Cote D, Morris CJ, Regular PM, Piersiak MG (2020) Effects of 2D seismic on snow crab
movement behaviour. Fish Res 230:105661. https://doi.org/10.1016/j.fishres.2020.105661
Darwall WRT, Freyhof J (2016) Lost fishes, who’s counting? The extent of the threat to
freshwater fish biodiversity. In: Closs GP (ed) Conservation of Freshwater Fishes. Cambridge
University Press, Cambridge, pp 1-36
Dudgeon D, Arthington AH, Gessner MO, Kawabata ZI, Knowler DJ, Lévêque C, Naiman RJ,
Prieur-Richard AH, Soto D, Stiassny MLJ, Sullivan CA (2006) Freshwater biodiversity:
importance, threats, status and conservation challenges. Biol Rev 81:163-182.
https://doi.org/10.1017/s1464793105006950
Dynesius M, Nilsson C (1994) Fragmentation and flow regulation of river systems in the
northern third of the world. Science 266:753-762. https://doi.org/10.1126/science.266.5186.753
Espinoza N, Farrugia TJ, Lowe CG (2011a) Habitat use, movements and site fidelity of the gray
smoothhound shark (Mustelus californicus Gill 1863) in a newly restored southern California
estuary. J Exp Mar Biol Ecol 401:63-74. https://doi.org/10.1016/j.jembe.2011.03.001
Espinoza N, Farrugia TJ, Webber DM, Smith F, Lowe CG (2011b) Testing a new acoustic
telemetry technique to quantify long-term, fine-scale movements of aquatic animals. Fish Res
108:364-371. https://doi.org/10.1016/j.fishres.2011.01.011
Farrugia TJ, Espinoza M, Lowe CG (2011) Abundance, habitat use and movement patterns of
the shovelnose guitarfish (Rhinobatos productus) in a restored southern California estuary. Mar
Freshw Res 62:648-657. https://doi.org/10.1071/MF10173
Fausch KD, Nakano S, Kitano S (1997) Experimentally induced foraging mode shift by
sympatric charrs in a Japanese mountain stream. Behav Ecol 8:414-420.
https://doi.org/10.1093/beheco/8.4.414
Fisheries and Oceans Canada (2014) Recovery strategy for white sturgeon (Acipenser
transmontanus) in Canada. In Species at Risk Act Recovery Strategy Series. Fisheries and
Oceans Canada, Ottawa, ON.
Fraser KC, Davies KTA, Davy CM, Ford AT, Flockhart DTT, Martins EG (2018) Tracking the
conservation promise of movement ecology. Front Ecol Evol 6:150.
https://doi.org/10.3389/fevo.2018.00150
Freshwater Fisheries Society of British Columbia (2014) Columbia White Sturgeon
Conservation Fish Culture Program. Report prepared for BC Hydro, Water License
Requirements, Castlegar, BC.

57

Gillespie MA, McDougall CA, Nelson PA, Sutton T, MacDonell DS (2020) Observations
regarding Lake Sturgeon spawning below a hydroelectric generating station on a large river
based on egg deposition studies. River Res Appl 36:2024-2042. https://doi.org/10.1002/rra.3731
Golder Associates Ltd (2006) Upper Columbia White Sturgeon Stock Monitoring and Data
Management Program: Synthesis Report. Report prepared for British Columbia Ministry of
Environment, Nelson, BC.
Golder Associates Ltd (2015) Preliminary VPS data analysis of white sturgeon behaviour below
the Hugh L. Keenleyside Dam and Arrow Lakes Generating Station. Reported prepared for
Terraquatic Resource Management and Jay Environmental, Nelson, BC.
Grant JW, Noakes DL (1987) Movers and stayers: foraging tactics of young-of-the-year brook
charr, Salvelinus fontinalis. J Anim Ecol 1:1001-1013. https://doi.org/10.2307/4962
Grill G, Lehner B, Lumsdon AE, MacDonald GK, Zarfl C, Liermann CR (2015) An index-based
framework for assessing patterns and trends in river fragmentation and flow regulation by global
dams at multiple scales. Environ Res Lett 10:015001.
Grill G, Lehner B, Thieme M, Geenen B, Tickner D, Antonelli F, Babu S, Borrelli P, Cheng L,
Crochetiere H, Macedo HE, Filgueiras R, Goichot M, Higgins J, Hogan Z, Lip B, McClain ME,
Meng J, Mulligan M, Nilsson C, Olden JD, Opperman JJ, Petry P, Liermann CR, Sáenz L,
Salinas-Rodríguez S, Schelle P, Schmitt RJP, Snider J, Tan F, Tockner K, Valdujo PH, van
Soesbergen A, Zarfl C (2019) Mapping the world’s free-flowing rivers. Nature 569:215-221.
https://doi.org/10.1038/s41586-019-1111-9
Haxton T, Cano T (2016) A global perspective of fragmentation on a declining taxon—the
sturgeon (Acipenseriformes). Endanger Species Res 31:203-210.
https://doi.org/10.3354/esr00767
Higginson AD, Ruxton GD (2015) Foraging mode switching: the importance of prey distribution
and foraging currency. Anim Behav 105:121-137. https://doi.org/10.1016/j.anbehav.2015.04.014
Hildebrand L, McLeod C, McKenzie S (1999) Status and management of White Sturgeon in the
Columbia River in British Columbia, Canada: An Overview. J Appl Ichthyol 15:164-172.
https://doi.org/10.1111/j.1439-0426.1999.tb00227.x
Hildebrand LR, Parsley M (2013) Upper Columbia White Sturgeon Recovery Plan - 2012
Revision. Prepared for the Upper Columbia White Sturgeon Recovery Initiative.
Hildebrand LR, Schreier AD, Lepla K, McAdam SO, McLellan J, Parsley MJ, Paragamian VL,
Young SP (2016) Status of white sturgeon (Acipenser transmontanus Richardson 1863)
throughout the species range, threats to survival, and prognosis for the future. J Appl Ichthyol
32:261-312. https://doi.org/10.1111/jai.13243
Holyoak M, Casagrandi R, Nathan R, Revilla E, Spiegel O (2008) Trends and missing parts in
the study of movement ecology. Proc Natl Acad Sci USA 105:19060-19065.
https://doi.org/10.1073/pnas.0800483105

58

Hussey NE, Kessel ST, Aarestrup K, Cooke SJ, Cowley PD, Fisk AT, Harcourt RG, Holland
KN, Iverson SJ, Kocik JF, Flemming JEM, Whoriskey FG (2015) Aquatic animal telemetry: a
panoramic window into the underwater world. Science 348:1255642.
https://doi.org/10.1126/science.1255642
Johnson DS, London JM, Lea MA, Durban JW (2008). Continuous-time correlated random walk
model for animal telemetry data. Ecology 89:1208-1215. https://doi.org/10.1890/07-1032.1
Jonsen ID, Flemming JM, Myers RA (2005) Robust state-space modeling of animal movement
data. Ecology 86:2874-2880. https://doi.org/10.1890/04-1852
Joo R, Boone ME, Clay TA, Patrick SC, Clusella‐Trullas S, Basille M (2019) Navigating
through the r packages for movement. J Anim Ecol 89:248-267. https://doi.org/10.1111/13652656.13116
Kasumyan AO (1999) Olfaction and taste senses in sturgeon behaviour. J Appl Ichthyol 15:228232. https://doi.org/10.1111/j.1439-0426.1999.tb00240.x
Katzner TE, Arlettaz R (2020) Evaluating contributions of recent tracking-based animal
movement ecology to conservation management. Frontiers Ecol Evol 7:519.
https://doi.org/10.3389/fevo.2019.00519
Killen SS, Brown JA, Gamperl AK (2007) The effect of prey density on foraging mode selection
in juvenile lumpfish: balancing food intake with the metabolic cost of foraging. J Anim Ecol
76:814-825. https://doi.org/10.1111/j.1365-2656.2007.01237.x
Kristensen K, Nielsen A, Berg CW, Skaug H, Bell BM (2016) TMB: automatic differentiation
and Laplace approximation. J Stat Softw 70:1-21. https://doi.org/10.18637/jss.v070.i05
Kynard B, Bolden S, Kieffer M, Collins M, Brundage H, Hilton EJ, Litvak M, Kinnison MT,
King T, Peterson D (2016) Life history and status of Shortnose Sturgeon (Acipenser
brevirostrum LeSueur, 1818). J Appl Ichthyol 32:208-48. https://doi.org/10.1111/jai.13244
Lennox RJ, Aarestrup K, Cooke SJ, Cowley PD, Deng ZD, Fisk AT, Harcourt RG, Heupel M,
Hinch SG, Holland KN, Hussey NE, Iverson SJ, Kessel ST, Kocik JF, Lucas MC, Flemming JM,
Nguyen VM, Stokesbury MJW, Vagle S, VanderZwaag DL, Whoriskey FG, Young N (2017)
Envisioning the future of aquatic animal tracking: technology, science, and application.
Bioscience 67:884-896. https://doi.org/10.1093/biosci/bix138
Liebe R, Rublee B, Sykes G, Manson R (2004) Adult White Sturgeon Monitoring - Nechako
River 2004. Report prepared for Alcan Primary Metal by Triton Environmental Consultants Ltd,
Kitimat, BC.
Makowski D, Ben-Shachar MS, Chen SHA, Lüdecke D (2019) Indices of effect existence and
significance in the Bayesian framework. Front Physiol 10:2767.
https://doi.org/10.3389/fpsyg.2019.02767

59

McClintock BT, Michelot T (2018) momentuHMM: R package for generalized hidden Markov
models of animal movement. Methods Ecol Evol 9:1518-1530. https://doi.org/10.1111/2041210X.12995
McClintock BT, Langrock R, Gimenez O, Cam E, Borchers DL, Glennie R, Patterson TA (2020)
Uncovering ecological state dynamics with hidden Markov models. Ecol Lett 23:1878-1903.
https://doi.org/10.1111/ele.13610
McElreath, R (2020) Statistical rethinking: a Bayesian course with examples in R and Stan.
Chapman and Hall/CRC. https://doi.org/10.1201/9780429029608
McLean M, Simpfendorfer C, Heupel M, Dadswell M, Stokesbury M (2014). Diversity of
behavioural patterns displayed by a summer feeding aggregation of Atlantic sturgeon in the
intertidal region of Minas Basin, Bay of Fundy, Canada. Mar Ecol Prog Ser 496:59-69.
https://doi.org/10.3354/meps10555
Michelot T, Langrock R, Patterson TA (2016) moveHMM: an R package for the statistical
modelling of animal movement data using hidden Markov models. Methods Ecol Evol 7:13081315. https://doi.org/10.1111/2041-210x.12578
Morales JM, Haydon DT, Frair J, Holsinger KE, Fryxell JM (2004) Extracting more out of
relocation data: building movement models as mixtures of random walks. Ecology 85:436-2445.
https://doi.org/10.1890/03-0269
Mueller T, Fagan WF (2008) Search and navigation in dynamic environments – from individual
behaviours to population distributions. Oikos 117:654-664. https://doi.org/10.1111/j.00301299.2008.16291.x
Nathan R, Getz WM, Revilla E, Holyoak M, Kadmon R, Saltz D, Smouse PE (2008) A
movement ecology paradigm for unifying organismal movement research. Proc Natl Acad Sci
USA 105:19052-9. https://doi.org/10.1073/pnas.0800375105
Nigro AA, Rieman BE, Elliot JC, Engle DR (1988) Status and habitat requirements of white
sturgeon populations in the Columbia River downstream from McNary Dam. Annual Progress
Report to Bonneville Power Administration, Portland, OR.
Nilsson C, Reidy CA, Dynesius M, Revenga C (2005) Fragmentation and flow regulation of the
world’s large river systems. Science 308:405-408 https://doi.org/10.1126/science.1107887
O'Brien WJ, Evans BI, Browman HI, O'Brian WJ (1989) Flexible search tactics and efficient
foraging in saltatory searching animals. Oecologia 80:100-110
https://www.jstor.org/stable/4219015
Parsley MJ, Beckman LG (1993) Spawning and rearing habitat use by white sturgeon in the
Columbia River downstream from McNary Dam. Trans Am Fish Soc 122:217-227.
https://doi.org/10.1577/1548-8659(1993)122<0217:SARHUB>2.3.CO;2

60

Patterson TA, Parton A, Langrock R, Blackwell PG, Thomas L, King R (2017) Statistical
modelling of individual animal movement: an overview of key methods and a discussion of
practical challenges. AStA Adv Stat Anal 101:399-438. https://doi.org/10.1007/s10182-0170302-7
Pohle J, Langrock R, van Beest FM, Schmidt NM (2017) Selecting the number of states in
hidden Markov models: pragmatic solutions illustrated using animal movement. J Agric Biol
Environ Stat 22:270-293. https://doi.org/10.1007/s13253-017-0283-8
R Core Team (2021) R: a language and environment for statistical computing. R Foundation,
Vienna, Austria. https://www.R-project.org
Ricciardi A, Rasmussen JB (1999) Extinction rates of North American freshwater fauna.
Conserv Biol 13:1220-1222. http://www.jstor.org/stable/2641755
Rennie MD, Collins NC, Shuter BJ, Rajotte JW, Couture P (2005) A comparison of methods for
estimating activity costs of wild fish populations: more active fish observed to grow slower. Can
J Fish Aquat Sci 62:767-780. https://doi.org/10.1139/f05-052
Robichaud D, English KK, Nelson TC (2017) Annual movements of acoustic-tagged white
sturgeon in the lower reaches of the Fraser river and its tributaries. Trans Am Fish Soc 146:611625. https://doi.org/10.1080/00028487.2017.1294542
Roy R, Beguin J, Argillier C, Tissot L, Smith F, Smedbol S, De-Oliveira E (2014) Testing the
VEMCO Positioning System: spatial distribution of the probability of location and the
positioning error in reservoirs. Anim Biotelemetry 2:1. https://doi.org/10.1186/2050-3385-2-1
Shaw AK (2020) Causes and consequences of individual variation in animal movement. Mov
Ecol 8:12. https://doi.org/10.1186/s40462-020-0197-x
Shepard ELC, Wilson RP, Rees WG, Grundy E, Lambertucci SA, Vosper SB (2013) Energy
Landscapes Shape Animal Movement Ecology. Am Nat 182:298-312.
https://doi.org/10.1086/671257
Smith F (2013) Understanding HPE in the VEMCO Positioning System (VPS).
http://vemco.com/wp-content/uploads/2013/09/understanding-hpe-vps.pdf
Taylor MK, Cooke SJ (2012) Meta-analyses of the effects of river flow on fish movement and
activity. Environ Rev 20:211-219. https://doi.org/10.1139/a2012-009
Terraquatic Resource Management (2011) Arrow Lakes generating station white sturgeon spawn
monitoring program. Unpublished report for Columbia Power Corporation, Castlegar, BC.
Tuomainen U, Candolin U (2011) Behavioural responses to human‐induced environmental
change. Biol Rev 86:640-657. https://doi.org/10.1111/j.1469-185x.2010.00164.x

61

Turschwell MP, Stewart‐Koster B, King AJ, Pusey B, Crook D, Boone E, Douglas M, Allsop Q,
Jackson S, Kennard MJ (2019) Flow‐mediated predator-prey dynamics influence fish
populations in a tropical river. Freshwater Biol 64:1453-1466. https://doi.org/10.1111/fwb.13318
Volkoff H, Rønnestad I (2020) Effects of temperature on feeding and digestive processes in fish.
Temp 7:1-14. https://doi.org/10.1080/23328940.2020.1765950
Webb MAH, Van Eenennaam JP, Crossman JA, Chapman FA (2019) A practical guide for
assigning sex and stage of maturity in sturgeons and paddlefish. J Appl Ichthyol 35:169-186.
https://doi.org/10.1111/jai.13582
West DT, Bayly MJ, Tamminga AD, Perkins T, Porto L, Parsley M, Hatfield T (2020) Lower
Columbia White Sturgeon Habitat Restoration Alternatives – Final Report (CLBWORKS-27).
Report prepared for BC Hydro and Power Authority by Ecofish Research Ltd.
Whoriskey K, Martins EG, Auger‐Méthé M, Gutowsky LFG, Lennox RJ, Cooke SJ, Power M,
Flemming JM (2019) Current and emerging statistical techniques for aquatic telemetry data: a
guide to analyzing spatially discrete animal detections. Methods Ecol Evol 10:935-948.
https://doi.org/10.1111/2041-210X.13188

62

CHAPTER 3

COARSE-SCALE MOVEMENT ECOLOGY AND BEHAVIOUR PATTERNS OF
HATCHERY-ORIGIN WHITE STURGEON IN THE REGULATED UPPER COLUMBIA
RIVER

63

Introduction
Movement ecology has become a powerful tool in informing management and recovery
measures for threatened species (Allen et al. 2016). Movement is essential to most animals, and
can be linked to the growth, survival, and reproductive success of individuals and populationlevel processes (Shaw 2020). Some species, however, face greater challenges to movement than
others. Migratory fish that cross international boundaries can be subject to compounding threats
when transboundary management and cooperation is absent or ineffective (Cooke et al. 2012;
Runge et al. 2014). Such threats can include overfishing, habitat degradation, loss of connectivity
between critical habitats (e.g. spawning, feeding), and alterations in water quality (e.g. water
temperature) and natural flow regimes (Dudgeon et al. 2006). Trends showing population
declines of many transboundary fish highlight this issue (Valbo-Jørgensen et al. 2008; Limburg
& Waldman 2009; Palacios-Abrantes et al. 2020). In response, the formation of international
conventions (e.g. Convention on Migratory Species) and regional river basin collectives (e.g.
Great Lakes Fishery Commission) has become a common strategy in facilitating cooperation and
species recovery between nations (Cooke et al. 2012). These groups, however, are only as
effective as the knowledge informing their management decisions. Research on the movement
ecology of migratory species would offer valuable insight for improving and developing
international mitigative measures, as it allows for a better understanding of how animals move
spatially, temporally, and in response to human activities (e.g. river regulation).
In this study, I examined the movements of hatchery-origin white sturgeon in the Canada
– United States Transboundary Reach of the Upper Columbia River (UCR). White sturgeon are a
species at risk almost everywhere they occur due to issues such as habitat fragmentation, habitat
degradation, and historical fisheries (Hildebrand et al. 2016). In the UCR, white sturgeon have

64

been experiencing recruitment failure and population declines over the last century. They were
listed as Endangered under the Canadian Species at Risk Act in 2006 (Fisheries and Oceans
2014), but remain a species of Least Concern under the Endangered Species Act in the United
States. Due to current recruitment estimates suggesting an inability to support a self-sustaining
population (Hildebrand & Parsley 2013), Canada and the United States implemented a joint
recovery effort and conservation aquaculture program in 2001. From its inception to 2018, the
conservation aquaculture program has released around 150,000 hatchery-origin juveniles (6 – 9
months of age) with an estimated survival of 15,000 in the Transboundary Reach (BC Hydro
2018b). While the program has been an overall success in delaying the extirpation of the species,
information is limited on how this new population of hatchery-origin sturgeon move within the
regulated and transboundary river section of the UCR. Better understanding movement and its
drivers for hatchery-origin sturgeon could offer more insights into their growth, survival, and
reproduction. Such insights would assist recovery programs in setting long-term population
targets and making decisions related to stocking, fishing, and discharge management.
Long-distance movements in sturgeon are often to locate feeding, overwintering, and
spawning habitat (Hildebrand et al. 2016; Kynard et al. 2016). Direct capture studies performed
annually in the spring and fall have shown that hatchery-origin white sturgeon have high site
fidelity and occur throughout the Transboundary Reach, with preferences for habitats
characterized by small substrate (e.g. sand, small gravel) and deep (>10 m), slow-moving water
(<1 m/s) (BC Hydro 2018a,b). Sturgeon presence in such habitats have been linked to their
ability to support productive feeding and overwintering areas (Hildebrand & Parsley 2013). The
greatest amount of movement observed from hatchery-origin sturgeon in the Transboundary
Reach has been immediately after hatchery release, with higher numbers of fish released in

65

Canada moving downstream and being first recaptured in the United States compared to fish
released in the United States being first recaptured in Canada (BC Hydro 2020). Similar
movement trends have also been observed in Kootenay River hatchery-origin sturgeon, with
post-hatchery dispersal linked to the availability of habitat with small substrate and slow-moving
water (Neufeld et al. 2009). Such habitat preferences may be the reasoning for dispersal trends in
the Transboundary Reach, as slow-moving (i.e. low-velocity) habitat is more abundant in the
United States compared to Canada. Information is lacking on how hatchery-origin sturgeon are
moving after this initial dispersal. Implementing movement analyses beyond direct captures
could offer a more detailed understanding of hatchery-origin sturgeon movement year-round and
in response to changing environmental conditions by river regulation.
Direct capture studies, due to their occurrence outside of the spawning period, have not
been able to provide detailed information on the presence or movement of hatchery-origin
sturgeon to spawning areas in the UCR. This is significant because hatchery-origin sturgeon are
beginning to reach sexual maturity. Males specifically are of interest as they typically mature at
smaller sizes and ages (~15 years) than females (~25 years), and they are able to spawn most
years following maturation while females require 3 – 5 year intervals between spawning events
(Hildebrand & Parsley 2013). Sexually mature sturgeon migrate to spawning areas in the UCR
typically between late June – July when water temperatures reach around 14 °C. However, unlike
river systems where sturgeon spawning movements can be defined by clear one-directional (e.g.
upstream or downstream) migrations (Balazik et al. 2020; Forsythe et al. 2012), sturgeon in the
UCR have access to multiple spawning areas both within and across international borders.
Assessing the movement of hatchery-origin sturgeon to spawning areas would help inform
recovery programs of the reproductive potential and possible genetic distribution of future

66

populations. Additionally, movement of hatchery-origin sturgeon from Canada to the United
States, whether to spawn or feed, could have direct implications for survival. Due to different
endangered species listings and management practices between countries, hatchery-origin white
sturgeon in the United States have been subject to fishing mortality since 2017 while remaining
protected from fishing in Canada. Better understanding movement between countries would also
assist in determining survival estimates and maintaining population targets in future hatchery
releases.
The objective of this study was to describe and quantify the coarse-scale movement
behaviour of hatchery-origin sturgeon residing throughout the Transboundary Reach of the UCR.
I applied hidden Markov models (HMMs) and generalized linear mixed models (GLMMs) to
acoustic telemetry data to investigate how movement behaviour varied spatially (between
habitats and countries), temporally (between seasons), and by biological factors (e.g. age, size,
sex) and river regulation (e.g. water temperature, discharge). Results will help inform
management decisions related to river regulation, hatchery release strategies, and the overall
cross-border recovery of this threatened species.

Methods
Study area
The study area resides on the Upper Columbia River between the Hugh L. Keenleyside
Dam (HLK) in British Columbia (443705 E, 5465875 N) and the Grand Coulee Dam in
Washington (352493 E, 5314347 N). This regulated 296 river kilometer (rkm) section is called
the Transboundary Reach due to it being divided at the Canada – United States international
border (rkm 1204.2) (Figure 3.1).

67

The Transboundary Reach is subdivided into 15 river zones, 11 of which were utilized by
my study population of hatchery-origin white sturgeon and will be the focus of this study (Figure
3.2). Zones 1 – 5 comprise the Canadian section of the Transboundary Reach, with Zone 2B
representing a tributary and regulated segment of the Kootenay River. Zones 6 – 10 comprise the
United States section of the Transboundary Reach. The range of these river zones are evenly
dispersed by country (~10 rkm in Canada and ~30 rkm in USA). Each river zone begins at the
termination of the previous zone. River zones are characterized by stream morphology and water
velocity. Zone 1 is predominantly deep (~20 m in the thalweg) and slow-moving (<1 m/s). Zones
2 – 6 are characterized by shallower depths (<10 m in the thalweg) and swifter flows (>1 m/s)
with occasional large, deep eddies (Crossman et al. 2016). In Zones 7 – 10, the UCR transitions
into a more reservoir-like environment, with wider reaches (~2,500 m) and slow-moving flows
(Hildebrand et al. 1999).
There are three main discharge sources and five known spawning areas in the
Transboundary Reach (Figures 3.1 and 3.2). In Zone 1, the most upstream impoundment is HLK
which regulates discharge of the UCR from the Arrow Lakes Reservoir. Discharge from HLK is
coordinated weekly between Canada and the United States under Columbia River Treaty
agreements (BC Hydro 2021). A known spawning area is located in Zone 1 immediately below
HLK (Fisheries and Oceans Canada 2014). Zone 2 contains a discharge input from the Kootenay
River which is regulated by Brilliant Dam in Zone 2B. A known spawning area in Zone 2 is
located around 4 rkm downstream of this confluence near Kinnaird, BC. Zone 5 contains a
discharge input from the Pend d’Oreille River which is regulated by Waneta Dam. A known
spawning area in Zone 5 is located 0.5 rkm downstream of this confluence near the international
border. The United States section of the Transboundary Reach (Zones 6 – 10) contains no main

68

discharge sources. However, known spawning areas have been identified near Northport, WA in
Zone 6 and China Bend, WA in Zone 7 (Fisheries and Oceans Canada 2014).
This study defined six environmental monitoring zones based on the locations of
temperature and discharge monitoring stations (Figure 3.2). Discharge records were obtained
from four gauge stations in Canada operated by BC Hydro and one gauge station in the United
States operated by USGS. Water temperature was monitored using underwater loggers (VEMCO
Minilog-II-T Temp Logger) at seven locations in Canada operated by BC Hydro and one
location in the United States operated by USGS. The distribution of environmental monitoring
zones was skewed, with the Canadian section of the Transboundary Reach containing the
majority of monitored zones and the United States section consisting mainly of one. While not
ideal, Canada does contain more input sources (e.g. tributaries), which would cause greater
fluctuations in temperature and discharge along this section of river.

Fish sampling and tagging
A total of 75 hatchery-origin white sturgeon were acoustically tagged for this study. Fish
were captured in the spring and fall of 2019. The sampling area extended from HLK in British
Columbia to Gifford in Washington (Zones 1 – 9). Sturgeon in the Canadian section of the
Transboundary Reach were captured by BC Hydro in the spring (May 20 – June 11) and
sturgeon in the United States section were captured by Colville Confederated Tribes in the fall
(September 4 – 11). Fish capture abided by capture protocols and set line configurations
commonly used for white sturgeon throughout the Columbia River (Golder 2006; Nigro et al.
1988). Four different hook sizes were used on each set line to facilitate the capture of various
sizes of white sturgeon (14/0, 16/0, 18/0, 20/0). Set lines were deployed at preselected sampling

69

locations across the sampling area. Sampling locations were determined by the physical
parameters (e.g. depth and velocity) of the site in relation to efficiency and safety of fish capture
(BC Hydro 2018a).
Juvenile white sturgeon from the conservation aquaculture program have been marked
with Passive Integrated Transponder (PIT) tags and released annually since 2001. Therefore, the
PIT tag number of captured sturgeon allowed us to conf irm their hatchery-origin and the age of
each fish prior to acoustic tagging. Acoustic tags were distributed across age, size, and river zone
(Figure 3.5). We balanced acoustic tag application within the age range of 9 – 18 years, as older
hatchery-origin sturgeon are reliably captured in the highest numbers during sampling
(Crossman et al. 2021). We attempted to distribute 100 tags evenly among river zones and
countries, however, only 25 of a possible 50 fish were tagged in the United States section of the
Transboundary Reach. The United States experienced lower collection efforts which focused on
the upper and lower river zones (i.e. Zones 6 and 9).
Sturgeon were measured for fork length (nearest 0.5 cm) and weight (nearest 0.02 kg).
The relative condition of each fish was calculated by dividing the observed weight of fish by the
expected weight given its measured fork length (Neff & Cargnelli 2004; Yard et al. 2015).
Expected weights were determined from a weight-length regression of all hatchery-origin
sturgeon collected during 2013 – 2019 stock assessments. Expected weights and relative
condition calculations were separated by country of capture, as sturgeon in the United States are
around 50% larger on average (Crossman et al. 2021). The sex of study fish was determined
during the surgical implantation of the acoustic tag. A 1.5 – 2.0 cm long incision was made
through the ventral body wall and adjacent to the sturgeon’s mid-line. An otoscope was inserted
to visually determine sex based on gonad tissue (Webb et al. 2019) and a biopsy of gonad tissue

70

was collected for histology to confirm sex and stage of maturity. Study fish were then implanted
with a uniquely-coded VEMCO V16-4x (69 kHz, 120-180 sec ping rate) acoustic tag. Incisions
were closed using half circle CP-2 reverse cutting-edge needles wedged to 2-0 polydioxanone
violet monofilament sutures (Ethicon, Z969H). After processing, sturgeon were returned to the
water at their location of capture. Procedures for capture, handling, and tagging of white
sturgeon were approved under a Species at Risk Act permit (permit number: 19-PPAC-00013)
and by the Animal Care and Use Committee of the University of Northern British Columbia
(protocol number: 2019-02).

Acoustic array
I used a pre-existing array of stationary VEMCO VR2W acoustic receivers managed by
BC Hydro in Canada and Spokane Tribal Fisheries and Colville Confederated Tribes in the
United States to detect the spatial and temporal movements of hatchery-origin white sturgeon
between May 21, 2019 – August 31, 2020. The array is dispersed at 3 – 10 rkm intervals across
the study area, covering all river zones and locations of interest (Figure 3.2). There were 31
acoustic receiver stations in Canada and 25 in the United States. Not all receivers were in
operation simultaneously in Canada. There were a total of eight receiver stations in Zone 2B
whose deployment was temporary and operated separately by Fortis BC and BC Hydro over the
study period. Additionally, Canada had multiple locations of interest (e.g. spawning areas,
international border, confluences) where more than one receiver was stationed within 0.5 rkm.
On average, the Canadian section of the Transboundary Reach had three receiver stations every
10 rkm and the United States section had two receiver stations every 10 rkm. The United States
likely has higher detection efficiency (DE) than Canada due to the prominence of slow-moving

71

reservoir habitat. Range testing performed at all receiver stations in the United States during
Summer 2015 and 2016 (~3,000 m3/s at the international border) demonstrated an average of
40% detection efficiency at distances of 165 – 330 m for receivers located upstream of Gifford,
Washington (Witte 2017). Range testing at eight receiver stations across the Canadian section of
the Transboundary Reach during Summer 2020 (~3,550 m3 /s at the international border)
achieved an average of 13% detection efficiency at similar distances.
Additionally, I used sentinel tags to assess detection efficiency in Canada over the entire
study period. Individual sentinel tags (range: 1 – 2 per river zone) were deployed at a distance
from 10 receiver stations across the reach to quantify temporal and spatial variation in detection
efficiency caused by environmental conditions (Gjelland et al. 2013; Kessel et al. 2014).
Deployment distances ranged from 140 – 360 m based on midpoint (50% detection efficiency)
range testing (Brownscombe et al. 2020) performed in Spring 2019 and feasibility for long-term
deployment. Sentinel tags were oriented at a horizontal angle near the river bottom (0.5 m) to
simulate typical tag orientation within sturgeon and sturgeon use of benthic habitats. Detection
efficiency was assessed by season and river zone (Figure 3.3). Mean detection efficiency for the
study period ranged from 41% – 64% among river zones, with Zone 5 having the highest
detection efficiency and Zone 3 having the lowest. Mean daily detection efficiency was highest
during the fall (72%) and lowest during the summer in both 2019 (40%) and 2020 (14%). Very
low detection efficiency during Summer 2020 was likely due to higher discharge during the
study period (Figure 3.4). Zone 1 performed the best during Summer 2020, demonstrating higher
mean daily detection efficiency (70%) than the other river zones (5%). This is likely due to Zone
1 discharge (HLK) being lower than river zones downstream with more water input sources.

72

Data filtering
Acoustic systems often generate erroneous detections (i.e. false positives) due to
environmental noise or acoustic tag collision (Simpfendorfer et al. 2015). Realism of movement
(<3 m/s) and frequency of detection (at least 120 secs between detections) approaches were
applied to identify potential erroneous detections and filter them from the dataset (Brownscombe
et al. 2019). In addition to filtering erroneous detections, these approaches highlighted the issue
of some receiver stations having overlapping ranges of detection in my dataset (i.e. fish being
detected at more than one receiver at the same time). To remedy this issue, detections from
receiver stations <1 rkm apart were aggregated to a single midpoint location. For receiver
stations >1 rkm apart, the station with the most detections within a 6-hour time interval
(described later) was determined the closest to the fish.

Movement behaviour analysis
I used hidden Markov models (HMMs; R package momentuHMM) to conduct the
analyses of movement behaviour (McClintock et al. 2018). HMMs are statistical models which
describe sequences of observable events that depend on unobservable or “hidden” discrete states
(McClintock et al. 2020). HMMs use the probability distribution of movement metrics,
computed from sequential animal locations, to decompose movement tracks into distinct hidden
states (assumed to represent different behaviour states of an animal) (Jonsen et al. 2005). This
study used speed (m/h; Gamma distribution with zero-inflation) and direction
(directed/undirected; Binomial distribution) as the movement metrics for a two-state HMM.
Direction was determined by whether a fish continued movement in an upstream or dow nstream
direction (directed) or exhibited a reversal or halt in movement direction (undirected). Speed was

73

determined by dividing the distance traveled (meters) by the time between positions (hours).
Selecting the appropriate number of behaviour states for a HMM can be challenging (Pohle et al.
2017). Based on the general movement tendencies described for white sturgeon as a species and
in the Transboundary Reach (Hildebrand et al. 2016; BC Hydro 2018a), a two-state model was
determined to be the best fit for the dataset. The two behaviour states identified by the HMM fit
to the white sturgeon dataset were characterized by 1) low speeds with undirected movement
(hereafter, residential behaviour) and 2) higher speeds with more directed movement (hereafter,
transitory behaviour).
HMMs are well suited to analyze movement datasets that are 1) recorded at regular time
intervals and 2) recorded with negligible positional error (Patterson et al. 2017). Fish detections
in my study often occurred on an irregular time interval due to fluctuations in imperfect detection
efficiency. To regularize our telemetry data, fish movement tracks were first split into unique
bouts if time between consecutive positions was greater than 96 hours (four days). Utilizing
bouts in this way helps define large gaps in detection as periods in which fish location cannot be
determined (Cote et al. 2020). Furthermore, bouts with less than four positions were removed to
avoid problems with model convergence (Bacheler et al. 2019). Remaining bouts were then
interpolated at a regular 6-hour time interval using the crawlWrap function in the R package
momentuHMM, which implements a continuous-time correlated random walk model to predict
locations at the selected time interval (McClintock et al. 2018). Positional error was not
accounted for in the crawlWrap function, as positions were determined from receiver stations
with defined GPS coordinates and erroneous detections were removed in my data filtering
process. Due to crawlWrap being unable to interpolate within the constraints of the river
boundary, positions were predicted on a linear simulation of the Transboundary Reach and

74

snapped to the nearest 0.5 rkm. This simulation decreased distorted calculations of speed caused
by natural bends in the river system.
After movement interpolation, HMMs were fit in momentuHMM. HMMs must be
provided starting parameters to estimate behaviour states. These starting parameters describe the
movement metrics of each state, which in my case would be represented by mean speed with a
zero mass parameter (i.e. proportion of speed values at 0) and the proportion of movement being
directed (Michelot et al. 2016). Starting parameters for this study were determined using the
getPar function in momentuHMM. This function helps estimate the best starting parameters for
the behaviour states when fitting an HMM to a dataset. Once a behaviour state was estimated for
each position using hidden Markov modelling, I assessed the probability of fish expressing
residential behaviour given the behaviour state at time t-1 using generalized linear mixed models
(GLMMs) with a binomial distribution and logit link function. Template Model Builder (TMB,
Kristensen et al. 2016) via the R package glmmTMB (Brooks et al. 2017) was used to perform
all GLMMs. Analyses in this study focused on the probability of residential behaviour, however,
it is important to note that the probability of transitory behaviour would be the complement to
these results.
Candidate models included combinations of three individual (relative condition, age, sex)
and four environmental (water temperature, discharge, season, river zone) covariates. Due to its
correlation with water temperature and discharge, season was not included in any candidate
models with these covariates. A random intercept by bout nested within individual was included
in all models. Relative condition was fixed at the value recorded during time of capture, age
increased at 0.5 year increments over the study period, and water temperature and discharge
represented mean values at the location (environmental monitoring zone) and time interval (6-

75

hour) of positioning. Season was divided into winter (December – February), spring (March –
May), summer (June – August), and fall (September – November).

Maximum displacement analysis
Maximum displacement analyses were performed with generalized linear mixed models
(GLMMs) using a gamma hurdle distribution with log link function. Template Model Builder
(TMB, Kristensen et al. 2016) via the R package glmmTMB (Brooks et al. 2017) was used to
perform all GLMMs. Maximum displacement was calculated as the distance (rkm) between the
highest and lowest position of a fish by country and seasonal period of each year. The filtered
dataset from the movement behaviour analysis was used to calculate these values. Candidate
models included combinations of three individual (relative condition, age, sex) and four
environmental (water temperature, discharge, season, country) covariates. Season was divided
into four periods as described earlier, relative condition was fixed at the value recorded during
time of capture, age increased to the nearest 0.5 year during a seasonal period, and water
temperature and discharge represented mean values at the location (country) and time interval
(season) of displacement. Due to its correlation with water temperature and discharge, season
was not included in any candidate models with these covariates. A random intercept by
individual was included in all models.

Model selection
I used bias-corrected Akaike Information Criterion (AIC c) scores to rank and select
candidate models. AIC is useful in model selection in that it estimates the relative quality of a
model within the candidate set for explaining the given data (McElreath 2020), with the lowest

76

AIC representing the best or top model. Top models were compared against the null model
(intercept and state at time t-1 variable) for validation. Due to model selection uncertainty
(Burnham et al. 2004), models included in a set defined by a cumulative weight AIC c residing
just at or below 0.95 were used for model averaged predictions. Model averaging was performed
using the R package MuMIn (Bartoń et al. 2020), and all analyses in this study were performed
in R (R Core Team 2021).

Results
Out of the 75 tagged hatchery-origin white sturgeon, 70 were included in my analyses.
Three fish were never detected and two fish did not meet the minimum requirements I identified
for the HMM (bouts of at least four positions). The final dataset resulted in a total of 250 bouts,
with an average of four bouts per fish (range: 1 – 10) and 40 days per bout (range: 1 – 432 days).
Fish were detected for an average of 144 days (range: 3 – 432 days), or 32% (range: 0.7% –
96%) of the study period.
Fish presence varied by season and river zone. More fish went undetected in the winter
and spring than the summer and fall (Figure 3.6). When fish were detected, Zone 1 had the most
fish presence on average, followed by Zone 2 and Zone 5 (Figure 3.7). Sturgeon spent an
average of 82% of their time in a single river zone (range: 32 – 100%) and occupied an average
of two river zones (range: 1 – 5) during the study period. Cross-zone movements occurred most
frequently in the summer and fall between Zones 1 – 2 and 8 – 9 (Figure 3.8). A total of three
fish moved across the Canada–United States international border, two of which were initially
tagged in Zone 5 and the other in Zone 6 (border adjacent zones).

77

A total of 27 fish resided for at least 1 day in a spawning area during the spawning
seasons (June 1 – July 31) of 2019 and 2020, spending an average of 18 days (range: 1.50 –
60.25 days) in a spawning area. Around 44% of those fish were female and 56% were male, with
an average age of 15 years (range: 9 – 19 years). Of these fish, five (18.5%) visited spawning
areas across years and all selected the same spawning area from the prior year. On average, fish
spent 52% of their time in a given year (range: 1 – 100%) residing within the same spawning
area, with an average maximum displacement of 13.8 rkm (range: 0 – 89 rkm) from the
spawning area. Only one fish (a 10-year old male) exhibited a cross-border movement and
resided in a spawning area outside their country of initial tagging.

Movement behaviour analysis
Hatchery-origin sturgeon spent 95% of their time expressing residential behaviour and
5% expressing transitory behaviour. Mean speed (m/h) for transitory behaviour was 343.4 (±
270.0) with a zero mass parameter of 0.18 (i.e. 18% of speed values were 0 m/h). Transitory
behaviour consisted of 44% directed movement and 56% undirected movement. Residential
behaviour had a zero mass parameter of 1, representing speed values of 0 m/h or continuous
residence at a single receiver station. Consequently, the proportion of directed movement for
residential behaviour was zero, representing no directed movement across receivers in the array.
Residential behaviour was therefore characterized as localized and undirected movement at a
single receiver, and transitory behaviour as larger movements between receivers that were both
directed and undirected.
The top model ranked by AIC c predicting the probability of residential behaviour given
the behaviour state of a fish at time t-1 (i.e. 6-hours before) included water temperature, relative

78

condition, sex, and a discharge and river zone interaction as covariates (Table 3.1). Five other
candidate models also closely explained the dataset and met criteria for inclusion in a 95%
confidence set for the best model. Consequently, these candidate models in addition to the top
model were included in model averaged predictions. All covariates except season appeared in at
least one of the averaged models. Water temperature and the discharge and river zone interaction
were included in all averaged models. My results demonstrated that the behaviour state of
hatchery-origin sturgeon varied considerably depending on the prior state. When having
expressed residential behaviour at time t-1, fish were very likely to express residential behaviour
at time t. However, when transitory behaviour was expressed at time t-1, the probability of
residential behaviour at time t depended more strongly on the values of other covariates.
Averaged model predictions saw the discharge and river zone interaction as having the
greatest influence on residential behaviour given transitory behaviour at time t-1 (Figure 3.10).
Notably, increases in discharge led to large increases in the probability of residential behaviour
in Zones 3 – 6. Zones 2 and 10 also demonstrated increases in residential probability in relation
to increased discharge. Remaining river zones related to discharge having a negative effect on
the probability of residential behaviour, however these results were not statistically significant
(i.e. coefficient standard errors overlapping 0) (Table 3.1). Water temperature had the next
largest influence. Warming water temperatures related to decreased probabilities of residential
behaviour given transitory behaviour at time t-1 (Figure 3.11). The biological covariates of sex
and relative condition had less influence, with the probability of residential behaviour decreasing
slightly if fish were female or in better relative condition. Age was the least influential covariate.
There was minimal influence on the probability of residential behaviour as fish got older.

79

Maximum displacement analysis
Maximum displacement of hatchery-origin white sturgeon varied across time and space.
On average, fish traveled farther during the summer and fall compared to the winter and spring
(Figure 3.12), and fish in the United States traveled three times farther than fish in Canada (39
rkm, 13 rkm). The top model ranked by AIC c predicting maximum displacement included water
temperature, age, and a country and discharge interaction as covariates (Table 3.2). Twelve other
candidate models including also closely explained the dataset and met criteria for inclusion in a
95% confidence set for the best model. Consequently, these 12 candidate models in addition to
the top model were included in model averaged predictions. All possible covariates appeared in
at least one of the averaged models. Country and time-varying environmental covariates (i.e.
season or water temperature and discharge) were included in all top-ranked models.
The country and discharge interaction was a strong predictor for the probability of white
sturgeon moving (i.e. binomial) and the extent of movement (i.e. gamma) when they did decide
to move (Figure 3.13). While fish were less likely to move in the United States compared to
Canada, averaged model predictions demonstrated higher maximum displacement in the United
States when fish did move (Figure 3.14). Increases in discharge led to a decreased likelihood of
movement and extent of movement, with maximum displacement in Canada decreasing more
rapidly than that in the United States. Warming water temperatures related to decreased
probabilities of movement but increased maximum displacement when fish did decide to move
(Figure 3.15). The covariates of season, sex, age, and relative condition were less inf luential and
not statistically significant (i.e. binomial and gamma coefficient standard errors overlapping 0)
(Figure 3.13 and 3.16). Averaged model predictions resulted in summer and fall having slightly
higher maximum displacement compared to winter and spring. Male hatchery-origin sturgeon

80

were predicted to displace slightly less than females. Fish that were older or in better relative
condition were also predicted to travel greater distances.

Discussion
Understanding the movement of threatened fish whose habitat range crosses international
boundaries is useful for the cooperative management and recovery of a species. In this study, I
used coarse-scale acoustic telemetry to track the movements of hatchery-origin white sturgeon
located within the Transboundary Reach of the Upper Columbia River. I identified distinct
residential and transitory patterns in movement behaviour, with the former representing localized
movement at a single receiver station and the latter representing directed and undirected
movements between receiver stations. Overall, hatchery-origin white sturgeon moved very little
and demonstrated strong site fidelity. Fish spent most of their time expressing residential
behaviour and residing in a single river zone. In my analyses, location (i.e. country, river zone),
discharge, and water temperature had the greatest influence on maximum displacement and
behaviour probability when fish were already in a transitory state. Sturgeon moving within the
United States, whose river zones can be characterized as more reservoir-like than riverine,
exhibited maximum displacements far greater than fish moving within Canada. Discharge also
interacted with fish location. Increases in discharge related to faster declines in maximum
displacement in Canada compared to the United States, as well as fish in more channelized river
zones having increased probabilities of expressing residential behaviour when previously in a
transitory state. These channelized river zones often resided near the international border and all
observed border-crossing fish were initially tagged from border adjacent zones. Warming water
temperatures were related to increases in maximum displacement and decreases in the

81

probability of residential behaviour when previously in a transitory state. Trends could be
observed seasonally, with warmer waters in the summer and fall being associated with increased
movement.
While movement ecology analyses can help us understand animal movement in terms of
biological and ecological processes, interpreting movement from coarse-scale acoustic telemetry
can be challenging. Acoustic telemetry arrays are imperfect and their performance in tracking
animal movement will depend on array design and environmental conditions. Array design,
including the number of receiver stations and how they are distributed throughout the study area,
influences animal detection and data resolution (Heupel et al. 2006). Larger distances between
receiver stations (e.g. receiver every 1 rkm versus 10 rkm) offer more space for animals and their
movements to go undetected. When animals are undetected, uncertainty arises on where an
animal is and what it is doing (Crossin et al. 2017). Environmental conditions, such as noise
caused from discharge or human activities, can also lead to error in positional accuracy (Kessel
et al. 2014). Increases in noise will have a negative effect on the detection range (i.e. distance
from receiver that tag can be detected) and detection efficiency (i.e. number of successful
detections of a tag out of possible detections) of an acoustic telemetry array. In this study, both
array design and environmental conditions appeared to influence fish detection. Periods of
sturgeon going undetected between receiver stations peaked during winter months (Figure 3.6).
Detection efficiency also declined during the summer as flows increased, especially in more
channelized river zones (Figures 3.3 and 3.4).
Even when fish are detected, intent of movement can be difficult to determine with
complete certainty. This is especially true for river systems, such as the UCR, where biological
and ecological processes (e.g. feeding, spawning, overwintering) overlap temporally and

82

spatially (Hildebrand et al. 2013). While incorporating empirical data (e.g. feeding or spawning
documentation) into movement analyses can benefit ecological interpretations, such data is often
difficult to acquire in swift and deep aquatic environments (Cooke et al. 2012). In this study,
these river system characteristics did limit the certainty of movement intent. Additionally, due to
acoustic telemetry array design and performance, the full extent of hatchery-origin sturgeon
movement could not be determined (e.g. missed detections of fish residing outside of receiver
range or moving during high flows). Recognizing the uncertainty in the data and modelling
process is essential for the honest and transparent communication of findings to recovery and
management agencies (Crossin et al. 2017). Such recognition does not immediately discount the
results of movement ecology analyses, but rather implores researchers to use their own intuition
and knowledge when interpreting the ecological meaningfulness of movement.
White sturgeon movement can be driven by multiple biological and ecological processes,
such as feeding, overwintering, and spawning (Hildebrand et al. 2016). Feed ing is necessary for
the growth, survival, and reproduction of an animal. Fish will move within and between habitats
to take advantage of feeding opportunities (Cooke et al. 2016). White sturgeon are primarily
benthic feeders, meaning they search out food on the river bottom. Depending on resource
availability and fish size, sturgeon will feed on a variety of invertebrates and other fish (McCabe
et al. 1993; Crossman et al. 2016; Parsley et al. 2010). While the intent of movement cannot be
determined with complete certainty, the timing of observed movements by hatchery-origin
sturgeon in this study do coincide with known sturgeon feeding ecology. In the Transboundary
Reach, movement activity related to feeding is thought to increase during warmer water
temperatures (i.e. summer and fall; Figure 3.4) and occur in low-velocity habitats where fish can
hold and feed on prey delivered by the river (Hildebrand & Parsley 2013). I observed that

83

increases in water temperature related to increased maximum displacement and reduced
probabilities of residential behaviour when previously in a transitory state (Figures 3.11 and
3.15). As water temperature is a driver of fish metabolism (Volkoff et al. 2020), these results
likely reflect how hatchery-origin sturgeon are increasing movement in response to metabolic
demands and feeding opportunities.
Spatially, maximum displacement and behaviour probabilities differed depending on fish
location and discharge. Hatchery-origin sturgeon moving within Canada displaced far less than
fish moving within the United States, with increases in discharge relating to faster declines in
maximum displacement in Canada compared to the United States (Figures 3.12 and 3.14).
Increases in discharge also related to sturgeon in more channelized river zones (e.g. Zones 3 and
4) having increased probabilities of expressing residential behaviour when previously in a
transitory state (Figure 3.10). These results may reflect differences in feeding habitat between
countries and river zones. While Canada consists of a more riverine environment with lowvelocity habitat interspersed between rapids, the United States is more reservoir-like with lowvelocity habitat covering entire river zones. Fish must make trade-offs when deciding between
allocating energy to movement or maintenance of other physiological functions such as growth
and reproduction (Rennie et al. 2005). Discharge has also been shown to alter the energetic
demands of fish movement, with greater energy required to move in higher flows (Shepard et al.
2013). Consequently, less energy expended moving in low-velocity water may be promoting the
greater displacement of hatchery-origin sturgeon in the United States to capitalize on feeding
habitat.
While water temperature, discharge, and fish location (i.e. country, river zone) were
influential in predicting movement, in general hatchery-origin sturgeon expressed strong site

84

fidelity. Sturgeon spent most of their time expressing residential behaviour and movement
typically did not extend beyond two river zones. These results support site fidelity predictions
from direct capture assessments of hatchery-origin sturgeon in the Transboundary Reach (BC
Hydro 2018a,b). It also supports the observed distribution and seasonal movements of wild adult
sturgeon in the Transboundary Reach who have been well documented since the 1990s through
the use of mark-recapture and acoustic telemetry (BC Hydro 2018a; Irvine et al. 2007;
Hildebrand et al. 1999). These assessments have shown that wild adults demonstrate high site
fidelity to deep and low water velocity areas, with individuals spending more than 60% of the
year at a single location and 90% of the year within a 10 rkm river section. Similar trends in site
fidelity have been observed in other sturgeon populations as well (Robichaud et al. 2017; Kynard
et al. 2016). I observed the least amount of movement in the winter and spring (Figure 3.11).
This reduction in movement is likely a reflection of overwintering, a biological process in which
fish activity levels and metabolic rates decline in response to cooling water temperatures
(Crawshaw 1984). Overwintering may also explain why larger numbers hatchery-origin sturgeon
went undetected during winter months (Figure 3.6), as overwintering in habitats between
receiver stations (i.e. outside of receiver range) would lead to long periods of no detection.
Spawning can also be a driver of movement in sturgeon. In this study, residence and
movement of hatchery-origin sturgeon to spawning areas was observed in 2019 and 2020.
Around 40% of the study population resided in a spawning area for at least one day during a
spawning period, and 7% were repeat visitors of a spawning area across years. While males were
of interest due to their earlier sexual maturity and greater likelihood of spawning activity, the sex
ratio of fish present in spawning areas was similar to the ratio of the entire study population. The
average age of both female and male hatchery-origin sturgeon occupying spawning areas was

85

around 15 years. This age is below average for expected female spawning in the UCR (~25
years) and on par for expected male spawning (~15 years), with the youngest age of a sexually
mature hatchery-origin surgeon in the UCR being a 9-year old male (Maskill et al. 2022,
unpublished). Although residence of hatchery-origin sturgeon in spawning areas suggests
possible spawning activity, especially by males, spawning areas in the Transboundary Reach can
also overlap with productive feeding habitat (Hildebrand et al. 2013). The presence of likely
sexually immature female hatchery-origin sturgeon in spawning areas might be representative of
feeding opportunity in spawning areas. Incorporating supporting independent data in future
studies, such as timestamped embryo collection and reproductive staging, could help better
determine spawning activity in relation to sturgeon movement (Buderman et al. 2021).
Long-distance migrations of hatchery-origin sturgeon to spawning areas were
uncommon. Most fish spent a majority of their time in a given year within a spawning area or
moved less than one river zone to reach a spawning area. In studies of other sturgeon
populations, spawning migrations have been observed in response to water temperature and
discharge. For lake sturgeon (Bruch & Binkowski 2002) and Kootenay River white sturgeon
(Paragamian & Kruse 2001), increases in water temperature during a spawning season was a
good predictor of female migration. Increases in discharge have been thought to cue spawning
migration (Forsythe et al. 2012; Dammerman et al. 2018), with increased movements by white
sturgeon following peak flows in the Kootenay River (Paragaminan et al. 1999). While spawning
activity could not be confirmed in this study, increases in water temperature, which occurred
during the summer spawning period (Figure 3.4), did relate with increased maximum
displacement and decreased probabilities of residential behaviour (when in a transitory state at
time t-1) (Figures 3.11 and 3.15). Discharge however, which also peaked during the summer

86

(Figure 3.4), related to decreased maximum displacement and increased probabilities of
residential behaviour (when in a transitory state at time t-1) in more channelized river zones
(Figures 3.10 and 3.14). While this might suggest that discharge is limiting the displacement of
hatchery-origin white sturgeon during spawning periods, my results are likely biased by
significant drops in detection efficiency during the summer in response to increased discharge
(Figure 3.3). I recommend that future research attempt to incorporate detection efficiency into
movement analyses, as it is possible that hatchery-origin sturgeon could be expressing greater
spawning movements, and movement in general, than observed in this study.
Although hatchery-origin sturgeon demonstrated high site fidelity overall, heterogeneity
in movement was present in the study population, with individual displacement ranging between
0 -89 rkm over the study period. Similar examples of heterogeneity in movement have been
observed in other sturgeon populations (Robichaud et al. 2017) and fish species (Moser et al.
2013; Rodríguez 2002). Variability in movement among individuals may be related to biological
factors such as age, relative condition, and sex. While biological covariates were not statistically
significant in predicting maximum displacement, my study observed decreases in the probability
of residential behaviour (when in a transitory state at time t-1) related to fish that were female or
in better relative condition (Figure 3.11). Age was the least influential predictor and there was
minimal influence on the probability of residential behaviour as fish got older. Other researchers
have observed that older lake sturgeon demonstrate greater maximum displacement (Trested et
al. 2011) and that relative condition was a good indicator of individual energetic state (Neff &
Cargnelli 2004). Outside of spawning migrations, sex has not been observed as a factor
influencing the movement of sturgeon. It is important to note that the duration of this study only
lasted 1.5 years. As individual sturgeon movement can differ across years, more time is likely

87

needed to better understand how biological as well as environmental variables influence
movement patterns and behaviour states.

88

Table 3.1. Models predicting the probability of residential behaviour given the behaviour state at
time t-1 ranked by increasing order of the bias-corrected Akaike Information Criterion (AIC c).
Differences in AIC c relative to the top model (ΔAIC c), weighted AIC c (wAIC c), cumulative
weighted AIC c (cwAIC c), log-likelihood (log(L)), and the number of parameters in a model (K)
are shown. The top six models were selected for averaged predictions.

zage, Age of fish; zdisc, Discharge; tm1, Probability of residential behaviour at time t-1; zcond,
Relative condition of fish; zone, River zone; seas, Season; sex, Sex of fish; ztemp, Water
temperature. All models include bout nested within individual as a random effect.

89

Table 3.2. Models predicting maximum displacement ranked by increasing order of the biascorrected Akaike Information Criterion (AIC c). Differences in AIC c relative to the top model
(ΔAICc), weighted AIC c (wAICc), cumulative weighted AIC c (cwAIC c), log-likelihood (log(L)),
and the number of parameters in a model (K) are shown. The top 13 models were selected for
averaged predictions.

zage, Age of fish; zdisc, Discharge; ctry, Country; zcond, Relative condition of fish; seas,
Season; sex, Sex of fish; ztemp, Water temperature. All models include individual as a random
effect.

90

Figure 3.1. Transboundary Reach of the Upper Columbia River from HLK in British Columbia,
Canada, through Gifford in Washington, USA. River kilometers of interest are labeled as white
dots.

91

Figure 3.2. Canadian section of the Transboundary Reach in British Columbia, Canada, (left) and
United States section of the Transboundary Reach in Washington, USA (right). River zones are
distinguished by grey and white sections of river. Dashed lines represent the extent of the six
environmental monitoring zones. White dots represent aggregated acoustic receiver locations
(some locations may contain more than one receiver). Yellow dots represent capture locations of
tagged fish. Shaded blue circles represent known white sturgeon spawning areas.

92

Figure 3.3. Detection efficiency (DE) from May 2019 – August 2020. A) Mean daily detection
efficiency by season and year. B) Mean daily detection efficiency by river zone. Solid blue lines
represent river zone mean and dashed blue lines represent river zone median for the study period.

93

Figure 3.4. Mean daily water temperature (A) and discharge (B) from May 2019 – August 2020.
Line type represents gauges at Hugh L. Keenleyside Dam (rkm 1261.2) and Border (rkm
1204.2).

94

Figure 3.5. Tagged hatchery-origin white sturgeon used in study. A) Fork length and weight of
tagged fish, with shades of gray describing sex, shape describing country of capture, and line
type representing expected values based on country of capture (relative condition of 0). B)
Number of fish tagged by age and sex. C) Number of fish tagged by habitat zone of capture and
sex. D) Boxplot of fish tagged by age and habitat zone of capture. Dashed line represents the
border between countries.

95

Figure 3.6. Distribution of tagged hatchery-origin white sturgeon by month and river zone
between May 2019 – August 2020. Fish were counted for the river zone in which they spent the
longest duration of time in a month. Last box represents the number of fish not detected during a
month.

96

Figure 3.7. Daily number of tagged hatchery-origin white sturgeon in each river zone between
May 2019 – August 2020. Dashed line represents the border between countries.

97

Figure 3.8. Presence of tagged hatchery-origin white sturgeon between May 2019 – August
2020. Presence indicated by black lines with instances of movement across zones highlighted by
points. Point colour represents river zone fish are crossing into and shape represents country
where the river zone is located. Crosses represent date and river zone of fish release after
tagging.

98

Figure 3.9. Model averaged coefficients and standard errors (SE) of the binomial generalized
linear mixed model parameters predicting the probability of residential behaviour given the
behaviour state at time t-1. Model averaged intercept estimate of -0.50 (± 0.34 SE).

99

3.10. Model averaged predictions of the probability of being in the residential behaviour state by
river zone, discharge, and the behaviour state at time t-1. Shaded areas denote 95% confidence
interval. To isolate the effect, other numeric model covariates were fixed at their mean and the
categorical covariate sex was set to Female.

100

Figure 3.11. Model averaged predictions of the effect of water temperature (A), sex (B), age (C),
and relative condition (D) on the probability of being in the residential behaviour state.
Behaviour state at time t-1 represented by colour with 95% confidence interval (shaded area and
whiskers). To isolate the effect, other numeric model covariates were fixed at their mean and the
categorical covariates sex and river zone were set to Female and Zone 1 respectively.

101

Figure 3.12. Maximum displacement (rkm) of tagged hatchery-origin white sturgeon by season
and country where movement took place between May 2019 – August 2020.

102

Figure 3.13. Model averaged coefficients and standard errors (SE) of the gamma hurdle
generalized linear mixed model parameters predicting maximum displacement. Results are
separated by binomial and gamma components of the gamma hurdle model. Model averaged
intercept estimates are -0.75 (± 0.52 SE) and 1.58 (± 0.42 SE) respectively.

103

Figure 3.14. Model averaged predictions of the effect of country (A) and discharge and country
(B) on maximum displacement (rkm) with 95% confidence interval (shaded area and whiskers).
To isolate effects, other model covariates were fixed at their mean. Categorical covariates of sex
and season were set to Female and Summer respectively.

104

Figure 3.15. Model averaged predictions of the effect of water temperature on maximum
displacement (rkm) with 95% confidence interval (shaded area). To isolate effects, other model
covariates were fixed at their mean. Categorical covariates of sex, season, and country were set
to Female, Summer, and USA respectively.

105

Figure. 3.16. Model averaged predictions of the effect of season (A), sex (B), age (C), and
relative condition (D) on maximum displacement (rkm) with 95% confidence interval (shaded
area and whiskers). To isolate effects, other model covariates were fixed at their mean.
Categorical covariates of sex, season, and country were set to Female, Summer, and USA
respectively.

106

References
Allen AM, Singh NJ (2016) Linking movement ecology with wildlife management and
conservation. Frontiers Ecol Evol 3:155. https://doi.org/10.3389/fevo.2015.00155
Balazik M, Barber M, Altman S, Reine K, Katzenmeyer A, Bunch A, Garman G (2020)
Dredging activity and associated sound have negligible effects on adult Atlantic sturgeon
migration to spawning habitat in a large coastal river. Plos One 15:e0230029.
https://doi.org/10.1371/journal.pone.0230029
Bartoń K (2020) MuMIn: multi-model inference. R package version 1.43.17. https://CRAN.Rproject.org/package=MuMIn
BC Hydro (2018a) Lower Columbia River Adult White Sturgeon Monitoring Program
(CLBMON-28). Year 10 Data Report. Report by BC Hydro, Castlegar, BC.
BC Hydro (2018b) Lower Columbia River Juvenile Detection Program (CLBMON-29). Year 10
Data Report. Report by BC Hydro, Castlegar, BC.
BC Hydro (2020) Lower Columbia River Adult White Sturgeon Monitoring Program
(CLBMON-28). Year 12 Data Report. Report by BC Hydro, Castlegar, BC.
BC Hydro (2021) Arrow Operations Summary.
Brooks ME, Kristensen K, Van Benthem KJ, Magnusson A, Berg CW, Nielsen A, Skaug HJ,
Machler M, Bolker BM (2017) glmmTMB balances speed and flexibility among packages for
zero-inflated generalized linear mixed modeling. R J 9:378-400. https://doi.org/10.3929/ethz-b000240890
Brownscombe JW, Lédée EJI, Raby GD, Struthers DP, Gutowsky LFG, Nguyen VM, Young N,
Stokesbury MJW, Holbrook CM, Brenden TO, Vandergoot CS, Murchie KJ, Whoriskey K,
Flemming JM, Kessel ST, Krueger CC, Cooke SJ (2019) Conducting and interpreting fish
telemetry studies: considerations for researchers and resource managers. Rev Fish Biol Fisher
29:369-400. https://doi.org/10.1007/s11160-019-09560-4
Brownscombe JW, Griffin LP, Chapman JM, Morley D, Acosta A, Crossin GT, Iverson SJ,
Adams AJ, Cooke SJ, Danylchuk AJ (2020) A practical method to account for variation in
detection range in acoustic telemetry arrays to accurately quantify the spatial ecology of aquatic
animals. Methods Ecol Evol 11:82-94. https://doi.org/10.1111/2041-210X.13322
Bruch RM, Binkowski FP (2002) Spawning behaviour of lake sturgeon (Acipenser fulvescens). J
Appl Ichthyol 18:570-579. https://doi.org/10.1046/j.1439-0426.2002.00421.x

107

Buderman FE, Gingery TM, Diefenbach DR, Gigliotti LC, Begley-Miller D, McDill MM,
Wallingford BD, Rosenberry CS, Drohan PJ (2021) Caution is warranted when using animal
space-use and movement to infer behavioural states. Mov Ecol 9:30.
https://doi.org/10.1186/s40462-021-00264-8
Burnham KP, Anderson DR (2004) Multimodel inference: understanding AIC and BIC in model
selection. Sociol Methods Res 33:261-304. https://doi.org/10.1177/0049124104268644
Cooke S, Paukert C, Hogan Z (2012) Endangered river fish: factors hindering conservation and
restoration. Endanger Species Res 17:179-191. https://doi.org/10.3354/esr00426
Cote D, Morris CJ, Regular PM, Piersiak MG (2020) Effects of 2D seismic on snow crab
movement behaviour. Fish Res 230:105661. https://doi.org/10.1016/j.fishres.2020.105661
Crawshaw LI (1984) Low temperature dormancy in fish. Am J Physiol 246:479-486.
https://doi.org/10.1152/ajpregu.1984.246.4.R479
Crossin GT, Heupel MR, Holbrook CM, Hussey NE, Lowerre-Barbieri SK, Nguyen VM, Raby
GD, Cooke SJ (2017) Acoustic telemetry and fisheries management. Ecol Appl 27:1031-1049.
https://doi.org/10.1002/eap.1533
Crossman JA, Jay KJ, Hildebrand LR (2016) Describing the diet of juvenile white sturgeon in
the Upper Columbia River Canada with lethal and nonlethal methods. N Am J Fish Manag
36:421-432. https://doi.org/10.1080/02755947.2015.1125976
Crossman JA, Korman J (2021) Evaluating the risk of family representation on the long-term
recovery of white sturgeon in the Transboundary Reach of the Columbia River. Report submitted
to the Department of Fisheries and Oceans, Canada.
Dammerman KJ, Webb M, Scribner KT (2019) Riverine characteristics and adult demography
influence female lake sturgeon (Acipenser fulvescens) spawning behaviour, reproductive
success, and ovarian quality. Can J Fish Aquat Sci 76: 1147-1160. https://doi.org/10.1139/cjfas2018-0141
Dudgeon D, Arthington AH, Gessner MO, Kawabata ZI, Knowler DJ, Lévêque C, Naiman RJ,
Prieur-Richard AH, Soto D, Stiassny MLJ, Sullivan CA (2006) Freshwater biodiversity:
importance, threats, status and conservation challenges. Biol Rev 81:163-182.
https://doi.org/10.1017/s1464793105006950
Fisheries and Oceans Canada (2014) Recovery strategy for white sturgeon (Acipenser
transmontanus) in Canada. In Species at Risk Act Recovery Strategy Series. Fisheries and
Oceans Canada, Ottawa, ON.

108

Forsythe PS, Scribner KT, Crossman JA, Ragavendran A, Baker EA, Davis C, Smith KK (2012)
Environmental and lunar cues are predictive of the timing of river entry and spawning-site arrival
in lake sturgeon Acipenser fulvescens. J Fish Biol 81:35-53. https://doi.org/10.1111/j.10958649.2012.03308.x
Gjelland KO, Hedger RD (2013) Environmental influence on transmitter detection probability in
biotelemetry: developing a general model of acoustic transmission. Methods Ecol Evol 4:665674. https://doi.org/10.1111/2041-210X.12057
Golder Associates Ltd (2006) Upper Columbia White Sturgeon Stock Monitoring and Data
Management Program: Synthesis Report. Report prepared for British Columbia Ministry of
Environment, Nelson, BC.
Heupel MR, Semmens JM, Hobday AJ (2006) Automated acoustic tracking of aquatic animals:
scales, design and deployment of listening station arrays. Mar Freshw Res 57:1-13.
http://dx.doi.org/10.1071/MF05091
Hildebrand L, McLeod C, McKenzie S (1999) Status and management of White Sturgeon in the
Columbia River in British Columbia, Canada: An Overview. J Appl Ichthyol 15:164-172.
https://doi.org/10.1111/j.1439-0426.1999.tb00227.x
Hildebrand LR, Parsley M (2013) Upper Columbia White Sturgeon Recovery Plan - 2012
Revision. Prepared for the Upper Columbia White Sturgeon Recovery Initiative.
Hildebrand LR, Schreier AD, Lepla K, McAdam SO, McLellan J, Parsley MJ, Paragamian VL,
Young SP (2016) Status of white sturgeon (Acipenser transmontanus Richardson 1863)
throughout the species range, threats to survival, and prognosis for the future. J Appl Ichthyol
32:261-312. https://doi.org/10.1111/jai.13243
Hussey NE, Kessel ST, Aarestrup K, Cooke SJ, Cowley PD, Fisk AT, Harcourt RG, Holland
KN, Iverson SJ, Kocik JF, Flemming JEM, Whoriskey FG (2015) Aquatic animal telemetry: a
panoramic window into the underwater world. Science 348:1255642.
https://doi.org/10.1126/science.1255642
Irvine RL, Schmidt DC, Hildebrand LR (2007) Population Status of White Sturgeon in the
Lower Columbia River within Canada. Trans Am Fish Soc 136:1472–1479.
https://doi.org/10.1577/T06-190.1
Johnson DS, London JM, Lea MA, Durban JW (2008). Continuous-time correlated random walk
model for animal telemetry data. Ecology 89:1208–1215. https://doi.org/10.1890/07-1032.1
Jonsen ID, Flemming JM, Myers RA (2005) Robust state-space modeling of animal movement
data. Ecology 86:2874-2880. https://doi.org/10.1890/04-1852

109

Joo R, Boone ME, Clay TA, Patrick SC, Clusella‐Trullas S, Basille M (2019) Navigating
through the r packages for movement. J Anim Ecol 89:248-267. https://doi.org/10.1111/13652656.13116
Kessel ST, Cooke SJ, Heupel MR, Hussey NE, Simpfendorfer CA, Vagle S, Fisk AT (2014) A
review of detection range testing in aquatic passive acoustic telemetry studies. Rev Fish Biol
Fisher 24:199-218. https://doi.org/10.1007/s11160-013-9328-4
Kristensen K, Nielsen A, Berg CW, Skaug H, Bell BM (2016) TMB: automatic differentiation
and Laplace approximation. J Stat Softw 70:1-21. https://doi.org/10.18637/jss.v070.i05
Kynard B, Bolden S, Kieffer M, Collins M, Brundage H, Hilton EJ, Litvak M, Kinnison MT,
King T, Peterson D (2016) Life history and status of Shortnose Sturgeon (Acipenser
brevirostrum LeSueur, 1818). J Appl Ichthyol 32:208-48. https://doi.org/10.1111/jai.13244
Lennox RJ, Aarestrup K, Cooke SJ, Cowley PD, Deng ZD, Fisk AT, Harcourt RG, Heupel M,
Hinch SG, Holland KN, Hussey NE, Iverson SJ, Kessel ST, Kocik JF, Lucas MC, Flemming JM,
Nguyen VM, Stokesbury MJW, Vagle S, VanderZwaag DL, Whoriskey FG, Young N (2017)
Envisioning the future of aquatic animal tracking: technology, science, and application.
Bioscience 67:884-896. https://doi.org/10.1093/biosci/bix138
Limburg KE, Waldman JR (2009) Dramatic declines in north Atlantic diadromous fishes.
BioScience 59:955-965. https://doi.org/10.1525/bio.2009.59.11.7
Makowski D, Ben-Shachar MS, Chen SHA, Lüdecke D (2019) Indices of effect existence and
significance in the Bayesian framework. Front Physiol 10:2767.
https://doi.org/10.3389/fpsyg.2019.02767
McCabe GT, Emmett RL, Hinton SA (1993) Feeding ecology of juvenile white sturgeon
(Acipenser transmontanus) in the Lower Columbia River. Report O. In: Beamesderfer RC, Nigro
AA (ed) Status and habitat requirements of the white sturgeon populations in the Columbia River
downstream from McNary Dam, 2nd vol. Final Report to Bonneville Power Administration,
Portland, OR, pp 245-263
McClintock BT, Michelot T (2018) momentuHMM: R package for generalized hidden Markov
models of animal movement. Methods Ecol Evol 9:1518-1530. https://doi.org/10.1111/2041210X.12995
McClintock BT, Langrock R, Gimenez O, Cam E, Borchers DL, Glennie R, Patterson TA (2020)
Uncovering ecological state dynamics with hidden Markov models. Ecol Lett 23:1878-1903.
https://doi.org/10.1111/ele.13610
McElreath, R (2020) Statistical rethinking: a Bayesian course with examples in R and Stan.
Chapman and Hall/CRC. https://doi.org/10.1201/9780429029608

110

Michelot T, Langrock R, Patterson TA (2016) moveHMM: an R package for the statistical
modelling of animal movement data using hidden Markov models. Methods Ecol Evol 7:13081315. https://doi.org/10.1111/2041-210x.12578
Morales JM, Haydon DT, Frair J, Holsinger KE, Fryxell JM (2004) Extracting more out of
relocation data: building movement models as mixtures of random walks. Ecology 85:436-2445.
https://doi.org/10.1890/03-0269
Moser ML, Keefer ML, Caudill CC, Burke BJ (2013) Migratory behaviour of adult Pacific
Lamprey and evidence for effects of individual temperament on migration rate. In: Ueda H,
Tsukamoto K (ed) Physiology and ecology of fish migration. CRC Press, Florida, pp 130-149
Mueller T, Fagan WF (2008) Search and navigation in dynamic environments – from individual
behaviours to population distributions. Oikos 117:654-664. https://doi.org/10.1111/j.00301299.2008.16291.x
Nathan R, Getz WM, Revilla E, Holyoak M, Kadmon R, Saltz D, Smouse PE (2008) A
movement ecology paradigm for unifying organismal movement research. Proc Natl Acad Sci
USA 105:19052-9. https://doi.org/10.1073/pnas.0800375105
Nigro AA, Rieman BE, Elliot JC, Engle DR (1988) Status and habitat requirements of White
Sturgeon populations in the Columbia River downstream from McNary Dam. Annual Progress
Report to Bonneville Power Administration, Portland, OR.
Neff B, Cargnelli L (2004) Relationships between condition factors, parasite load and paternity
in bluegill sunfish, Lepomis macrochirus. Environ Biol Fish 71:297-304.
https://doi.org/10.1007/s10641-004-1263-8
Neufeld MD, Rust PJ (2009) Using passive sonic telemetry methods to evaluate dispersal and
subsequent movements of hatchery-reared white sturgeon in the Kootenay River. J Appl Ichthyol
25:27-33. https://doi.org/10.1111/j.1439-0426.2009.01336.x
Palacios-Abrantes J, Sumaila UR, Cheung WWL (2020) Challenges to transboundary fisheries
management in North America under climate change. Ecol Soc 25:41.
https://doi.org/10.5751/ES-11743-250441
Paragamian VL, Kruse G, Wakkinen VD (1999) Kootenai River white sturgeon investigations.
Annual Progress Report of Idaho Department of Fish and Game to Bonneville Power
Administration, Boise, ID.
Paragamian VL, Kruse G, Wakkinen VD (2001) Spawning habitat of Kootenay River white
sturgeon, post-Libby Dam. N Am J Fish Manag 21:22-33. https://doi.org/10.1577/15488675(2001)021<0022:SHOKRW>2.0.CO;2

111

Parsley MJ, van der Leeuw BK, Elliott DG (2010) Characterization of the contents and histology
of the gastrointestinal tracts of white sturgeon (Acipenser transmontanus) captured from Upper
Lake Roosevelt, Washington, October 2008. U.S. Geological Survey Open-File Report 20101193. https://doi.org/10.3133/ofr20101193
Patterson TA, Parton A, Langrock R, Blackwell PG, Thomas L, King R (2017) Statistical
modelling of individual animal movement: an overview of key methods and a discussion of
practical challenges. AStA Adv Stat Anal 101:399-438. https://doi.org/10.1007/s10182-0170302-7
R Core Team (2021) R: a language and environment for statistical computing. R Foundation,
Vienna, Austria. https://www.R-project.org
Rennie MD, Collins NC, Shuter BJ, Rajotte JW, Couture P (2005) A comparison of methods for
estimating activity costs of wild fish populations: more active fish observed to grow slower. Can
J Fish Aquat Sci 62:767-780. https://doi.org/10.1139/f05-052
Rodríguez MA (2002) Restricted movement in stream fish: the paradigm is incomplete, not lost.
Ecology 83:1-13. https://doi.org/10.2307/2680115
Robichaud D, English KK, Nelson TC (2017) Annual movements of acoustic-tagged white
sturgeon in the lower reaches of the Fraser River and its tributaries. T Am Fish Soc 146:611-625.
https://doi.org/10.1080/00028487.2017.1294542
Runge CA, Martin TG, Possingham HP, Willis SG, Fuller RA (2014) Conserving mobile
species. Front Ecol Environ 12:395-402. https://doi.org/10.1890/130237
Scott WB, Crossman EJ (1973) Freshwater fishes of Canada, Bulletin 184. Fisheries Research
Board of Canada, Department of Fisheries and Oceans, Scientific Information and Publications
Branch, Ottawa, ON.
Shaw AK (2020) Causes and consequences of individual variation in animal movement. Mov
Ecol 8:12. https://doi.org/10.1186/s40462-020-0197-x
Shepard ELC, Wilson RP, Rees WG, Grundy E, Lambertucci SA, Vosper SB (2013) Energy
Landscapes Shape Animal Movement Ecology. Am Nat 182:298-312.
https://doi.org/10.1086/671257
Simpfendorfer CA, Huveneers C, Steckenreuter A, Tattersall K, Hoenner X, Harcourt R, Heupel
MR (2015) Ghosts in the data: false detections in VEMCO pulse position modulation acoustic
telemetry monitoring equipment. Anim Biotelemetry. 31:482-492.
https://doi.org/10.1186/s40317-015-0094-z

112

Trested DG, Chan MD, Bridges WC, Isely JJ (2011) Seasonal movement and mesohabitat usage
of adult and juvenile lake sturgeon in the Grasse River, New York. T Am Fish Soc 140:10061014. https://doi.org/10.1080/00028487.2011.603981
Valbo-Jørgensen J, Marmulla G, Welcomme RL (2008) Migratory fish stocks in transboundary
basins — implications for governance, management and research. In: Lagutov V (ed) Rescue of
Sturgeon Species in the Ural River Basin. Springer, Dordrecht, pp 61-86.
https://doi.org/10.1007/978-1-4020-8924-4_5
Webb MAH, Van Eenennaam JP, Crossman JA, Chapman FA (2019) A practical guide for
assigning sex and stage of maturity in sturgeons and paddlefish. J Appl Ichthyol 35:169-186.
https://doi.org/10.1111/jai.13582
Whoriskey K, Martins EG, Auger‐Méthé M, Gutowsky LFG, Lennox RJ, Cooke SJ, Power M,
Flemming JM (2019) Current and emerging statistical techniques for aquatic telemetry data: a
guide to analyzing spatially discrete animal detections. Methods Ecol Evol 10:935-948.
https://doi.org/10.1111/2041-210X.13188
Witte BT (2017) Acoustic tracking of redband rainbow trout (Oncorhynchus myskiss gairdneri)
in Lake Roosevelt (2015-2016), evaluation of an acoustic receiver array, and low-voltage
electroanesthesia for tag implantation. EWU Masters Thesis Collection 433.
http://dc.ewu.edu/theses/433
Yard MD, Korman J, Walters CJ, Kennedy TA (2015) Seasonal and spatial patterns of growth of
rainbow trout in the Colorado River in Grand Canyon, Arizona. Can J Fish Aquat Sci 73:125139. https://doi.org/10.1139/cjfas-2015-0102

113

CHAPTER 4

CONCLUSION

The goal of this study was to investigate the movement patterns and behaviour states of
white sturgeon who occupy a regulated and transboundary section of the Upper Columbia River.
Acoustic telemetry was utilized to capture movement and behaviour at two different spatial
scales. In Chapter 2, an overlapping acoustic telemetry array covering 2 rkm was implemented to
track the fine-scale movements of wild adult white sturgeon immediately downstream of a
hydroelectric operation on the UCR. In Chapter 3, an acoustic telemetry array covering 296 rkm
was implemented to track the coarse-scale movements of hatchery-origin white sturgeon residing
throughout the Transboundary Reach of the UCR. Hidden Markov models (HMMs) and
generalized linear mixed models (GLMMs) were then applied in both chapters to identify
ecologically meaningful movement behaviours and determine how movement and behaviour is
influenced by biological factors and environmental conditions (e.g. water temperature and
discharge) that are frequently altered by river regulation. Results from this study described in
detail white sturgeon interactions with dams and river regulation and will help inform recovery
measures and management decisions for this threatened species.
Chapter 2 identified two distinct behaviour states from the fine-scale movement dataset
of wild adult white sturgeon occupying critical habitat immediately below a hydroelectric
operation. These states consisted of a residential and transitory behaviour, with the former
representing short movements with more persistent direction and the latter representing longer
movements with more frequent turns. The influence of biological factors and environmental

114

conditions on behaviour was assessed using two response variables: 1) the probability of fish
expressing residential behaviour given the probability of residential behaviour at time t-1 (i.e.
15-min before), and 2) the mean weekly probability of residential behaviour given the mean
weekly probability of residential behaviour at time t-1 (i.e. 1 week before). In both analyses,
white sturgeon were very likely to remain in the same behaviour state. Environmental covariates
influenced behaviour only when there was less certainty of a behaviour state in the previous
timestep (i.e. probability of 50%). At a fine timescale, the interaction of season and zone was the
strongest predictor of residential behaviour given probabilities at time t-1 were 50%. HLK Eddy,
for example, was sensitive to seasonal changes and demonstrated lower probabilities of
residential behaviour in the summer and fall. At a coarser timescale, mean weekly probabilities
of residential behaviour were influenced by water temperature, discharge, and habitat zone.
Warming water temperatures related to decreased probabilities of residential behaviour and
higher flows from ALH related to increased probabilities of residential behaviour. Fish located in
the HLK Tailrace and Downstream habitat zones also demonstrated decreased mean weekly
probabilities of residential behaviour when probabilities at time t-1 were 50%.
While the exact biological and ecological processes that dictate behaviour cannot be
known for certain (Patterson et al. 2017), the movement behaviours in Chapter 2 were likely
correlated to feeding ecology and foraging strategies as observed for other sturgeon (McLean et
al. 2014) and fish species (Killen et al. 2007; Grant et al. 1987; Fausch et al. 1997; O’Brien et al.
1989). Residential behaviour and transitory behaviour fluctuated temporally and spatially.
Residential behaviour, possibly representing a low-energy sit-and-wait foraging strategy, was
expressed most frequently in the winter and spring as energy levels and metabolism likely
declined due to cooler water temperatures (Volkoff et al. 2020). On the other hand, transitory

115

behaviour, possibly representing a high-energy foraging strategy that can cover larger areas, was
expressed more frequently as metabolic demands for energy increased with warming water
temperatures. Additionally, I observed that the spatial distribution of behaviours shifted
seasonally, likely in response to facility discharge changing the downstream energetic landscape
(e.g. energy demands for movement) and resource (e.g. prey availability) environment (Taylor et
al. 2012; Turschwell et al. 2019; Shepard et al. 2013). Patches of residential behaviour, for
example, were observed to shift seasonally so as to possibly capitalize on low-velocity feeding
habitat adjacent to fast waters. Overall, my results demonstrated that hydroelectric operations in
the study area did appear to have some influence on white sturgeon movement behaviour.
Results from Chapter 2 can help inform recovery measures and management decisions.
Discharge, for example, could be adjusted in future mitigative measures so as to optimize flows
during periods of increased sturgeon feeding (e.g. summer and fall). These measures would be
especially important for habitat such as the HLK Eddy which has been identified as critical for
white sturgeon feeding (Fisheries and Oceans 2014). Patches of residential behaviour in the HLK
Eddy (likely representing feeding activity) were most prominent during the summer when
discharge was highest at both HLK and ALH facilities. In the fall, however, patches of
residential behaviour largely disappeared from the HLK Eddy as flows deceased and came
mainly from the ALH facility. I hypothesize that elevated summer flows provides upstream food
resources and extending the period of HLK discharge into the fall, or for a longer period of time,
may therefore improve white sturgeon feeding in the HLK Eddy during periods of increased
metabolic demand. I recommend future research utilize more detailed spatial analyses (e.g.
coordinates versus habitat zones) to further investigate the relationship between white sturgeon
behaviour and environmental variables related to hydroelectric operations.

116

In Chapter 3, two behaviour states were identified from my coarse-scale movement
dataset of hatchery-origin white sturgeon residing in the Transboundary Reach of the UCR.
These states also consisted of a residential and transitory behaviour, with the former representing
localized movement at a single receiver station and the latter representing longer movements
between receiver stations. The influence of biological factors and environmental conditions on
behaviour was assessed using the probability of fish expressing a behaviour state given the state
at time t-1 (i.e. 6-hours before). In my analysis, sturgeon spent most of their time expressing
residential behaviour and were unlikely to change behaviour when already in a residential state
(i.e. state at time t-1). However, when transitory behaviour was expressed at time t-1, the
probability of residential behaviour at time t depended more strongly on environmental
covariates. With increases in discharge, I observed sturgeon in more channelized river zones
(e.g. Zones 3 and 4) having increased probabilities of residential behaviour (when in a transitory
state at time t-1). Additionally, warming water temperatures were related to decreased
probabilities of residential behaviour when previously in a transitory state. Similar results were
observed in my assessment of maximum displacement. Warming water temperatures were
related to increased maximum displacement. Hatchery-origin sturgeon moving within Canada,
whose river zones are characterized as more channelized, also displaced far less than fish moving
within the United States, with increased flows related to faster declines in maximum
displacement in Canada compared to the United States.
While the intent of movement exhibited by hatchery-origin white sturgeon in the UCR
cannot be known for certain, similar movement patterns have been observed in other sturgeon
and fish species. For example, increased movement activity in response to water temperature has
been documented in wild white sturgeon in the UCR (Hildebrand & Parsley 2013) and other

117

sturgeon populations (Robichaud et al. 2017; Kynard et al. 2016). Movement activity during
warmer seasons (e.g. summer) is typically driven by feeding and spawning (Hildebrand et al.
2016). While spawning activity could not be confirmed for hatchery-origin sturgeon in this
study, feeding ecology may explain observed movements. Water temperature is a driver of fish
metabolism (Volkoff et al. 2020) and it would be reasonable to assume hatchery-origin sturgeon
are increasing movement during warmer water temperatures in response to increasing metabolic
demands. Additionally, discharge has been shown to alter the energetic landscape and demands
of fish movement (Shepard et al. 2013), with fish making trade-offs when deciding between
allocating energy to movement or maintenance of other physiological functions such as growth
and reproduction (Rennie et al. 2005). Consequently, the increased movement of sturgeon
observed in the United States may be related to feeding ecology, as lower energy is required to
move across its low-velocity and reservoir-like habitat for feeding opportunities. My results
could suggest a possible movement component for the growth differences observed in hatcheryorigin sturgeon between countries and river zones in the Transboundary Reach (Crossman et al.
2021; Van Poorten & McAdam 2010). Future research may benefit from integrating stable
isotope analyses to better understand feeding-related movements (Brownscombe et al. 2022), as
well as timestamped embryo collection or reproductive staging for possible spawning-related
movements (Buderman et al. 2021).
Overall, hatchery-origin white sturgeon in the UCR demonstrated strong site fidelity.
Fish spent most of their time expressing residential behaviour and movement typically did not
extend beyond two river zones. These results support trends observed in wild adult white
sturgeon located in the UCR (BC Hydro 2018a; Irvine et al. 2007; Hildebrand et al. 1999), as
well as other sturgeon populations (Robichaud et al. 2017; Kynard et al. 2016). There were only

118

a few instances of hatchery-origin sturgeon crossing the international border (4% of study
population), and all instances were from fish initially tagging in border adjacent river zones.
These results suggest that fishing mortality risk of sturgeon located in Canada migrating to the
United States is not substantial enough to warrant adjustments in the hatchery stocking program.
However, due to the short duration of the study period (i.e. 1.5 years), more time is
recommended to fully understand the movements of hatchery-origin sturgeon in the
Transboundary Reach.
A longer study period would also benefit our understanding of how environmental
covariates influence movement. Discharge, specifically, varied significantly between years. Peak
flows which occurred in the summer were far greater in 2020 than 2019. While increased
discharge has been found to increase the movement of sturgeon related to spawning (Forsythe et
al. 2012; Dammerman et al. 2018; Paragaminan et al. 1999), this study observed a negative
effect of discharge on maximum displacement. Higher discharge has been shown to negatively
affect detection efficiency in an acoustic telemetry array (Gjelland et al. 2013; Kessel et al.
2014). Therefore, it is possible that my observed declines in movement during increased flows
are a result of such detection efficiency bias. It is recommended that future studies track and
correct for detection efficiency so as to better understand the influence of environmental
variables on movement (Brownscombe et al. 2020).
A main takeaway from Chapter 3 is that hatchery-origin white sturgeon are moving and
residing throughout the Transboundary Reach. Specifically, in Canadian section of the
Transboundary Reach, habitats along the entire 57 rkm stretch are used year-round by white
sturgeon, with strong individual based site fidelity. Accordingly, maintaining the functionality
and connectivity of all habitats, such as eddies and spawning areas (Irvine et al. 2007), should be

119

critical for future recovery measures and management decisions. Currently, the majority of river
zones in the Transboundary Reach are not federally protected in Canada or the United States (no
critical habitat in river zones 2 – 4 and 6 – 10; Fisheries and Oceans 2014). Since hatchery-origin
white sturgeon are moving across and expressing high fidelity to river zones not cited as critical
and federally protected, future recovery planning should consider expanding protections to
include more, if not all, river zones of the Canadian section of the Transboundary Reach. In the
United States section where the population isn’t listed, my coarse-scale approach from Chapter 3
applied over a longer period could be used to help inform guidelines for the ongoing fishery to
ensure pertinent information on habitat use is available.

120

References
BC Hydro (2018a) Lower Columbia River Adult White Sturgeon Monitoring Program
(CLBMON-28). Year 10 Data Report. Report by BC Hydro, Castlegar, BC.
Brownscombe JW, Griffin LP, Chapman JM, Morley D, Acosta A, Crossin GT, Iverson SJ,
Adams AJ, Cooke SJ, Danylchuk AJ (2020) A practical method to account for variation in
detection range in acoustic telemetry arrays to accurately quantify the spatial ecology of aquatic
animals. Methods Ecol Evol 11:82-94. https://doi.org/10.1111/2041-210X.13322
Brownscombe JW, Shipley ON, Griffin LP, Morley D, Acosta A, Adams AJ, Boucek R,
Danylchuk AJ, Cooke SJ, Power M (2022) Application of telemetry and stable isotope analyses
to inform the resource ecology and management of a marine fish. J Appl Ecol.
https://doi.org/10.1111/1365-2664.14123
Buderman FE, Gingery TM, Diefenbach DR, Gigliotti LC, Begley-Miller D, McDill MM,
Wallingford BD, Rosenberry CS, Drohan PJ (2021) Caution is warranted when using animal
space-use and movement to infer behavioural states. Mov Ecol 9:30.
https://doi.org/10.1186/s40462-021-00264-8
Dammerman KJ, Webb M, Scribner KT (2019) Riverine characteristics and adult demography
influence female lake sturgeon (Acipenser fulvescens) spawning behaviour, reproductive
success, and ovarian quality. Can J Fish Aquat Sci 76: 1147-1160. https://doi.org/10.1139/cjfas2018-0141
Fausch KD, Nakano S, Kitano S (1997) Experimentally induced foraging mode shift by
sympatric charrs in a Japanese mountain stream. Behav Ecol 8:414-420.
https://doi.org/10.1093/beheco/8.4.414
Fisheries and Oceans Canada (2014) Recovery strategy for white sturgeon (Acipenser
transmontanus) in Canada. In Species at Risk Act Recovery Strategy Series. Fisheries and
Oceans Canada, Ottawa, ON.
Forsythe PS, Scribner KT, Crossman JA, Ragavendran A, Baker EA, Davis C, Smith KK (2012)
Environmental and lunar cues are predictive of the timing of river entry and spawning-site arrival
in lake sturgeon Acipenser fulvescens. J Fish Biol 81:35-53. https://doi.org/10.1111/j.10958649.2012.03308.x
Gjelland KO, Hedger RD (2013) Environmental influence on transmitter detection probability in
biotelemetry: developing a general model of acoustic transmission. Methods Ecol Evol 4:665674. https://doi.org/10.1111/2041-210X.12057
Grant JW, Noakes DL (1987) Movers and stayers: foraging tactics of young-of-the-year brook
charr, Salvelinus fontinalis. J Anim Ecol 1:1001-1013. https://doi.org/10.2307/4962

121

Hildebrand L, McLeod C, McKenzie S (1999) Status and management of White Sturgeon in the
Columbia River in British Columbia, Canada: An Overview. J Appl Ichthyol 15:164-172.
https://doi.org/10.1111/j.1439-0426.1999.tb00227.x
Hildebrand LR, Parsley M (2013) Upper Columbia White Sturgeon Recovery Plan - 2012
Revision. Prepared for the Upper Columbia White Sturgeon Recovery Initiative.
Irvine RL, Schmidt DC, Hildebrand LR (2007) Population Status of White Sturgeon in the
Lower Columbia River within Canada. Trans Am Fish Soc 136:1472–1479.
https://doi.org/10.1577/T06-190.1
Kessel ST, Cooke SJ, Heupel MR, Hussey NE, Simpfendorfer CA, Vagle S, Fisk AT (2014) A
review of detection range testing in aquatic passive acoustic telemetry studies. Rev Fish Biol
Fisher 24:199-218. https://doi.org/10.1007/s11160-013-9328-4
Killen SS, Brown JA, Gamperl AK (2007) The effect of prey density on foraging mode selection
in juvenile lumpfish: balancing food intake with the metabolic cost of foraging. J Anim Ecol
76:814-825. https://doi.org/10.1111/j.1365-2656.2007.01237.x
Kynard B, Bolden S, Kieffer M, Collins M, Brundage H, Hilton EJ, Litvak M, Kinnison MT,
King T, Peterson D (2016) Life history and status of Shortnose Sturgeon (Acipenser
brevirostrum LeSueur, 1818). J Appl Ichthyol 32:208-48. https://doi.org/10.1111/jai.13244
McLean M, Simpfendorfer C, Heupel M, Dadswell M, Stokesbury M (2014). Diversity of
behavioural patterns displayed by a summer feeding aggregation of Atlantic sturgeon in the
intertidal region of Minas Basin, Bay of Fundy, Canada. Mar Ecol Prog Ser 496:59-69.
https://doi.org/10.3354/meps10555
Paragamian VL, Kruse G, Wakkinen VD (1999) Kootenai River white sturgeon investigations.
Annual Progress Report of Idaho Department of Fish and Game to Bonneville Power
Administration, Boise, ID.
Patterson TA, Parton A, Langrock R, Blackwell PG, Thomas L, King R (2017) Statistical
modelling of individual animal movement: an overview of key methods and a discussion of
practical challenges. AStA Adv Stat Anal 101:399-438. https://doi.org/10.1007/s10182-0170302-7
O'Brien WJ, Evans BI, Browman HI, O'Brian WJ (1989) Flexible search tactics and efficient
foraging in saltatory searching animals. Oecologia 80:100-110
https://www.jstor.org/stable/4219015
Rennie MD, Collins NC, Shuter BJ, Rajotte JW, Couture P (2005) A comparison of methods for
estimating activity costs of wild fish populations: more active fish observed to grow slower. Can
J Fish Aquat Sci 62:767-780. https://doi.org/10.1139/f05-052

122

Robichaud D, English KK, Nelson TC (2017) Annual movements of acoustic-tagged white
sturgeon in the lower reaches of the Fraser River and its tributaries. T Am Fish Soc 146:611-625.
https://doi.org/10.1080/00028487.2017.1294542
Shepard ELC, Wilson RP, Rees WG, Grundy E, Lambertucci SA, Vosper SB (2013) Energy
Landscapes Shape Animal Movement Ecology. Am Nat 182:298-312.
https://doi.org/10.1086/671257
Taylor MK, Cooke SJ (2012) Meta-analyses of the effects of river flow on fish movement and
activity. Environ Rev 20:211-219. https://doi.org/10.1139/a2012-009
Turschwell MP, Stewart‐Koster B, King AJ, Pusey B, Crook D, Boone E, Douglas M, Allsop Q,
Jackson S, Kennard MJ (2019) Flow‐mediated predator-prey dynamics influence fish
populations in a tropical river. Freshwater Biol 64:1453-1466. https://doi.org/10.1111/fwb.13318
Van Poorten BT, McAdam SO (2010) Estimating differences in growth and metabolism in two
spatially segregated groups of Columbia River white sturgeon using a field based bioenergetics
model. Open Fish Sci J 3:132-141. http://dx.doi.org/10.2174/1874401X01003010132
Volkoff H, Rønnestad I (2020) Effects of temperature on feeding and digestive processes in fish.
Temp 7:1-14. https://doi.org/10.1080/23328940.2020.1765950