-
Identifier
-
Thesis_MES_2022Su_PavlinovicA
-
Title
-
Does boat presence affect the behavior of Sounders in inland waters? A study on gray whales (Eschrichtius robustus) in North Puget Sound
-
Date
-
September 2022
-
Creator
-
Pavlinovic, Alex
-
extracted text
-
Does boat presence affect the behavior of Sounders in inland waters? A
study on gray whales (Eschrichtius robustus) in North Puget Sound.
by
Alexander Thomas Pavlinovic
A Thesis
Submitted in partial fulfillment
Of the requirements for the degree
Master of Environmental Studies
The Evergreen State College
{September} {2022}
©{2022} by {Alexander Thomas Pavlinovic}. All rights reserved.
This Thesis for the Master of Environmental Studies Degree
by
Alexander Thomas Pavlinovic
has been approved for
The Evergreen State College
by
_______________________________
John B. Kirkpatrick, Ph.D.
Member of the Faculty
_______________________________
Date
ABSTRACT
Does boat presence affect the behavior of Sounders in inland waters? A study on
gray whales (Eschrichtius robustus) in North Puget Sound.
Alexander Thomas Pavlinovic
Keywords: Gray Whales, Anthropogenic Effects, Behaviors
Understanding the impacts of vessel presence on whale behavior is a crucial aspect of
cetacean conservation. “The Sounders” are a small group of gray whales that stop to feed
in North Puget Sound during their annual migration. This leg of their journey exposes
them to whale-watching boat-based operations, recreational, and commercial vessel
traffic. Research on other cetacean species has indicated that boat presence may have
adverse effects, such as disrupted foraging and avoidance of heavily-trafficked areas.
There are limited studies on gray whales regarding the behavioral impacts of boats
presence. To address this gap, land-based research was conducted from Hat Island,
Washington, using a theodolite to record the locations and movements of gray whales.
The presence or absence of vessels within one km were also tracked. These observations
began on March 14 and ended on May 14, 2021. Whales were tracked for 51 days for a
total of 78 hours. 39% of all whale observations occurred while boats were within 1000
m. Results indicated that whale’s speed, inter-breath intervals, deviation, and direction
indices differed when boats were within 1000 m of the whale. The Sounders’ prolonged
periods of foraging close to shore provided an ideal opportunity to study the impacts of
boats on gray whales, the results of which may better inform the conservation and
regulation of the species
Table of Contents
Table of Contents……………………………………………………………..…………..iv
List of Figures…………………………………………………………………………….vi
List of Tables…………………………………………………………………………….vii
Acknowledgments………………………………………………………………………viii
1 Introduction………………………………………………………………………….....01
2 Literature Review………………………………………………………………………03
2.1 Species Description…………………………………………………………..03
2.2 Life History…………………………………………………………………..04
2.3 Feeding……………………………………………………………………….04
2.4 Distribution & Abundance…………………………………………………...05
2.5 Unusual Mortality Event……………………………………………………..06
2.6 Pacific Coast Feeding Group………………………………………………...07
2.7 Sounders/ North Puget Sound Feeding Group……………………………….08
2.8 Whale-Watching Regulations………………………………………………..10
2.9 Anthropogenic Effects……………………………………………………….10
2.10 Shore-Based Observations………………………………………………….11
2.10.1 Theodolite Observations………………………………………….12
3 Methods………………………………………………………………………………...13
3.1 Study Area…………………………………………………………………...13
3.1.1 Observation Posts…...……………………………………………...14
3.1.2 Hydrographic Map Around Hat Island.………………...………….16
iv
3.2 Theodolite & Visual Observations…………………………………………..17
3.2.1 Tracking Whales…………………………………………………...18
3.2.2 Tracking Boats……………………………………………………..22
3.2.3 Calculating & Cleaning Variables…………………………………24
3.2.4 Focal Follow Types………………………….……………………..26
3.2.5 Measuring Anthropogenic Effects & Data Analysis………………27
4 Results…………………………………………………………………………….……31
4.1 Direction, Deviation, & Speed Indices…………………...………………….32
4.2 Whale Inter-breath intervals (IBI)…………………………………………...40
4.3 Summary……………………………………………………………………..45
5 Discussion……………………………………………………………………………...46
5.1 Interpreting the Data Analysis……………...……………………………….46
5.2 Research Implications……...………………………………………………..51
5.3 Management Implications…………………………………………………...53
5.4 Project & Analysis Limitations ...…………………………………………...54
5.5 Future Research….……………...…………………………………….….....57
5.6 Final Thoughts………………….………………………………………..….57
6 Bibliography……...………………………………………………………………......59
v
List of Figures
Figure 2-1: Image of Gray Whale………………………………………………..………03
Figure 2-2: Home Range of Gray Whales…………………………………………...…..05
Figure 2-3: Box plot showing Strandings of Gray Whales………………………………06
Figure 2-4: Image Showing Ghost Shrimp………………………………………………09
Figure 3-1: Hat Island’s Location in the Pacific Northwest……………………………..13
Figure 3-2: Hat Island’s Location near the Snohomish River Delta…………………..…14
Figure 3-3: Hydrographic Map Surrounding Hat Island…………………………………16
Figure 3-4: Team Set up for Tracking Whales & Boats………………...……………….18
Figure 3-5: How DEV, DI, & IBI were calculated……………..………………………..26
Figure 4-1: Observation Posts Locations on Hat Island…………………………………32
Figure 4-2a.: Violin Plot Showing Distribution of Direction Indices (DI)………............34
Figure 4-2b.: Violin Plot Showing Distribution of Deviation Indices (DEV)……...……35
Figure 4-2c.: Violin Plot Showing Distribution of Speed………………..………………36
Figure 4-3a.: Violin Plot Showing the distribution of Inter-breath Interval Durations….41
Figure 4-3b.: Violin Plot Showing the distribution of Dive Durations…………………..42
Figure 4-3c.: Violin Plot Showing the distribution of Blows Durations………………...43
vi
List of Tables
Table 2-1: Sounders Recorded in the Puget Sound by Year……………………………..08
Table 4-1: Summary Statistics of Response Variables…………………………………..33
Table 4-2: Configurations of Final and Best fit Generalized Least Squares Models for
Inter-breath Intervals, Direction, Deviation, and Speed Indices…………………………39
vii
Acknowledgments
In the field, things did not always go to plan. There were advances and setbacks, but I
appreciate all the help I received. There truly are countless people who helped make this
project happen. I will try to recognize them all below properly, but I apologize in advance
for anyone I missed.
There were many people I talked about the idea of this project with or listened to my
ramblings about gray whales, including Dr. Pauline yu, Don Chalmers, Dyanna
Lambourn, Joe Jauquet, Jeff Harris, Dr. Tony Orr, and the staff and interns of Cascadia
Research Collective.
I want to thank the research assistants of Irissa Danke and Quinn Wilson. These folks
were outside, rain or shine, with me recording data and helping build the project.
I want to recognize the volunteers who helped with the fieldwork in 2020, including
Lynn Pavlinovic, Joe Jauquet, Elsa Toskey, Mackenzie Davidson, Steve Dilley, Dan
Lowe, Norm Rathvon, Maggie Santangelo, Kathleen and Bruce C. Thompson Phillip
Meade, Katie Wold, Elizabeth and Harry Fuchs.
Regarding theodolite advice, help, and troubleshooting, I want to thank Marena Salerno
Collins, Florence Sullivan, and Lisa Hildebrand. They all answered countless questions
and responded to many emails.
Also, related to theodolites, I want to thank David Anderson. He helped me troubleshoot
the theodolite many times. His thesis was also an excellent reference and sometimes a
framework for writing my own.
I want to thank Lori Christopher for all her help. Lori not only got me onto Hat Island,
which is private, she also toured me around the island. During this tour, we located and
gained access to use all of the observation posts. Lori helped secure a vehicle and rental
for the team on the island. She also continually made us aware of a gray whale cited near
Hat Island. She was instrumental in the success of the project.
I also want to give a big thank you to the Hat Island community. Not only did they
welcome my many helpers and me into their community, they also let us use their
property to observe the whales and even fed us from time to time. For landowners, I want
to thank Kevin and Jody Meyers, Karen and Stanley Van Spanje, and the owners of K26,
W18, and M27. Shawn Potter ferried us countless times to and from the mainland. I also
want to thank Kelly and Kerry Dukes, Larry and Linda Bender, Laurie Gray, Kyle
Opstead, and Darla and David Younce.
Without the financial support from the Washington Chapter of the WIldife Society and
the Master's in Environmental Studies program, this project wouldn't have been possible.
viii
Bill Vogel contributed countless hours to help me. He proofed papers and grant
applications. He was always available to bounce ideas off. He helped secure volunteers
and funding. From the beginning to the end of the project, Bill supported and helped with
this project.
Lynn Pavlinovic, my mother, helped a lot. She proofed papers and helped secure funding.
While in the field, Lynn watched my dog, Rosie. She constantly provided moral support
and made an excellent field assistant.
I want to thank Kiirsten Flynn and John Calambokidis for their support. They reviewed
papers and presentations. Both helped secure equipment and supported me throughout the
thesis project. Before this project started, I bounced countless ideas about possible
master's projects. After picking one, he provided expertise and guidance.
Dr. John Kirkpatrick was another critical piece of this project. It all started with him
agreeing to help me apply for grants. Little did he know that we would end up applying to
12 of them. Next, it continued with him helping me prepare for my fieldwork and
supporting me while I was on Hat Island. Most recently, he has reviewed many drafts of
my thesis. He trusted me to do the work but was there to support me if it didn't. By the
end of this thesis, we will have exchanged hundreds of emails and spent just as many
hours on zoom or phone. While we didn't always agree on things, I feel lucky to have had
him as my reader.
I want to thank my partner, Elsa Toskey. They spent countless hours helping me figure
out statistical issues. Elsa proofed papers and presentations for me; they were excellent
field assistant. Elsa provided emotional and moral support, always loving, and putting up
with me throughout this long process.
ix
1. Introduction
The Sounders or North Puget Sound (NPS) gray whales (Eschrichtius robustus) have
been documented detouring into Puget Sound to feed on ghost shrimp every year since
the 1990s. Since the original two whales appeared to have discovered this feeding area,
the group has grown to approximately 15 individuals as of 2021 (Table 2-1). While there,
they interact with commercial, recreational, and boat-based whale-watching operations.
With little known about how these interactions may affect the Sounders or gray whales in
general, this project aimed to evaluate the presence of boats on the behaviors of the
Sounders. With little known about the influence of anthropogenic activities on gray
whales, this project focused on the Sounders, and provided an excellent opportunity to
assess these impacts.
With few studies about gray whales and whether boat's presence impacts them, we
have been forced to rely on studies, primarily on other cetacean species. Usually, boats
have adverse effects on whale's behaviors (Senigaglia et al., 2016). Documented results
included disrupted foraging, erratic movements, and energy expenditure (Senigaglia et
al., 2016), which caused them to avoid and abandon high-disruption areas (Lusseau &
Bejder, 2007) and altered gray whale’s communication (Burnham & Duffus, 2019).
Effects on whale behavior depended on different factors (Senigaglia et al., 2016) and
appeared to vary by species (Weinrich & Corbelli, 2009).
Using the methods from Williams et al. 2002, the hypothesis was boats presence
impacted whale behavior, specifically speed, direction (DI), and deviation (DEV) indices,
dives, blows, and Inter Breath Intervals (IBI). The question posed was, when boats were
1
present within 1000 m or less of a whale, would their presence impact the whale's
behaviors? To answer this query, whale behaviors were recorded and divided into two
categories. First, when a boat was present within 1000 m or less of the whale being
tracked. Second, when no boat was present within 1000 m or less of the whale being
tracked. To separate the two groups, when boats were within 1000 m of the whale, their
presence and details were recorded. This data was gathered using a theodolite to track the
whales and boats and a laptop to record the data. After comparing the two datasets, there
were noteworthy results.
This project focused on an understudied species and attempted to answer an
unanswered question. Little research has been done on boat’s impacts on gray whales and
none on the Sounders. Because the Sounders feed around Hat Island, Washington, for
several months of the year, it was an ideal place to answer this question. These results are
applicable broadly across the geographic home range of gray whales (Figure 2-2) and
could be used to manage the species better.
2
2. Literature Review
2.1 Species Description
Figure 2-1: This image shows a gray whale (NMFS, 2020a).
Gray or grey whales are a unique and common cetacean species found in the
North Pacific Ocean (Swartz & Jones, 2016). Typically, they are seen traveling or
foraging near shore (Swartz & Jones, 2016). Gray whales are the only remaining member
of the Eschrichtiidae Family, but they have a close phylogenetic relationship with the
Balaenopteridae Family made up of rorquals (Swartz & Jones, 2016).
As their name indicates, gray whales generally are a gray hue (Figure 2-1). Their
backs have whiteish spots (Figure 2-1). Because they travel slowly, they have extensive
amounts of barnacles and cyamids (whale lice) on their backs (Osborne et al., 1988).
Unlike most Mysticetis, gray whales do not have a dorsal fin but a dorsal ridge of sixtwelve knuckles (Osborne et al., 1988). Like other Mysticetis, these whales have two
blow holes and use baleen to consume their prey. Gray whales have large pectoral fins
and flukes (NMFS, 2020a). Their maximum length is 15 m, and they have a maximum
weight of 35 tons (Sumich, 2014). They are also sexually dimorphic, with females
slightly larger than males (Sumich, 2014).
3
2.2 Life History
Gray whales are estimated to live for several decades, with mothers birthing
claves every other year (NMFS, 2020a). After a 12-13 month gestation period, a gray
whale calf is born (NMFS, 2020a). After spending seven to eight months with its mother
migrating and nursing, the calf is weaned (Swartz & Jones, 2016). If the calf survives, it
will become sexually mature between the ages of six and twelve, with the average of
eight years old being the same for both sexes (Swartz & Jones, 2016). After becoming
sexually active, males attempt copulation at every opportunity, while females typically
enter estrus every two years (Swartz & Jones, 2016). This latter situation allows the
females to regain some of their critical fat stores, which are depleted while nursing the
calves (Swartz & Jones, 2016). While the average lifespan of a gray whale is unknown,
one deceased female was estimated to be between 75 and 80 years old (NMFS, 2020a).
2.3 Feeding
Gray whales are flexible foragers utilizing many species to flourish. Their primary
way of foraging uses suction feeding, but they have been known to employ gulp and skim
feeding (Swartz & Jones, 2016). Their primary and preferred food source comprises
benthic organisms, including a minimum of 60 benthic amphipod species and 80 to 90
other benthic invertebrate species (Swartz & Jones, 2016). They will also consume
swarming species, including krill, mysids, shrimp, sardines, anchovies, and other species
(Swartz & Jones, 2016).
With their feeding only taking place during the roughly five months spent in the
feeding grounds, these whales must fast and rely on their fat reserves for the rest of the
4
year (Swartz & Jones, 2016). To build these reserves, gray whales have been known to
forage 24 hours a day (Swartz & Jones, 2016).
2.4 Distribution & Abundance
Figure 2-2: This image shows the approximate home range of gray whales (NMFS, 2020a).
Historically, gray whales were found throughout the Atlantic and Pacific Oceans
(Duffield, 2009). Presently, gray whales are found only in the north Pacific Ocean. Gray
whales of the Eastern North Pacific (ENP) stock have been recorded making an annual
round trip migration of 15,000-25,000 km, making their species the longest migrating
Mysticetis (Swartz & Jones, 2016).
According to NMFS, there are two remaining stocks of gray whales. First is the
ENP stock, which breeds in lagoons in Baja, Mexico, and primarily feeds in the Chukchi
Sea and Bering Straits (Figure 2-2) (NMFS, 2020a). As of 2015, before the Unusual
Mortality Event (UME), this group contained approximately 27,000 whales (NMFS,
5
2021a). Second, there is the Western North Pacific (WNP) stock, comprised of less than
300 individuals, which feeds in the same area and is thought to breed somewhere off
Korea (Figure 2-2) (NMFS, 2020a). Until recently, these two stocks were thought to be
mostly isolated from each other (NMFS, 2020a). However, a recent project demonstrated
that some WNP gray whales traveled to the calving Lagoons in Baja, Mexico (NMFS,
2020a).
2.5 Unusual Mortality Event
Figure 2-3: This box plot shows the strandings of gray whales between Alaska, Washington,
Oregon, and California (NMFS, 2022).
From 2019 to August 2022, 578 gray whales have stranded and died on the west
coasts of the United States, Canada, and Mexico (Figure 2-3). These whales are
considered part of the ENP (NMFS, 2022). Because of the deaths in 2019 and 2020, this
event was declared a UME, which continues into the present year of 2022. To add
6
perspective, these strandings were estimated to be only 14% of the estimated mortality of
4700 whales (CRC, 2022). For perspective, the normal mortality for the Eastern North
Pacific population of gray whales is 29 whales stranded yearly (NMFS, 2021b).
2.6 Pacific Coast Feeding Group
Unlike the majority of the ENP, the Pacific Coast Feeding Group (PCFG) is a
group of 232 whales that don’t make migration to Alaska (Calambokidis et al., 2019). In
2017, this group of 232 whales made up less than 1% of the approximately 27,000 whales
of the ENP stock. These animals feed from Northern California, USA, to Vancouver
Island, Canada (Calambokidis et al., 2019). While there, they part in risky feeding
behaviors in shallow waters off the coasts of these countries. Two studies comparing the
PCFG and ENP whales genetically found significant mtDNA haplotype frequency
differences (Calambokidis et al., 2019). This development provides the strongest
evidence so far that the PCFG whales may be significantly isolated enough from the ENP
stock to allow for maternally inherited mtDNA (Calambokidis et al., 2019).
7
2.7 Sounders/ North Puget Sound Feeding Group
N Puget Sound gray
whale (Sounders)
histories for whales
seen more than 2 years
Yr
s
20
21
20
20
32
31
31
30
31
31
24
23
23
22
22
2
22
2
2
4
3
3
2
2
2
M
F
F
Lucyfer
20
19
20
18
20
17
20
16
20
15
20
14
20
13
20
12
20
11
20
10
20
09
20
08
20
07
20
06
20
05
20
04
20
03
20
02
20
01
20
00
19
99
19
98
19
95
19
94
M
F
M
M
M
M
M
19
93
Sex
Shackleton
Earhart
Dubnuck
Patch
Little Patch
19
92
Name
21
22
44
49
53
56
185
356
383
396
531
543
723
1193
1213
2246
2249
2255
2259
2261
2356
19
91
ID
19
90
Confirmed N Puget Sound gray whale IDs seen in multiple years compiled by Cascadia Research
M
NPS area
PCFG area only
Table 2-1: This table shows the North Puget Sound gray whales recorded in the Puget Sound
and/or Pacific Coast Feeding Group areas. It also shows which years they were present in these
areas (CRC, 2022).
In 1990, the first members of the North Puget Sound (NPS) feeding group, known
as the Sounders, were recorded in the North Puget Sound. These individuals detour into
Washington waters for approximately three months before following the other members
of the ENP stock to forage in Alaskan waters. The founding group was made up of CRC
ID 21 (Shackleton) and 22 (Earhart) (Table 2-1). To be a Sounder, the whale must have
shown up for more than two years. Since then, the group has increased. Between 1999
and 2000, the group grew to 11 individuals (Table 2-1). Between 2019 and 2021, the
group again increased to approximately 15 individuals (Table 2-1).
Interestingly, some of the PCFG whales have transitioned to being Sounders
(Table 2-1). During all three-time segments, UMEs were impacting the ENP. So, it could
8
be speculated that these events pushed the gray whales to explore new areas to survive
the difficult events.
To improve their conditions, the Sounders take part in risky behavior in shallow
waters to suction feed on ghost shrimp, also known as sand shrimp (Figure 2-4) (Pruitt &
Donoghue, 2016). In the wake of these efforts, the gray whales leave feeding pits
(approximately 3 x 2 m in size) (Pruitt & Donoghue, 2016).
This strategy of feeding on the shrimp has been beneficial for the Sounders.
Between 2020 and 2022, while many other whales in the ENP stock were in poor
condition, the Sounders and other whales recorded in Puget Sound improved their
condition (Fearnbach & Durban, 2022). While these changes may have happened during
different years, there were no drone studies to measure the whale’s body conditions until
2020.
Figure 2-4: This image shows three ghost shrimp taken in a sample from the Puget Sound (Pruitt
& Donoghue, 2016).
9
2.8 Whale-Watching Regulations
Whale-watching regulations for commercial trips and the public on the West
Coast of the United States break into two categories of killer whales and other cetacean
species. For killer whales, their protection ranges from a minimum distance between the
boats and whales of 200 to 400 y/m (NMFS, 2020b). Other cetaceans, including gray
whales, have a minimum distance between the boats and whales of 100 y/m or .1 km
(NMFS, 2020b).
2.9 Anthropogenic Effects
Few studies have investigated anthropogenic effects on gray whales. Therefore,
we mainly rely on similar studies on other cetaceans. Whether documented effects on
whale’s behaviors are present depends on various factors (Senigaglia et al., 2016). In
most studies, researchers concluded that the presence of boats adversely affects whales,
but this was not consistent across all species and regions (Senigaglia et al., 2016). Some
of these disturbances resulted in fasting and increased erratic movement, which combined
to cause calorie deficits (Senigaglia et al., 2016). Regarding acoustics, boats have been
known to alter gray whale’s (Burnham & Duffus, 2019) and humpback whale’s
communications (Fournet et al., 2018). For Northern Resident killer whales, both males
and females showed avoidance tactics during the periods when boats approached them
(Williams et al., 2002). In Iceland, minke whales responded to the presence of boats by
decreasing their IBI and increasing their sinuous movements (Christiansen et al., 2013).
From a study on WNP stock of gray whale’s responses to seismic surveys, the
researchers found that some of the whales sped up and traveled in a more linear path
away from the sounds when present, but these impacts were not found across all
10
individuals studied (Gailey et al., 2016). Cumulatively, these disturbances can cause
whales to avoid and abandon high-disruption areas (Lusseau & Bejder, 2007).
Sullivan and Torres)produced the most similar paper to this thesis in 2018, which
attempted to assess the impacts of vessel disturbance on gray whales off Oregon. This
paper had robust methods of using a theodolite and laptop running the software
Pythagoras to record the locations of boats and whales and then compare the distances
between the two. I disagreed with the way they performed their data-analysis.
Specifically, I disagreed with their use of the residence in space and time concept, which
divided observations of gray whales into three behaviors of search, travel, and forage,
which was an oversimplification of the gray whale’s behaviors. Also, they chose to
exclude resting from the behaviors, which I disagreed with. I felt that they needed to also
measure other response variables of speed, DEV, DI, and IBI. Partially, because of this
research, I wanted to perform a similar research project to assess the impacts of vessel
disturbance on gray whales.
2.10 Shore-Based Observations
For cetacean research, shore-based observation methods have been used for
decades (Piwetz et al., 2018). These noninvasive methods allow researchers to record
information about nearshore marine mammals’ habitats, behaviors, and movements
without affecting the studied individuals (Piwetz et al., 2018). For studying
anthropogenic effects on marine mammals, this tool allows researchers to study these
impacts without affecting the animals themselves.
11
2.10.1 Theodolite Observations
The theodolite was built for surveying and construction, but marine mammal
scientists have used it for research (Piwetz et al., 2018). In the 1970s, Roger Payne and
colleagues repurposed it to study marine mammals (Piwetz et al., 2018). Since then,
technology has come a long way. Originally, the angles and other information provided
by a theodolite were written down, which would later be used to calculate the
observational data (Piwetz et al., 2018). Now, theodolites can be connected to a laptop
running software, which organizes and calculates the data from the observations almost
immediately (Piwetz et al., 2018). For computer software, Pythagoras can be used, which
calculates the locations and records the behaviors of the marine mammals being studied
(Gailey et al., 2016).
12
3. Methods
3.1 Study Area
The Pacific Northwest's Hat Island (Figure 3-1), also known as Gedney Island, is
in the North Puget Sound of Washington State (Figure 3-2). It provided an ideal
location for observations of gray whale interactions with boats because gray whales
forage around this island every year from March to May (Calambokidis et al., 2002)
Figure 3-1: This image taken from Google Earth shows Hat Island’s location relative to the
Pacific Northwest (Google Earth, 2020a).
13
Figure 3-2: This image shows Hat Island relative to the Snohomish River Delta (Google Earth,
2020b).
3.1.1 Observation Posts
Three observation posts (OPs) were utilized for this project to observe boats,
whales, and their interactions. In 2019, a team visited the island to scout the island for
the upcoming field season. Because Hat Island was a private island, permission was
required to ride the ferry to the island and gain access to private property. From this
trip, five OPs were initially selected. After the first week of the project in 2020, the
number of OPs was reduced to three.
From this point forward, the OPs will be referred to by a letter followed by a
number (e.g., F15) used on the island (historically, the island was divided into
14
approximately 21 different areas of houses, with each area assigned a specific letter).
For each OP, the GPS location and heights were entered into the laptop, and reference
azimuth points (e.g., tops of flag poles) were selected. Additionally, known points (at
sea level) (e.g., boat navigation piling) were chosen to check the accuracy of the
theodolite after it was set up.
The most used OP to observe the whales from was F15 (48°00'57" N, 122°19'06"
W, 53 m elevation). This location provided the best vantage point, with almost 180
degrees of view. Also, it offered an excellent view of the Snohomish River Delta,
where the gray whales fed almost daily. When observing from this OP, the lowest tide
was -.18 m, and the highest was 3.1 m, with a mean height of 1.65 m.
The next most used OP was W18 (48°00'17" N, 122°18'21" W, 38 m elevation).
From this post, the whales could be observed South and West of Hat Island. From this
area, the whales were often seen traveling more than foraging. When observing from
this OP, the lowest tide was -.83 m, and the highest was 3.31 m with a mean height of
1.07 m.
The least used post was M27 (48°1'14" N, 122°19'44" W, 78 m elevation). This
OP provided a view north of the island, including Camano Head and East Tulalip
Bay. From this site, the whales were seen traveling more often than foraging. When
observing from this OP, the lowest tide was -.16 m, and the highest was 3.33 m with a
mean height of 1.49 m.
15
3.1.2 Hydrographic Map Around Hat Island
Print Map - TopoZone
https://www.topozone.com/map-print/?lat=48.5239945&lon=-122.5479...
Using the map below, we can see that Puget Sound's floor is quite variable in its
Hat Island Topo Map in Skagit County Washington
depths around Hat Island (USGS, 1993). While some places near the island are
+shallower than -10 m, the majority of depths are greater (USGS, 1993). In contrast to
−
this shallowness, a deep area northwest of the island is greater than -166 m.
Figure 3-3: This image shows hydrographic information about the Puget Sound around Hat
Island (USGS, 1993) in the metric system
(m depth).
! Print
this map
Map provided by TopoZone.com
16
3.2 Theodolite & Visual Observations
From March 15 to May 14, 2021, theodolite and visual observations were used to
record locational data on boats and whales and behavioral data on whales. From the
locational data, the proximity of the vessels to the whales was recorded. These
observations were limited by the visibility rated on a zero to five scale, usually a four
or five. A zero value represented the observers' inability to see six feet in front of
them. The Beaufort Sea state also limited the team, usually a two when tracking. One
day, the group attempted to observe with the Beaufort scale at a five. After a time,
these efforts ceased, because of an inability to follow the whales, due to the height of
the waves. All observations took place during daylight hours.
For fieldwork, there was a team ranging from two to four individuals (Figure 3-4).
For recording locational data on boats and whales, a Sokkia DT5A theodolite was
used with a 5 s level of precision and a 30-power monocular magnification (Gailey et
al., 2016). This theodolite was connected to a Toughbook computer running the
software Pythagoras (v1.2), which calculated the precise locations of whales and
boats. This act was completed using a cord attached to the RS232 output point on the
theodolite and a USB port on the laptop. This software also allowed us to record IBIs
on the computer. Additional observations were made with binoculars or a spotting
scope.
17
Figure 3-4: This image shows the team set up for tracking whales. The theodolite operator would
be using the device on the yellow stand. The computer operator in the red chair would be
recording the data. The spotter utilizing the spotting scope would be communicating data to the
computer operator.
3.2.1 Tracking Whales
Tracking or conducting focal follows of a specific whale can be broken into five
distinct steps.
1. First, the whale or whales had to be located, which was done by using Facebook
whale-watching groups and receiving tips from Hat Island Residents. If those
options were not producing whales to observe, the team would rotate through the
different OPs and scan the surrounding area with binoculars.
18
2. Once a whale was located, the timing of its breaths was recorded each time it
surfaced. These efforts are all under the title of Inter-breath Intervals (IBIs),
which include blows, dives, and missed blows and dives, as defined below.
Surfacings: This is an all-inclusive term, including missed dives, missed blows, IBIs,
dives, and blows.
Missed dives & blows: When the observers or operators miss the timing of a "blow
or dive" but know that it occurred. The recording of these behaviors results in the IBI
durations were not calculated.
IBI: The time between whales blows for any duration, including blows dives.
Dive: When the whale blows after being under for 60 seconds or more.
Blow: When the whale blows after being under for less than 60 seconds.
If the timing of the dive duration being recorded were incorrect, then a note would
be recorded to remove it later. For example, if the residue of the blow was visible, but
the actual blow was not seen. For data accuracy, the exact moment when the whale
first blew was recorded. If this was even a second late, a missed dive or blow would
be recorded instead.
After leveling the theodolite, the visibility, Beaufort Sea state, and tide heights
were recorded into Pythagoras. These numbers were updated and recorded on the
laptop at least every thirty minutes. Then, the check accuracy point was used. If the
theodolite and laptop provided us with the correct latitude and longitude, then
tracking would begin. If not, then the theodolite was releveled. Also, if the theodolite
19
was bumped, we would check the accuracy of the theodolite. Due to the height of
observation posts, which increased the accuracy of the theodolite (Würsig et al.,
1991), tracking was conducted out to 7600 m. Beyond this distance, visibility became
an issue. So, there were no efforts to track whales or boats beyond 8000 m.
Next, whale locations were recorded using a theodolite. Each recorded data point
had a fixed type, time, day, latitude, and longitude.
If an inaccurate fix was recorded, either the delete last fix option was selected in
Pythagoras, or a note would be taken to remove it from the data set later.
Theodolite accuracy was checked on one occasion. On March 30, 2021, the first
session to check the accuracy of the theodolite took place from stations M27 and
W18. This test was completed with the help of a Cascadia Research Collective vessel.
An individual on the ship recorded a GPS point; then, it would be compared to the
land-based crew's data. For a theodolite, the further the distance between the
theodolite and the boat or whale being tracked, the less accurate the theodolite is.
Test 1
On March 30, 2021, the accuracy of the theodolite was tested from M27 and
W18. The maximum distance between the boat and the land-based team was 2307 m.
After comparing the data sets, the difference in accuracy between the boat and
land-based teams' data was 0-22.2 m, except for P9 and P5b. The average difference
for latitudinal values was 0.00009375 (much less than 11.1 m) and for longitudinal
values was 0.0000875 (much less than 11.1 m). For P9, the theodolite was bumped,
20
which caused the difference between the land-based and boat-based observations to
increase to 22-33 m for the longitude and latitude values. For P5b, the locations did
not match up because of a communication issue between the land and boat-based
teams about when to record the point.
This tracking session gave us confidence in the equipment used for this research.
The slight difference between the data recorded by the boat versus the land-based
team could be attributed to the groups fixing the points at slightly different times.
This situation likely happened because the two teams had identical values during
some fixes.
3. When a whale displayed a singular behavior, the team recorded it (see definitions
below). If false behaviors were recorded, the computer operator would put a note
in to remove it later.
Pectoral Fin showing: When the pectoral fin breaks the surface of the water in any
way, i.e., half showing or fully exposed, usually when the whale was on its side
assumed to be feeding.
Peduncle showing: When the area where the tail fluke connects to the body was
visible after a blow/surface right before a dive.
Spy-hopping: When the whale surfaces vertically so that only its head was seen.
Fluke Up: When after a surface, a whale lifted its fluke completely to clear the
surface of the water in preparation for a dive.
21
Side Fluke: When a whale was seen on its side, and the fluke broke the surface in a
vertical position, often seen during foraging bouts.
Breaching: When all or most of a whale's body broke the surface of the water.
4. When possible, the gray whale being tracked was identified. Each gray whale that
has visited Puget Sound has been given a corresponding CRC identification
number (Table 2-1). Unfortunately, due to the distance at which the whales were
tracked, this situation happened approximately 1 in 25 times. When it happened,
the team utilized a Nikon DSLR with a 400mm telephoto lens.
5. When operated properly, all these steps listed above became a streamlined
process. The whale would first be identified and selected, and the station would
be set up. If a boat or boats were within 1 km, it would be mentioned in the notes.
First, the breath would be recorded on the laptop. Next, a fix for the whale would
be taken. These two actions would take about two sections and involve using two
different windows in Pythagoras. If the whale produced any singular behaviors,
those would be recorded. While the whale was subsurface, a fix of the boat or
boats would be taken. After this, the computer operator got ready to record the
breathtaking place, which would quickly be followed by the fix.
3.2.2 Tracking Boats
Anytime a boat was within 1000 m of the whale, this vessel was recorded either
by putting it in the notes or getting a fixed point of the ship. These boats were broken
into distinct categories based on types and whether they were participating in whalewatching activities. Recreational/ non-whale-watching boats were this if…
22
•
They were not actively pursuing the whale.
o
No change in speed or direction to match the path of the whale.
•
They were significantly distanced from the whale >1000m.
•
They were traveling with no delineation in speed or direction in response to a
whale's presence.
While Whale-watching boats were whale-watching if…
•
They were actively pursuing a whale with obvious changes in boat speed and
directionality to match the whale's path.
•
They were within a close range of the whale, <1000m, with signs of using
cameras or binoculars.
•
They were known as commercial whale-watching boats on active tours.
•
Their initial speed and direction changed their orientation towards the whale in
response to a whale's blow/surface.
Additionally, vessels were divided into categories based on their hull design and
purpose (Wladichuk et al., 2018), including ridged-hull inflatable boats (RHIBs),
monohulls, catamarans, landing craft, and sailboats. These same vessels were also
divided between being used for recreational, commercial, or research purposes.
23
3.2.3 Calculating & Cleaning Variables
After removing eccentric and individual (singular) data points, IBIs, speeds,
direction (DI), and deviation indices (DEV) were calculated. For IBIs, if the notes
said that the dive or blow was missed or the resulting duration was longer than 10
minutes, then that IBI duration was removed. For the movement metrics, if the
whale's speed exceeded 15.5 KM/hour, then the corresponding data point would be
removed.
From the timing of breaths, IBIs were calculated as the time between two breaths
(Christiansen et al., 2013). If a missed dive or blow occurred, then IBI was not
calculated for that period (Christiansen et al., 2013). Because of the shallowness of
the study area, any dives ≥ 10 minutes were excluded.
From the GPS points and time, speed, direction (DI), and deviation indices (DEV)
were calculated for the whales. These metrics were only calculated to get the best
analyses if there were a minimum of two surfacings before the specific data point
(Christiansen et al., 2013). For each of these groups of three, the three metrics were
calculated. This situation allowed us to get the most out of the data. Speed was
calculated by dividing the distance traveled between two surfacings by the time taken
to complete this action. After reviewing the literature, the highest speed at which a
gray whale was recorded traveling was 15.5 KM/hour (Sumich, 1983). So, any speeds
recorded greater than that were thrown out.
24
DEV and DI show the linearity and path predictability of a whale's path (Williams
et al., 2002). DEVs ranged from 0°-180°, with 0° being a straight path and 180° being
a change in the whale's path to the opposite direction (Figure 3-5).
The values for DI range from 0-1, with 1 showing the whale traveled in a straight
line and 0 indicating that the whale traveled in the opposite direction. For equations to
calculate DI and DEV, please see the description of Figure 3-5 (Christiansen et al.,
2013).
25
Figure 3-5: This image and description show how DEV, DI, and IBI were calculated, “Example
of a movement track of a minke whale with 3 surfacings (Pt, Pt−1 and Pt−2) and 2 inter-breath
intervals (IBI) (l1 and l2); Pt is the present position, Pt−1 the previous position, etc. and L is the
net distance traveled between Pt and Pt−2. The deviation index for l2 is α. The directness index
(DI) for l2 is calculated by DI = 100 · [L/(l1 + l2)]” .
3.2.4 Focal Follow Types
The data recorded about the whales were divided into two distinct categories (boat
& no boat). Boats were believed to be present within 1000 m of the whale (boat) if…
•
Boats were fixed within 1000 m of the whale.
•
Notes mention boats present within 1000 m of the whale.
No boats were believed to be present within 1000 m of the whale (no boat) if…
26
•
Boats were not fixed within 1000 m of the whale or were fixed beyond 1000 m of
the whale.
•
Notes contain no mention of boats present within 1000 m of the whale.
For blows and dives, they were divided into boat or no boat categories. If a boat was
present within 1000 m of the whale for most of the blow or dive duration, it was put in
the boat category. If a boat was present within 1000 m of the whale for less than half of
the blow or dive duration, it was put in the no boat category.
The movement metrics were divided into different focal follows. If a whale was
tracked for the first 10 minutes without a boat present, it would be in the no-boat
category. However, if a boat appears on the scene at the 11-minute mark, it is broken into
a new focal follow in the boat category. For example, if fixes one through twenty are of
the same whale, one through ten are with no boat present within 1000 m, and eleven
through twenty are with a boat present within 1000 m, then these fixes are separated into
two separate focal follows for analysis.
3.2.5 Measuring Anthropogenic Effects & Data Analysis
Speed, DI, and DEV data were recorded about whales to measure possible
anthropogenic impacts. The datasets were broken into two distinct categories of when
there were and were not boats within 1000 m of a whale. The two categories of boats
present versus no boats present had two subcategories. Movement metrics (speed, DEV,
DI) were analyzed as one subcategory. The second subcategory was Inter-Breath Indices
(IBI). To properly analyze the data points, these two groups had to be separated due to
differences in their data structures.
27
For the movement metrics, the differences between the two groups were tested
using multivariate statistics. After the movement metrics were cleaned, organized, and
calculated, these three-movement metrics were graphically assessed using violin plots
between the two groups (boat versus no boat). Second, all three-movement metrics were
tested for normality using the Shapiro-Wilk normality test. Next, we used a multivariate
Hotelling's T2 test to determine the possible impacts of ships on all three-movement
metrics together. This test allowed us to account for correlation between behaviors,
increase the test's power, and preserve the type I (experiment-wise) error rate.
To assess the difference between individual movement metrics, both Welch's twosample t-test and Mann-Whitney U-tests tests were used. First, to assess the difference
between individual movement metrics parametric, two-sample t-tests were used. Even
though the distributions of the movement metrics were non-normal, because of the
Central Limit Theorem, these tests were suitable for the datasets. According to this
theorem, if the datasets contain a large enough sample size, which movement metrics
(956 boat, 1843 no boat) had, then the lack of normality in the data is not problematic,
and the t-test was suitable (Lumley et al., 2002). For this test, Welch's t-tests were chosen
over Student's because of an unequal variance between the two groups. Additionally, to
compare the results of the t-test to a nonparametric test, our three metrics were analyzed
using Mann-Whitney U-tests. This test does not require the datasets to be normally
distributed.
To account for any correlation structure among the movement metrics and to
calculate the effect of each movement metric in separating the two groups of whales with
and without boats present, standardized discriminant function coefficients were analyzed.
28
To standardize the data, each movement metrics distribution was converted to a mean of
0 and a standard deviation of 1. Doing this allows for a direct comparison of the
contribution of the movement metrics to the difference between the groups (boat versus
no boat).
To assess whether boat presence impacted IBIs, Welch's univariate t-test and
Mann-Whitney U-tests were used. After the IBIs were cleaned and organized and IBIs
were calculated, this data was divided into three categories of all IBIs, dives, and blows.
The three dive duration metrics were graphically assessed between the two groups (boat
versus no boat). All three categories were tested for normality using the Shapiro-Wilk
normality test. Then, the parametric Welch's two-sample t-tests were used to determine
the difference between the boat versus no boat data for IBIs, dives, and blows. Even
though the distributions of the IBIs were non-normal, because of the Central Limit
Theorem, these tests were suitable for the datasets. According to this theorem, if the
datasets contain a large enough sample size, which IBIs (1523 boat, 2549 no boat), dives
(555 boat, 725 no boat), and blows (978 boat, 1864 no boat) had, then the lack of
normality in the data did not matter, and the t-tests were suitable (Lumley et al., 2002).
For the univariate t-tests, Welch's t-tests were chosen over Student's because of an
unequal variance between the two groups. To compare the results of the t-test to a
nonparametric test, our three metrics were analyzed using Mann-Whitney U-tests. This
test does not require the datasets to be normally distributed.
To account for the limited number of focal follows (movement metrics = 98, IBIs
=116) and three observation posts (F15, W18, M27) used to observe the whales,
Generalized Least Squares (GLS) models were used. To create the best fit models for our
29
variables (speed, DEV, DI, IBI), many different models were compared using AIC and
BIC scores. The final models were selected by using these scores. GLS was not utilized
for blows or dives. These comparisons included accounting for the correlation among
focal follows by using them as the autocorrelation structure for models, which made the
model fit better. Additionally, boat and OP were used as the variance structures to create
a better-fitting model for the movement metrics. This change was not made for IBI
because the models fit worse with that variance structure added.
30
4. Results
Data was recorded from March 15 to May 14, 2021, on Hat Island,
Washington. These efforts resulted in 76.5 hours of observational data on whales.
To analyze speed indices, direction indices, deviation indices, dive durations,
blow durations, and IBI durations, R (R v4.1.3) was used.
This section will first cover the results and analyses for the movement
metrics of speed, DEV, and DI. Next, the results and analysis of IBIs, dives, and
blows will be presented. Finally, this chapter will end with a summary of the
analysis performed for this thesis.
31
Figure 4-1: This image shows the locations of the observation posts on Hat Island, Washington
(Google Earth, 2020c).
4.1 Direction, Deviation, & Speed Indices
Over 98 focal follows, 4537 fixes of a whale’s locations (1297 boat, 3363 no boat),
and 1127 fixes of boats were recorded. F15 (Figure 4-1) was the site where we observed
the whales from the most, which produced the most fixes totaling 2871. The next most
productive OP was W18 (Figure 4-1), with 909 fixes. The least used OP was M27
(Figure 4-1), totaling 757 fixes. After removing the fixes used to check the accuracy of
the theodolite, removing the incorrect fixes, and the fixes which could not be used to
calculate movement metrics, 3187 fixes remained. The distance between the theodolite at
32
the observation post and the whale or boat being tracked ranged from 256 to 7601 m,
with the average distance being 3264 m. From this data, 2799 estimates (956 boat, 1843
no boat) of DEV, DI, and Speed were calculated. These estimates were broken into 98
focal follows ranging from 19 seconds to 2 hours and 9 minutes.
Description
Boat/ No
Boat
IBI (S)
Boat
66.71
No Boat
Dives (S)
Boat
No Boat
Blows (S)
Boat
No Boat
Speed (km/h)
Boat
No Boat
DEV
Boat
No Boat
DI
Boat
Mean
SD +/-
Min
Max
N
2
478
1523
53.09
65.18424
49.42066
2
448
2549
130
71.19985
60
478
555
84
60.66844
60
448
725
30.5
2
59
978
30.71
12.9423
13.91502
2
59
1864
3.5361
1.905158
0.0155
15.1265
956
2.92
1.954908
0
15.346
1843
28.8345
35.6329
0.0205
179.9494
956
41.844
45.2577
0
179.997
1843
0.94689
0.1276795
0.05444
1
956
No Boat
0.91527
0.1633662
0.01684
1
Table 4-1: This table shows the data distribution for all the response variables, the presence
(boat) or
absence (no boat) of vessels within a km of the whale being tracked after the data had been
cleaned.
s = seconds, km/h = kilom traveled per hour, IBI =inter-breath interval, DEV = deviation indices,
DI = direction indices, SD = standard deviation, and n = number of observations used.
First, all the variables were plotted using violin plots (Figures 4-2) to show their
distribution by the presence of boat(s) using the package ‘ggplot2’. The mean directness
index of whale movement with boats was 0.95 and without boats was 0.92. The mean
33
1843
deviation index of whale movement with boats was 28.8 and without boats was 41.8. The
mean speed index of whale movement with boats was 3.54 km per hour and without
boats was 2.92 km per hour. The violin plot with the greatest apparent difference between
boat and no boat was for the deviation indices with a larger spread and median for no
boat than boat.
a.
34
b.
35
c.
Figures 4-2 (a-c): These three violin plots show the distribution of the movement
metrics (a) direction, (b) deviation, and (c) speed.
Second, all the movement metrics were tested for normality using the Shapiro-Wilk
normality test in base R. For speed, p < 2.2 x 10-16, which was less than 0.05, the null
hypothesis was rejected. For DEV, p < 2.2 x 10-16, which was less than 0.05, the null
hypothesis was rejected. For DI, p < 2.2 x 10-16, which was less than 0.05, the null
hypothesis was rejected. All three-movement metrics were not normally distributed.
Third, based on the multivariate Hotelling’s T2 test in the ‘ICSNP’ package in R, the
null hypothesis was rejected. This test allowed us to determine the possible impacts of
ships on all three-movement metrics together and account for the correlation between
behaviors. These results showed whale’s movement metrics between the whales in the
36
presence of boats and whales, not in the presence of boats, were not equal (p = 2.2 x 1016
, a = 0.05, T2 = 40.441, df1 = 3, df2 = -2795). While considering the movement metrics
of a gray whale, there was a significant difference between when boats were present
within a km of a whale or not.
Fourth, parametric univariate Welch’s T-test in base R was used to assess the
difference between individual movement metrics. Each movement metrics of speed,
direction, and deviation were significantly different when boats were present within a km
of a whale versus not. For speed, p = 1.6 x 10-15, which was less than 0.05, the null
hypothesis was rejected. This evidence suggests that whale’s swimming speeds
significantly differed when boats were present versus not (t = 8.038, df = 1977.3). For
deviation, p < 2.2 x 10-16, which was less than 0.05, the null hypothesis was rejected at
the 5% level. The evidence suggests that the whale’s deviation indices significantly
differed when boats were present versus not (t = -8.33, df = 2363.8). For direction, p =
2.0 x 10-08, which was less than 0.05, the null hypothesis was rejected at the 5% cutoff
level. This test suggested that whale’s direction indices significantly differed when boats
were present versus not (t = 5.63, df = 2377).
Fifth, nonparametric Mann-Whitney U-tests were used to assess the difference
between individual movement metrics in base R. The metrics speed, direction, and
deviation were significantly different when boats were present within a km of a whale
versus not. For speed, p < 2.2 x 10-16, which was very small, the null hypothesis was
rejected. This evidence suggests that whale’s swimming speeds significantly differed
when boats were present versus not (w = 1080535). For deviation, p = 3.084 x 10-08 was
very small, and the null hypothesis was rejected. The evidence suggested that the whale’s
37
deviation indices significantly differed when boats were present versus not (W =
768694). For direction, p = 2.704 x 10-05, which was very small, the null hypothesis was
rejected. This test suggested that whale’s direction indices significantly differed when
boats were present versus not (W = 966048).
Sixth, to account for any correlation structure among the movement metrics and to
calculate the effect of each movement metric in separating the two groups in the
Hotelling’s T2 test, standardized discriminant coefficient function analysis was used in
the ‘MASS’ package in R. DEV appeared to have the most significant impact on
separating the two groups (boat versus no boat) with a linear discriminant coefficient
value of 0.905. The next most significant variable was speed, with a linear discriminant
coefficient value of -0.695. Last and least was DI, with a linear discriminant coefficient
of 0.266.
Last, based on generalized least squares models analysis using the ‘nlme’ package in
R, boats presence still impacted the whale’s movement metrics, even after accounting for
the correlation among focal follows by adding focal follows as the autocorrelation
structure. Additionally, by adding Boat x OP as the variance structure, a better-fitting
model was established for the movement metrics. For speed, p < 1 x 10-04, which was less
than 0.05, and thus the null hypothesis was rejected. Evidence suggested that the whale’s
speed indices significantly differed when boats were present versus not (t = -5.66040, df
= 2795). For observation posts, when observing whales from OPW18 (p < 1 x 10-04, t = 8.40020, df = 2795), but not OPM27 (p = .8124, t = 0.23730, df = 2795), evidence led us
to believe that this significantly affected speed indices. Using the AIC and BIC scores as
38
guidance to select the best-fitting model, the final model for speed had an AIC score of
11032 and a BIC score of 11098 (Table 4-2).
Fixed
Variables Effects
IBI
Speed
DEV
DI
Variance
Structure
Correlation Structure
AIC
BIC
Pboat
PM27
Boat* OP
corAR1(form = ~ 1|Focal_Follow)
44053 44135
< .001 0.539
Boat* OP
corAR1(form = ~ 1|Focal_Follow) Boat * OP
11032 11098
< .001 0.8124
Boat* OP
corAR1(form = ~ 1|Focal_Follow) Boat * OP
28258 28323
< .001 < .001
Boat* OP
corAR1(form = ~ 1|Focal_Follow) Boat * OP
-2797 -2762
< .003 < .001
Table 4-2: This table shows the configurations of the final and best fit generalized least squares
models for each of our variables. The table format came from (Christiansen et al., 2013). IBI
=inter-breath interval, DEV = deviation indices, DI = direction indices, Boat = boat or boats
present within a km of the whale, OP = observation post from which the observations were taking
place, focal follow = tracking session number, AR= auto-regression, AIC = Akaike’s information
criterion, BIC = Bayesian information criterion.
For DEV, p = 1 x 10-04 was less than 0.05, and the null hypothesis was rejected. The
evidence suggests that the whale’s deviation indices significantly differed when boats
were present versus not (t = 2.017075, df = 2795). For OPs, when observing whales from
OPW18 (p < 1 x 10-04, t = 2.408271, df = 2795) and OPM27 (p < 1 x 10-04, t = 2.714495,
df = 2795) evidence led us to believe that this significantly affected deviation indices.
Using the AIC and BIC scores as guidance to select the best-fitting model, the final DEV
model had an AIC score of 28258 and a BIC score of 28323 (Table 4-2).
For DI, p = 3.4 x 10-03 was less than 0.05, and the null hypothesis was rejected. The
evidence suggests that the whale’s deviation indices significantly differed when boats
were present versus not (t = -2.93583, df = 2795). For OPs, when observing whales from
OPW18 (p < 1 x 10-04, t = 4.90659, df = 2795) and fOPM27 (p < 1 x 10-04, t = 4.84677, df
= 2795) evidence led us to believe that this significantly affected direction indices. Using
the AIC and BIC scores as guidance to select the best-fitting model, the final model for
DI had an AIC score of -2797 and a BIC score of -2762 (Table 4-2).
39
PW18
< .035
< .001
< .001
< .001
4.2 Whale Inter-breath intervals (IBIs)
From the initial fieldwork, 6222 focal behaviors were recorded. From this
dataset, 4355 surfacings or missed surfacings (1816 boat, no boat 2539) were recorded
over 116 focal follows. These focal follows ranged in duration from 40 seconds to 2
hours and 42 minutes. F15 (Figure 4-1) provided the most surfacings with 3033. The next
most productive OP was W18 (Figure 4-1), with 793 surfacings. The least used OP was
M27 (Figure 4-1), with 529 surfacings. After excluding the missed dives and blows and
cleaning the data, 4072 IBIs (1523 boat, 2549 no boat) were calculated. The range for IBI
was 2 to 478 seconds. IBIs were divided into dives (555 boat, 725 no boat) and blows
(978 boat, 1864 no boat). The range for dives was 60 to 478 seconds, and blows were 2 to
59 seconds.
First, IBIs, dives, and blows were plotted using violin plots (Figures 4-3) to show
their distribution by the presence of boats using the package ‘ggplot2’. The mean whale
IBI durations were 67 seconds with boats and 53 seconds without boats. The mean whale
dive durations with boats were 130 seconds and without boats was 84 seconds. The mean
whale blow durations with boats were 30.5 seconds and without boats was 30.71 seconds.
Visually, the violin plot with the most difference between boat and no boat was for dive
durations with a larger spread and median for boat than no boat.
40
a.
41
b.
42
c.
Figures 4-3(a-c): These three violin plots show the distribution of the durations for (a)
IBIs, (b) dives, and (c) blows.
Second, IBIs, dives, and blows were tested for normality using the Shapiro-Wilk
normality test in base R. For IBIs, p < 2.2 x 10-16, which was less than 0.05, the null
hypothesis was rejected. For dives, p < 2.2 x 10-16, which was less than 0.05, the null
hypothesis was rejected. For blows, p < 2.2 x 10-16, which was less than 0.05, the null
hypothesis was rejected. All three of the breathing metrics datasets were not normally
distributed.
43
Third, to assess the difference between the two categories of IBIs, dives, and
blows, parametric univariate Welch’s T-tests in base R were used. IBIs and dives were
significantly different when boats were present within a km of a whale versus not, but
blows were not. For IBIs, p = 2.526 x 10-12, which was less than 0.05, the null hypothesis
was rejected. This evidence suggests that whale’s IBI durations significantly differed
when boats were present versus not (t = 7.0361, df = 2566.2). For dives, p = 5.062 x 1008
, which was less than 0.05, the null hypothesis was rejected at the 5% level. The
evidence suggests that the whale’s dive durations significantly differed when boats were
present versus not (t = 5.4879, df = 1084.6). For blows, p = 0.6841, which was not less
than 0.05, the null hypothesis was not rejected at the 5% level. This test proved that
whale’s blow durations were not significantly different when boats were present versus
not (t = 5.63, df = 2377).
Fourth, nonparametric Mann-Whitney U-tests were also used to assess the difference
between individual movement metrics in base R. IBIs and dives were significantly
different when boats were present within a km of a whale versus not, but blows were not.
For IBIs, p = 6.074 x 10-08, which was very small, the null hypothesis was rejected. This
evidence suggests that whale’s IBI durations significantly differed when boats were
present versus not (W = 2137666). For dives, p = 4.362 x 10-10 was very small, and thus
the null hypothesis was rejected. The evidence suggests that the whale’s dive durations
significantly differed when boats were present versus not (W = 242084). For blows, p =
0.8573, which was greater than 0.05, the null hypothesis was not rejected. This test
proved that whale’s blow durations did not appear significantly different when boats were
present versus not (W = 915232).
44
Last, based on generalized least squares models analysis using the ‘nlme’
package in R, boats presence still impacted the whale’s IBI durations, even after
accounting for the correlation among focal follows by adding focal follows as the
autocorrelation structure. For IBI, p < 1 x 10-04, which was less than 0.05, the null
hypothesis was rejected. The evidence suggested that the whale’s IBI durations
significantly differed when boats were present versus not (t = -6.789453, df = 4067). For
OPs, when observing whales from OPW18 (p < 1 x 10-04, t = -2.113918, df = 4067), but
not OPM27 (p = 0.5388, t = 0.614735, df = 4067), evidence led us to believe that this
significantly affected IBIs. Using the AIC and BIC scores as guidance to select the bestfitting model, the final model for IBI had an AIC score of 44053 and a BIC score of
44135.
4.3 Summary
In this section, the results of the data and analyses were presented. For the movement
metrics, the analysis began with visual comparisons between the boat and no boat
categories. As the analysis proceeded, increasingly more complex analyses were applied
to the movement metrics. The final analysis was generalized least squares (Table 4-2),
which accounted for the correlation among focal follows by adding focal follows as the
autocorrelation structure. Similar analyses were applied to inter breath-intervals,
excluding multivariate statistics and discriminant function coefficient analysis. Across
many different analyses, excluding the variable blow, significant differences were found
when comparing gray whale’s speeds, DEVs, DIs, IBIs, and dives when boats were
present within 1 km of the whale versus when there were no boats present within that
distance.
45
5. Discussions
5.1 Interpreting the Data Analysis
Initially, the question was whether the presence of boats would impact whale’s
behaviors of speeds, DIs, DEVs, IBIs, dives, and blows. It was hypothesized that boat
presence would impact the whale’s behaviors listed above. However, at what distance
would boat’s presence impact whale’s behaviors? To answer this query, three distance
intervals of 100 m or less, 101-400 m, and 401-1000 m were planned to be used. In trying
to answer those original three questions, only a combined version of all three were
answered. When boats were within 1000 m or less of a whale, most gray whale’s speeds,
DIs, DEVs, IBIs, and dives were impacted by the presence of boats, but not blows.
Intending to record 200 hours and having recorded 76 hours of observations, it was
decided not to subdivide the dataset further into the three categories of the distance
between the whale and boat of 100 m or less, 101-400 m, and 401-1000 m.
For the movement metrics of speed, DEV, and DI, whale’s behaviors were
significantly different between when boats were and were not within 1000 m of the whale
being studied using multivariate tests. To analyze the impacts of boat’s presence on the
three-movement metrics together, a multivariate Hotelling’s T2 test was used. With p =
2.2 x 10-16 at the 5% level, this number tells us that the difference in the whale’s
movement metrics were significantly different in the presence of boats versus not. To
calculate the effect of each movement metric in separating the two groups in Hotelling’s
T2 test, standardized discriminant coefficient function analysis was used, resulting in the
linear discriminant coefficient values of DEV = 0.905, speed = -0.695, and DI = 0.266.
These values tell us that DEV had the most significant impact on separating the two
46
groups (boat versus no boat), then speed, and the least was DI. While analyzing the
movement metrics as a group, whale’s speeds, DIs, and DEVs recorded when boats were
present within 1000 m were significantly different than when boats were absent within
that distance.
To assess the differences in the boat versus no boat categories for the individual
movement metrics, parametric Welch’s T-tests and nonparametric Mann-Whitney U-tests
were employed. From the Welch’s T-tests, speed p = 1.6 x 10-15, deviation p < 2.2 x 10-16
and direction p = 2.0 x 10-08, at the 5% level, the two categories of each of the movement
metrics differed significantly. Using the nonparametric Mann-Whitney U-tests, speed p <
2.2 x 10-16, DEV p = 3.084 x 10-08, DI p = 2.704 x 10-05, all very small values, there were
significant differences in the two categories of whale movement metrics when boats were
within 1000 m versus not. Both these tests seemed to support the idea that there was a
significant difference between the boat versus no boat categories for DI (Figure 4-2a.),
DEV (Figure 4-2b.), and Speed (Figure 4-2c.). Still, since not all observations were fully
randomized – for example, a single focal follow of an individual may have generated
multiple measurements – autocorrelation needed to be considered.
Using generalized least squares models, boat’s presence was assessed to still
impact the whale’s movement metrics of speed, p < 1 x 10-04, DEV p = 1 x 10-04, and DI
p = 3.4 x 10-04, all tested at the 5% level. This test allowed lumping all the individual
fixes into their corresponding focal follows (movement metrics = 98), using focal follows
as an autocorrelation structure. Also, the variables Boat X OP were added as the variance
structure to create a better-fitting model. Despite making both additions, whale metrics
recorded when boats were present were still significantly different from when boats were
47
absent. This test gave a high level of confidence that the boat’s presence affected the
whale’s movement metrics because of the ability to treat all the fixes in a focal follow as
correlated.
Using generalized least squares model analyses, the impacts of recording the
movement metrics from different locations or OPs were investigated. For speed, when
observing whales from OPW18 p < 1 x 10-04, but not OPM27 p = .8124, evidence showed
that this significantly affected speed indices. For DI, when observing whales from
OPW18 p < 1 x 10-04 and fOPM27 p < 1 x 10-04, evidence showed that this affected
direction indices considerably. For DEV, when watching whales from OPW18 p < 1 x
10-04 and OPM27 p < 1 x 10-04, evidence indicated that this affected the deviation index
considerably. These results are not surprising because the whales performed different
behaviors in front of different OPs. For example, the whales were more likely to forage in
front of F15 and travel in front of W18. Since all observations were combined for a
complete analysis, including sites with different behaviors, the fact that the generalized
least squares model still yielded significant results underscores the robustness of the
observed differences in whale behavior.
For IBIs, dives, and blows, both parametric Welch’s T-tests and nonparametric
Mann-Whitney U-tests were used to assess the differences in the boat versus no boat
categories. From the Welch’s T-tests, IBIs p = 2.526 x 10-12, dives p = 5.062 x 10-08, and
blows p = 0.6841, at the 5% level, the whale metrics of IBIs and dives were significantly
different between when boats were and were not within 1000 m of the whale being
studied, but not blows. Using the nonparametric Mann-Whitney U-tests, IBIs p = 6.074 x
10-08, dives p = 4.362 x 10-10, and blows p = 0.8573, IBI and dive values were very small
48
values, but not the blow values. So, the whale metrics of IBIs and dives significantly
differed between when boats were and were not within 1000 m of the whale being
studied, but not blows. Both tests found that the boat versus no boat categories differed
significantly for IBIs (Figure 4-3a.) and dives (Figure 4-3b.), but not blows (Figure 43c.).
Using generalized least squares models, the whale’s IBIs recorded when boats
were present were significantly different than when boats were absent p < 1 x 10-04,
which was less than 0.05, even after adding focal follows as an autocorrelation structure.
This analysis was not completed for dive or blow durations. Even after lumping all the
individual IBIs into their corresponding focal follows (dive durations =116), using focal
follows as an autocorrelation, boat’s presence still impacted whale’s movement metrics.
This test gave a high level of confidence that the boat’s presence affected the whale’s
IBIs because of the ability to treat all the fixes in a focal follow as correlated.
Using generalized least squares model analyses, the impacts of recording the IBIs
from different OPs were investigated. When observing whales from OPW18 p < 1 x 10 04
, but not OPM27 p = 0.5388, evidence led us to believe that this significantly affected
IBIs. These results were not surprising because whales performed different behaviors in
front of other OPs. When whales foraged, they had a more variable IBI pattern than when
traveling.
To be cautious, only variables analyzed with generalized least squares models
were used. To properly analyze these datasets, the IBI durations and fixes were not
analyzed individually. Instead, they were analyzed with all data points in their
49
corresponding focal follows (movement metrics = 98, dive durations =116), completed
using generalized least squares models.
After running the statistical tests, there was a high level of confidence that whale
metrics of speed, DI, DEV, and IBI recorded when boats were present within 1000 m
significantly differed from when boats were absent. Over 76.5 hours of observations and
countless other hours either observing whales or spent in the field, these efforts resulted
in 98 movement metrics, and 116 IBI durations focal follows, made up of 2799 estimates
of movement metrics and 4072 estimates of IBI durations. For speed, whale metrics
recorded when boats were within a km of the whale had a mean speed of 3.54 km versus
2.92 km per hour when boats were absent (Figure 4-2c.). So, gray whales probably sped
up in the presence of vessels. For DI, on a 0-1 scale, with 1 being a perfectly straight path
between points and 0 signifying a 180-degree turn, whale metrics recorded when boats
were within a km of the whale had a mean value of 0.95 versus 0.92 and had a more
extensive spread (Figure 4-2a.). For DEV, on a scale of 0-180, with 0 being a perfectly
straight path between points and 180 signifying a 180-degree turn, whale metrics
recorded when boats were within a km of the whale had a mean value of 28.8 versus a
mean value of 41.8 when boats were absent (Figure 4-2b.). Therefore, gray whales likely
traveled a more linear path in the presence of boats. Whether or not these paths were
toward, away from, or tangential to boat traffic was not resolved in this data set.
However, it is consistent with the observed increase in speed mentioned above. For IBIs,
whale metrics recorded when boats were within a km of the whale had a mean value of
67 seconds versus a mean value of 53 seconds when boats were absent (Figure 4-3a.). So,
whales probably had longer IBI durations in the presence of vessels.
50
The assumption was that causality works one way—that boat presence impacted
whale behavior, not that whale behavior influenced the presence/absence of boats. Even
after including all the potential confounding variables, time of day, a small group of
whales, varying depths of the water below the whales, number, and activity of vessels,
the previous statement is true. Boat size and activity (for example, noise level or engine
condition) could have large differences— for example, a full-throttle motorboat versus a
small sailing vessel. Also, given that there were less than 15 Sounders and around 100
focal follows, the same whales must have been tracked multiple times. For seafloor
depths, if a whale was tracked while it was in shallower water, it may have reacted
differently than if the whale was in deeper water. Observations were from 10:00-18:00
hours, and whales could have acted differently at different times. However, the fact that
observations were significant only meant the impact of confounding appears to be
relatively small. Still, this would be a valuable area for future research, with potential
policy implications. Given this, our analysis of the waters surrounding Hat Island shows
that certain gray whale’s behaviors appear to have been altered by the occurrence of boats
within 1000 m.
5.2 Research Implications
While these gray whales were in Puget Sound, they interacted with several boatbased whale-watching operations and large numbers of recreational and commercial
vessels. The results showed that boat’s presence likely impacted gray whales. With the
Sounders and the number of gray whales utilizing the Puget Sound increasing and
individual whales staying longer, this feeding area is becoming more critical to the
51
Eastern North Pacific Stock of gray whales (CRC, 2022). In the future, this area may
need to be protected.
The most recent unusually high mortality event for the Eastern North Pacific
population of gray whales started in 2019 (NMFS, 2022). Since that year and continuing
into this year, during that time frame, it is estimated that 25% of the approximately
20,000 whales died (CRC, 2022). To put this into perspective, the average mortality for
the 18 years before this UME for the Eastern North Pacific population of gray whales
was 29 whales stranded yearly (NMFS, 2021). As this population was already under
pressure, limiting the stress put on this population may be critical for preventing further
losses.
This research fits in well with previous research on the impacts of boats on
cetaceans and can be applied to other research on studying anthropogenic effects
(Christiansen et al., 2013; Senigaglia et al., 2016; Williams et al., 2002). Using methods
based on Williams et al. (2002), speed, inter-breath intervals, direction, and deviation
indices were measured when boats were and were not present within 1000 m of the
whales. Like their research on killer whales, the presence of vessels correlated with
changes in the gray whale’s behaviors. These similar reactions have also been with minke
whales (Christiansen et al., 2013) and many other cetacean species (Senigaglia et al.,
2016).
Interestingly, Gailey et al. (2016) found that individual gray whales reacted to
seismic surveys by speeding up and traveling in a more linear path away from the source
of the sounds, but this did not happen with every whale. This fact is notable because this
reaction was the same as observed in this study, albeit with recreational vessels. Also,
52
while the Sounders foraged in shallow water, they were much more prone to being
disturbed by boats (CRC, 2022). Future research on the Sounders population may be able
to address what individual differences exist in their reactions to the presence of human
activities.
The issue seems to be not if the whales were disturbed by the presence of boats
but how they would be disturbed. In Senigaglia et al. (2016), the impacts fell under the
categories of DEV, DI, IBI, speed, and activity budget, with the effects dependent on the
species, location, and methods. This research showed boats impacted gray whales in four
of the five categories listed in Senigaglia et al. (2016), a meta-analysis studying the
impacts of boat’s presence on whales. It fits in well with the scientific literature. Like
most previous papers published on assessing the anthropogenic effects of boats on
cetaceans, this research can be used to help guide similar projects in the future.
5.3 Management Implications
At the time of the research, the required separation distance between boats and gray
whales was 100 m/y (NMFS, 2020), but the cutoff used in this study was much further, at
1,000 m. Even at this distance, changes were observed. Among other marine mammals,
killer whales had special and increased protection from boaters of at least 200 m and up
to 400 m of distance (NMFS, 2020), but even this distance between the boats and whales
was relatively near. With gray whales found throughout the North Pacific Ocean, this
research could be used to better inform managers of the best ways to protect this species.
It has been noted that the Sounders were more sensitive to boats presence when
foraging (CRC, 2022). If a boat approached a foraging whale, it would likely stop
foraging (CRC, 2022). With a short window of high tide for the whales to forage,
53
interruptions of this critical time could cause large changes in whales caloric intake and,
thus, overall health. This information suggested that a “one distance fits all” approach
may not be as valuable as protection put into place for the whales in some of these areas.
Future research comparing different locations, such as the different observation posts
used in this study, could shed light on optimal management strategies.
5.4 Project & Analysis Limitations
Like all research projects, this one had limitations. Initially, the goal of this project
was to gather more observational data. Also, the plan was to have two years of field
seasons. Due to COVID canceling the pilot project in 2019, this work only took place in
2020. Also, in 2020, the ENP stock, which the Sounders are part of, faced a UME
(NMFS, 2022). So, arguably, the whale’s behaviors could have been different in these
years.
Because there were roughly 15 Sounders, individuals would have been recorded
multiple times. The whales were tracked at great distances between the whale and
theodolite, and often beyond 2000 m. Photo identification pictures could not be taken of
most of the whales tracked. Due to an often inability to identify the individual whale and
the possibility that different whales responded differently to the presence of boats, there
was no way to determine if the different whales reacted differently to the presence of
boats. For example, an older whale that has come into Puget Sound for 20+ years may be
more used to avoiding and dealing with boats than a younger whale. To solve this issue in
future projects, identification pictures of the whales being tracked would need to be
taken.
54
Observations were only from Hat Island. This situation could have caused a bias, due
to geographic limitations. The furthest a whale was tracked from Hat Island was
approximately 8000 m, which was not a small distance. Because of this, whales could
only be tracked within the greater Hat Island area. If whales behaved differently in other
parts of Puget Sound, this data would not be present in this study. Also, with the Puget
Sound surrounding Hat Island having varying depths, this variable could have affected
the whale’s behaviors. The influence of bathymetry as a confounding variable remains
unknown.
Using a theodolite for this research added a possible variable to this research. This
tool was used to record fixes and later used to calculate the response variables of speed,
DEV, and DI. The distance between the theodolite at the observation post and the whale
or boat being tracked ranged from 256 to 7601 m. Before a whale or boat was tracked, a
known point was used to check the accuracy of the theodolite. If the theodolite was
inaccurate, it would be reset. This check revealed several times that the theodolite had not
been set up correctly, and this data was removed from the analysis. Also, there was a
session used to check the accuracy of the theodolite, which produced an average
difference for the latitudinal and longitudinal values of 5.588 x 10-5 (less than 11.1 m).
This slight difference between the data recorded by the boat versus the land-based team
could be attributed to the groups fixing the points at slightly different times, which likely
happened because the two teams had identical values during some fixes. These processes
gave us confidence in the equipment used for this research.
There were variables focused on boats that could have impacted the data, specifically,
the number, size, type, and type and number of engines on the boats. To simplify matters,
55
the analysis compared the datasets of when there were boats within 1000 m of the whales
versus not. If there were multiple boats within that distance, a note was made, and fixes
on all the boats within 1000 m of the whales were attempted. When boats were fixed,
they were divided into categories based on their hull design and purpose (Wladichuk et
al., 2018), including ridged-hull inflatable boats (RHIBs), monohulls, catamarans,
landing craft, and sailboats. These same vessels were also divided between being used for
recreational, commercial, or research purposes. Also, if possible, information about the
boat’s engines was recorded, particularly size, number, and horsepower. This information
was not included in the analysis presented here because this data was not available for
every interaction with boats within 1000 m. However, if more boats with larger engines
were near the whale being tracked, it is possible that the impact on the whales would
have increased.
Field efforts were limited to daylight hours with good visibility. If it was raining too
hard, the whales could not be tracked. The whales could not be observed if the Beaufort
scale was too high. Therefore, this dataset primarily shows the whale’s behaviors on clear
and low-wind days. The observations usually occurred between 10:00 – 18:00 hours. It is
possible that whales may have acted differently at other times of the day or night or
during other weather events.
Initial methodologies for this project planned to divide the data about whale’s when
boats were within 1000 m of the whales into three categories of 100 m or less, 101-400
m, and 401-1000 m distance between ships and whales. The project’s original goal for
this analysis was to record 200 hours of observational data about the whales. By the end
of the field season, 76.5 hours of observational data were recorded. Because of the
56
amount of data collected, it was decided not to create the three subcategories and not to
dilute the significance of our datasets. Nonetheless, there was still high confidence in the
data.
5.5 Future Research
The research question has been answered; whale’s behaviors recorded when boats
were present within 1000 m or less between them significantly differed from when boats
were absent. For future research, it would be valuable to have more data to perform more
granular analyses. It would be interesting to compare the impacts on the whales when
boats were between 1000-401 m, 101-400 m, and 100 m or less. It could be speculated
that the closer the distance between the whales and boats, the more significant the impact.
This sort of “dose-response” impact would not only validate the cause-effect relationship
but could also be highly valuable for crafting future policy.
It would also be fascinating to look at changed behaviors and see if changes
caused the gray whales to burn extra calories. While the difference in swimming speed,
for example, was shown to be statistically significant, the practical significance cannot be
assessed by this study. It could have a large or a small caloric impact, but that remained
unknown. Measuring these impacts on whale’s diet, caloric intake, physiological stress,
and other factors would be extremely valuable in future work.
5.6 Final Thoughts
By now, as the reader, I hope you have learned something. If not, I hope this has
provided you with the perfect document to put you to sleep!
Back to the science, even while knowing the limitations and weaknesses of this
study, I am confident in the results of this study. In 2020, gray whales in the North Puget
57
Sound speed, inter-breath intervals, direction, and deviation indices were different when
boats were present within 1000 m or less of the whale to when there were no boats
present within 1000 m of the whale. While there are many unanswered questions, as
noted above, the data showed a significant trend. With my coauthors’ help, we hope to
publish a version of this thesis in a scientific journal.
58
6. Bibliography
Burnham, R., & Duffus, D. (2019). Gray Whale Calling Response To Altered
Soundscapes Driven By Whale Watching Activities In A Foraging Area. 23.
Calambokidis, J., Darling, J. D., Deecke, V., Gearin, P., Gosho, M., Megill, W.,
Tombach, C. M., Goley, D., Toropova, C., & Gisborne, B. (2002). Abundance,
range and movements of a feeding aggregation of gray whales (Eschrichtius
robustus) from California to southeastern Alaska in 1998. 10.
Calambokidis, J., Perez, A., & Laake, J. (2019). Updated analysis of abundance and
population structure of seasonal gray whales in the Pacific Northwest, 1996-2017
(p. 72).
Cascadia Research Collective [CRC]. (2022). Unpublished Information and Data
Regarding Gray Whales and the Sounders [Personal communication].
Cascadia Research Collective [CRC]. (2022, April). North Puget Sound Gray Whales
(Sounders) History.
Christiansen, F., Rasmussen, M., & Lusseau, D. (2013). Whale watching disrupts feeding
activities of minke whales on a feeding ground. Marine Ecology Progress Series,
478, 239–251.
Duffield, D. A. (2009). Extinctions, Specific, IV Extinct Population, A Atlantic Gray
Whale. In Encyclopedia of Marine Mammals (Second, pp. 402–404). Academic
Press.
Fearnbach, H., & Durban, J. (2022, April). 2022 Condition Assessment of “Sounders”
gray whales. SR3 Sealife Response, Rehabilitation, and Research Improving the
Health of Marine Wildlife.
59
Fournet, M., Matthews, L., Gabriele, C., Haver, S., Mellinger, D., & Klinck, H. (2018).
Humpback whales Megaptera novaeangliae alter calling behavior in response to
natural sounds and vessel noise. Marine Ecology Progress Series, 607, 251–268.
Gailey, G., Sychenko, O., McDonald, T., Racca, R., Rutenko, A., & Bröker, K. (2016).
Behavioural responses of western gray whales to a 4-D seismic survey off
northeastern Sakhalin Island, Russia. Endangered Species Research, 30, 53–71.
Google Earth. (2020a). Location of Hat Island in Puget Sound (V. 7.3.4.8573) [Satelite
Image].
Google Earth. (2020b). Location of Hat Island relative to the Pacific Northwest (V.
7.3.4.8573) [Satelite Image].
Google Earth. (2020c). Location of Observation Posts on Hat Island (V. 7.3.4.8573)
[Satelite Image].
Lumley, T., Diehr, P., Emerson, S., & Chen, L. (2002). The Importance of the Normality
Assumption in Large Public Health Data Sets. Annual Review of Public Health,
23(1), 151–169.
Lusseau, D., & Bejder, L. (2007). The Long-term Consequences of Short-term Responses
to Disturbance Experiences from Whalewatching Impact Assessment. 10.
National Marine Fisheries Service [NMFS]. (2020a, January 9). Gray Whale. NOAA.
National Marine Fisheries Service [NMFS]. (2020b, January 9). Marine Life Viewing
Guidelines. NOAA.
National Marine Fisheries Service [NMFS]. (2022, June 3). 2019-2022 Gray Whale
Unusual Mortality Event along the West Coast and Alaska (Alaska, West Coast).
NOAA.
60
National Marine Fisheries Service [NMFS]. (2021a, February 1). Gray Whale Population
Abundance (West Coast). NOAA.
National Marine Fisheries Service [NMFS]. (2021b, December 7). West Coast Gray
Whales Declined During Unusual Mortality Event, Similar to Past Fluctuations in
Numbers (Alaska, West Coast). NOAA.
Osborne, R., Calambokidis, J., & Dorsey, E. M. (1988). A guide to marine mammals of
Greater Puget Sound (D. Haley, Ed.). Island Publishers.
Piwetz, S., Gailey, G., Munger, L., Lammers, M. O., Jefferson, T. A., & Würsig, B.
(2018). Theodolite Tracking in Marine Mammal Research: From Roger Payne to
the Present. Aquatic Mammals, 44(6), 683–693.
Pruitt, C., & Donoghue, C. (2016). Ghost shrimp: Commercial harvest and gray whale
feeding, North Puget Sound, Washington. (p. 30) [Technical].
Senigaglia, V., Christiansen, F., Bejder, L., Gendron, D., Lundquist, D., Noren, D.,
Schaffar, A., Smith, J., Williams, R., Martinez, E., Stockin, K., & Lusseau, D.
(2016). Meta-analyses of whale-watching impact studies: Comparisons of
cetacean responses to disturbance. Marine Ecology Progress Series, 542, 251–
263.
Sullivan, F. A., & Torres, L. G. (2018). Assessment of vessel disturbance to gray whales
to inform sustainable ecotourism: Vessel Disturbance to Whales. The Journal of
Wildlife Management, 82(5), 896–905.
Sumich, J. L. (1983). Swimming velocities, breathing patterns, and estimated costs of
locomotion in migrating gray whales, Eschrichtius robustus. Canadian Journal of
Zoology, 61(3), 647–652.
61
Sumich, J. L. (2014). E. robustus: The biology and human history of gray whales.
Swartz, S. L., & Jones, M. L. (2016). Gray Whale (Eschrichtius robustus). In Marine
Mammal Encyclopedia (3rd ed., p. 25). Academic Press, Inc.
U.S. Geological Survey [USGS]. (1993). Hydrographic Map Around Hat Island
[Topographic M].
Weinrich, M., & Corbelli, C. (2009). Does whale watching in Southern New England
impact humpback whale (Megaptera novaeangliae) calf production or calf
survival? Biological Conservation, 142(12), 2931–2940.
Williams, R., Trites, A. W., & Bain, D. E. (2002). Behavioural responses of killer whales
(Orcinus orca) to whale-watching boats: Opportunistic observations and
experimental approaches: Behavioural responses of killer whales to whalewatching. Journal of Zoology, 256(2), 255–270.
Wladichuk, J., D. Hannay, A. MacGillivray, Z. Li. 2018. Whale Watch and Small Vessel
Underwater Noise Measurements Study: Final Report. Document 01522, Version
3.0. Technical report by JASCO Applied Sciences for Vancouver Fraser Port
Authority ECHO Program.
Würsig, B., Cipriano, F., & Würsig, M. (1991). Dolphin movement patterns: Information
from radio and theodolite tracking studies. In Dolphin societies: Discoveries and
puzzles (pp. 79–111). Berkeley: University of California Press.
62
