-
Title
-
SEABIRD INDICATORS FOR CHERRY POINT AQUATIC RESERVE: INTEGRATING COMMUNITY SCIENCE DATA INTO MARINE CONSERVATION
-
Date
-
2022 June
-
Creator
-
Stehr, Erin
-
Identifier
-
Thesis_MES_2022Sp_StehrE
-
extracted text
-
SEABIRD INDICATORS FOR CHERRY POINT AQUATIC RESERVE:
INTEGRATING COMMUNITY SCIENCE DATA INTO MARINE CONSERVATION
by
Erin Stehr
A Thesis
Submitted in partial fulfillment
Of the requirements for the degree
Master of Environmental Studies
The Evergreen State College
June 2022
©2022 by Erin Stehr. All rights reserved.
This Thesis for the Master of Environmental Studies Degree
by
Erin Stehr
has been approved for
The Evergreen State College
by
_______________________________
Kevin Francis, Ph. D.
Member of Faculty
_______________________________
Date
ABSTRACT
Seabird Indicators for Cherry Point Aquatic Reserve: integrating community science data into
marine conservation
Erin Stehr
Seabirds are ecosystem indicators currently used in the Salish Sea to track progress
towards management goals and monitor the status of marine habitats. Depending on available
data, life history, and regional trends, certain seabird species are better indicators than others.
Cherry Point Aquatic Reserve (CPAR), one of eight Aquatic Reserves managed by the
Department of Natural Resources Aquatic Reserves Program, encompasses 3,050 acres of
nearshore habitats in the eastern Strait of Georgia. Community scientists have collected seabird
data at CPAR since April 2013 at three shore-based locations. While this data was collected to
inform CPAR management, it had never been reviewed, analyzed, or incorporated into Aquatic
Reserves Program frameworks. This thesis thoroughly reviews the CPAR dataset and creates
replicable data management, quality control and analysis methods by which the Aquatic
Reserves Program can better incorporate other community science efforts. Additionally, this
thesis discusses the application of seabird indicators to small management areas and recommends
that Surf Scoter, Pelagic Cormorant, and a forage fish specialist like Pacific or Red-throated
Loon be the focus of ongoing monitoring and analyses to best track CPAR ecosystem health and
resilience. Finally, this thesis outlines recommendations for other agencies and groups that may
wish to improve their seabird data collection processes and data quality. Salish Sea conservation
and restoration efforts increasingly incorporate community science data to strengthen
conservation outcomes. This thesis occupies the space where the objectives of management
agencies, academic researchers, and volunteer/community scientists overlap and provides a
model for the effective conservation and management of marine habitats.
Key Words: seabirds, indicators, marine conservation, Aquatic Reserves, collaborative
management, community science, Salish Sea, Puget Sound
CONTENTS
List of Figures ................................................................................................................................ vi
List of Tables ................................................................................................................................ vii
Acknowledgements ...................................................................................................................... viii
Introduction ..................................................................................................................................... 1
Literature Review............................................................................................................................ 6
Definitions of the Salish Sea and Puget Sound ........................................................................... 7
What is an indicator? ................................................................................................................... 8
Seabirds as ecosystem indicators ................................................................................................ 9
Seabird indicators used by the Puget Sound Partnership ..................................................... 10
Limitations and a ‘coarse’ approach .................................................................................... 12
Seabirds as fisheries indicators ................................................................................................. 12
Seabirds and herring ............................................................................................................. 13
Cherry Point Aquatic Reserve ................................................................................................... 16
DNR Aquatic Reserves Program ........................................................................................... 17
Applying seabird indicators to CPAR ....................................................................................... 18
Salish Sea seabird trends ........................................................................................................... 20
MESA historical baseline ...................................................................................................... 20
Recent literature using the MESA baseline ........................................................................... 22
Community science seabird monitoring efforts ........................................................................ 26
Interpreting Salish Sea seabird trends ....................................................................................... 28
CPAR marine bird survey effort ............................................................................................... 32
Purpose and value of this thesis ................................................................................................ 33
Methods......................................................................................................................................... 36
Site description and survey methods ......................................................................................... 36
Data management ...................................................................................................................... 40
Calculating seabird density ....................................................................................................... 42
Data exploration and analyses ................................................................................................... 42
Results and Discussion ................................................................................................................. 44
Surf Scoter ................................................................................................................................. 50
Common Loon........................................................................................................................... 54
Horned Grebe ............................................................................................................................ 57
iv
Pelagic Cormorant ..................................................................................................................... 58
Bufflehead ................................................................................................................................. 60
Brant .......................................................................................................................................... 61
Western Grebe ........................................................................................................................... 62
All targeted species ................................................................................................................... 64
Feeding guilds ........................................................................................................................... 66
Conclusions ................................................................................................................................... 71
Future inquiries and applications .............................................................................................. 72
Recommendations for community science seabird efforts ....................................................... 74
Final summary ........................................................................................................................... 77
Notes ............................................................................................................................................. 78
References ..................................................................................................................................... 79
Appendices .................................................................................................................................... 85
Appendix A - Data Sheets ......................................................................................................... 85
v
LIST OF FIGURES
Figure 1. Map of the Salish Sea and Puget Sound .......................................................................... 8
Figure 2. Herring Spawn Depostion Maps from 1970 to 2021..................................................... 15
Figure 3. Map of All Aquatic Reserves and a Close-up of CPAR ................................................. 18
Figure 4. MESA Vessel-based, Aerial, and Shore-based Survey Locations ................................. 22
Figure 5. PSAMP Aerial Surveys that Overlapped with MESA Aerial Surveys ........................... 23
Figure 6. WWU Shoreline Census and Ferry Survey Locations ................................................... 24
Figure 7. Padilla Bay Shoreline Survey Points and Vessel-based Survey Routes ........................ 25
Figure 8. Map of Current CPAR Seabird Survey Locations and Survey Areas ............................ 36
Figure 9. Diagram of CPAR Seabird Field Survey Operating Procedure ................................... 38
Figure 10. Total CPAR Survey Time and Total Seabird Counts Per Season ............................... 45
Figure 11. Presence and Absence of All Species Targeted by the CPAR Seabird Surveys .......... 46
Figure 12. Encounter Rate for All Species Targeted by the CPAR Seabird Surveys .................... 47
Figure 13. Mean Density for All Species Targeted by the CPAR Seabird Surveys ...................... 48
Figure 14. Surf Scoter Density by Season and Month .................................................................. 50
Figure 15. Surf Scoter Density by Season and Month Focusing on Summary Statistics .............. 51
Figure 16. Surf Scoter Density Across Each Season..................................................................... 53
Figure 17. Common Loon Density by Season and Month ............................................................. 54
Figure 18. Mean Loon Density by Season .................................................................................... 56
Figure 19. Horned Grebe Density by Season and Month ............................................................. 57
Figure 20. Pelagic Cormorant Density by Season and Month ..................................................... 58
Figure 21. Bufflehead Density by Season and Month ................................................................... 60
Figure 22. Brant Density by Season and Month ........................................................................... 61
Figure 23. Western Grebe Density by Season and Month ............................................................ 62
Figure 24. Total Seabird Density by Season and Month .............................................................. 64
Figure 25. Mean Density of Original Seven Species by Season ................................................... 66
Figure 26. Species Composition of the Four Feeding Guilds ....................................................... 67
Figure 27. Mean Piscivore Density by Season and Month ........................................................... 68
Figure 28. Mean Benthivore Density by Season and Month ......................................................... 69
vi
LIST OF TABLES
Table 1. Percent Change to Seabird Abundance According to Five Relevant Studies for all
Seabird Species Currently Surveyed at CPAR. ............................................................................. 30
Table 2. All Seabird Species Targeted by the CPAR Survey Effort Including Their Common
Name, Scientific Name, Four-letter Species Code, Feeding Guild, and the Month/Year of
Addition to the Targeted Species List. .......................................................................................... 39
Table 3. Assessment Criteria for the Pass/Fail Test Used to Removed Low Quality Data Before
Analysis. ........................................................................................................................................ 41
Table 4. Common Errors, Recommended Solutions, and Actions Taken for the Data Collection
and Review Processes Associated with the CPAR Marine Bird Surveys...................................... 74
Table 5. Recommendations to Improve the Scale and Application of Marine Bird Monitoring
Efforts Including the Pros and Cons Associated With Each Recommendation. ........................... 75
vii
ACKNOWLEDGEMENTS
Cherry Point Marine Bird Survey Volunteers
Thank you all for the data that made this thesis possible: Amy Mower, Anne Moore, Annie
Provost, Andrea Warner, Barbara Rofkar, Ben Albers, Bill Beers, Christine Hanley, Christy
Mann, Casey McGee, Charis Weathers, Doug Brown, Doug Couvelier, Deborah Kaye, Dean
Rofkar, Doug Stark, Darrel W., Gayle Stebbings, E. Kilanowski, Eleanor Hines, Judy Akins,
Jane Aspines, Jacqueline Anderson, John Aspines, John Bowers, Jana Turner, Jonathan Hucke,
Jo Wedon, John Yearsly, Lilya Jaeren, Kali Klotz-Brooks, Katie Novak, Katrina Poppe, Kristin
Anderson, Katie Ayres, Kristine Margaret, Kelley Palmer-McCarty, Lyle Anderson, Mary
Blackstone, Mary Durbrow, Margarette Grant, Marg Leone, Melanie Lichterman, Melissa
McBride, Makie Matsumo-Hervol, Mary Mueller, Maru Rawlins, Matt Schwartz, Michelle
Landis, Mike Senett, Natalie Lord, Pam Borso, Pat Britain, Patrice Clarck, Phile Calise, Peter
Neubeck, Pauline Sterin, Paul Woodcock, Ryan Campbell, Ryan Sarhan, Robert Kaye, Rondi
Nordal, Sue Euler, Stephen harper, Susan Gardner, Sue Parrot, Skylar Summer, Twink Coffman,
Tami DuBow, Trevor Robinson, Victoria Souze, Veronica Wisniewski, Wendy Steffensen.
RE Sources
Thank you to Eleanor Hines and Rondi Nordal for the data stewardship and access.
MES Program
Thank you to the Bilezikian Family for the financial support of your fellowship.
Thank you to all the faculty and peers within MES. Special appreciation to Melissa Sanchez and
Lisa Hillier for your reviews, humor, and friendship.
So many thanks to Kevin Francis who, from the beginning, made me feel welcome and valued
within the MES program. Your support and perspective through this process kept me sane.
DNR Aquatic Reserves Program
Thank you to the staff. Birdie Davenport for your personal support and the program-wide
support you bring in your wake. Allison Brownlee for your data science brain and R advice.
Erica Bleke for your insightful reviews and commiseration. Betty Bookheim for your feedback.
And all the Puget SoundCorps who I have supervised while juggling my thesis.
Family and Friends
Thank you to my family. You make so many things possible with your love and support. Thank
you to my friends for your support and understanding of my social absences. And thank you to
my partner for being both and more.
viii
INTRODUCTION
As the human population has increased and altered local environments, Salish Sea
ecosystems and habitats have declined. Marine management and monitoring groups across Puget
Sound and the Salish Sea focus on restoration and conservation, but marine ecosystems are
interconnected, complex and hard to measure. To track progress towards management goals and
assess ecosystem health, scientists have identified measurable variables that function as
indicators for the underlying system. Marine bird, or seabird, abundance is one indicator of
ecosystem health, biodiversity, and resilience. Long-term datasets on indicators are hard to
maintain for many agencies due to funding restrictions or changes in agency direction.
Community science can fill these data gaps and seabird data in particular is being incorporated
into analyses and literature (Toft et al., 2017).
Marine birds are effective indicators since they are well studied and their abundance and
distribution reflect underlying prey availability and habitat health (Gaydos & Pearson, 2011). As
top predators in marine systems, seabirds are controlled by bottom-up processes (McLeod et al.,
2009; Piatt, Sydeman, et al., 2007). At its most simplified, poor habitat leads to low prey
availability which results in fewer birds. Indicators respond to habitat change and perturbation in
different ways. Seabirds are non-specific and lagging indicators, meaning that their abundance
responds to widespread change and a delay exists between that change and seabird response.
In the Salish Sea, seabird abundance has declined precipitously since the first
comprehensive baseline surveys conducted in the late 1970s (Wahl et al., 1981) and have
continued to decline since the 1990s (Vilchis et al., 2015). Historical seabird populations were
likely even larger as industrialization and colonization of the Salish Sea were well underway by
the time of the 1970s baseline surveys (Bower, 2009). Now areas of particularly high seabird
1
density have shifted as continued urbanization impacts prey and habitat resources. Four local
seabird species are now listed at the state or federal level: Marbled Murrelet (state endangered,
federally threatened), Common Loon (state sensitive), Western Grebe (state candidate) and
Tufted Puffin (state endangered). Federal, Tribal, state, and community organizations are a few
of the entities collecting data to help track seabirds in the Puget Sound and Salish Sea.
To combat further species and habitat declines, the Puget Sound Partnership was created
and tasked with leading the effort to restore and protect Puget Sound. They compile data from
community scientists, Tribes, and other state agencies to track “vital signs” or indicators. Seabird
abundance is one of these vital signs. Community science efforts like the British Columbia
Coastal Waterbird Surveys, Puget Sound Seabird Surveys and Salish Sea Guillemot Network are
a few of the many community science networks providing valuable data far beyond any single
agency’s capability and scientists are increasingly facing the happy challenge of incorporating
this kind of data into monitoring and decision-making frameworks. Community science can be
particularly valuable for small programs such as Washington State Department of Natural
Resources (DNR) Aquatic Reserves Program.
DNR Aquatic Reserve’s Program manages eight Aquatic Reserves, each of which was
established for their scientific, natural, and cultural importance. One of the reserves is Cherry
Point Aquatic Reserve (CPAR). CPAR encompasses 3,050 acres of intertidal and subtidal
habitats in the eastern Strait of Georgia along the western coast of Whatcom County in
Washington State. This reserve has dedicated community involvement in the form of a
Community Stewardship Committee and several self-organized monitoring efforts, one of which
is a seabird survey.
The CPAR marine bird surveys were established in 2013 with the intent of providing
2
seabird abundance data specific to the Aquatic Reserve to inform reserve managers. The
community scientists conducting these surveys are self-organized and incredibly dedicated. RE
Sources, a nonprofit in Bellingham, Washington, is the steward for this data and the community
organizer for the Stewardship Committee. The CPAR Birders conduct monthly surveys at three
locations along the Cherry Point shoreline. They organize training for new members and
implement a consistent methodology developed from the historical Marine Ecosystems Analysis
(MESA) surveys conducted in 1978/79 (Bower, 2009). The resulting dataset is now a valuable
source of information spanning nine years and including 29 marine bird species.
In this thesis I attempt to answer one main question: How can this small data set,
focusing on a specific area, and collected by community scientists, inform conservation and
restoration of marine ecosystems? I take the three-pronged approach to citizen science analysis
suggested by Toft et al. (2017) incorporating three potential audiences for these analyses:
volunteers, managers, and scientists. Each audience has different but overlapping objectives.
Volunteers (i.e., community scientist) may be most interested in which species are most often
encountered and when. Managers may be interested in which species or groups are the best
indicators for this area and warrant continued focus. Scientists, or academic biologists, may be
interested in changes to marine bird and/or individual species density over the duration of this
survey effort.
Before I could begin analyses, I restructured the database, referenced old scans, and
wrangled the data to make it more accessible to data scientists. Data visualization and analysis
identified species most often encountered at Cherry Point. I looked at change between and over
seasons and months, grouped seabirds by feeding guild (piscivore, herbivore, benthivore,
omnivore and planktivore) (Bower, 2009), and looked for shifts in migration timing. The
3
combined efforts of the data collectors and myself establish a status report on CPAR marine
birds which can be used as a baseline for future comparison. I identify areas for further analysis
and inquiry, suggest methodological alterations to increase the quality of the dataset, and create a
system by which the CPAR bird data is accessible to Aquatic Reserve Program staff and
managers. In this way, CPAR seabird data can be used to inform adaptive management and track
progress towards management goals.
While seabirds can indicate environmental health, biodiversity, habitat condition, and
climate change on a large scale (Pearson & Hamel, 2013), using them as indicators for smaller
areas like aquatic reserves has limitations. Certain species may be better than others as indicator
selection is based on life history, particular ties to the management area and availability of data.
Many seabird species are migratory and have mixed life history which means that declines in
some species may be due to degradation of the lakes they depend on for breeding grounds (i.e.,
Common Loon), a decrease in old growth trees required for nesting (i.e., Marbled Murrelet), or
negative impacts to other areas of their migratory route.
Despite the concerns around migratory species as indicators, scoters may be a valuable
indicator for Cherry Point. The Cherry Point herring stock used to be an unrivaled resource for
local seabird populations, but since the 1970s this herring stock has critically declined. Surf
Scoter are particularly tied to herring spawn (Boyd et al., 2006, Lok et al., 2012) and during this
same time, the population of Surf Scoters foraging at Cherry Point has declined by 90%
(WDNR, 2010). The duration of time Surf Scoters spend at CPAR has also decreased with the
amount of spawn (Sandell presentation for the Cherry Point Implementation Meeting, Nov. 30,
2021).
At a little more than 12 km2, Cherry Point Aquatic Reserve is a small part of the entire
4
18,000 km2 Salish Sea, but with the extra monitoring and support provided by community
scientists, the information available to managers is magnified. This thesis examines the
application of seabird indicators to small management areas and provides an example to
community and agency scientists alike on how to incorporate community driven data into
monitoring and decision-making frameworks. In this way, this thesis contributes to the overall
effort to restore and conserve the Salish Sea.
5
LITERATURE REVIEW
Seabirds are ecosystem indicators used to represent the resilience and function of the
underlying marine system. Their abundance can indicate prey availability, habitat suitability,
exposure to pollutants and ecosystem stressors such as algal blooms or warm water events.
Seabird survival and abundance reflects the structure and function of the marine environment
(Pearson & Hamel, 2013). They are a diverse group foraging at different trophic levels and
utilizing various habitats. Some seabirds specialize in a specific habitat or prey source which ties
them to that resource’s availability. Other seabirds are generalists relying on many prey and
habitat types. While some seabirds are year-round residents of the Salish Sea, many are migrants
that rely on Salish Sea resources for only a portion of their life history.
The first section of this literature review discusses the use of seabirds as indicators and
two main applications for their use: focusing on ecosystems and focusing on fisheries. Then it
examines the application of seabird indicators to small management areas like Cherry Point
Aquatic Reserve (CPAR) and suggests theoretically appropriate seabird indicators for CPAR
using information from current ecosystem indicator publications and documented seabird
response to herring spawn events.
The second part of this literature review explores seabird trends in the Salish Sea and
what these trends may mean for aquatic habitats. Understanding these trends can help
contextualize results from this thesis’ data exploration and analyses. The Marine Ecosystem
Analysis (MESA) seabird surveys are the historical baseline for current seabird abundance
studies in the Salish Sea. Those methods have developed into the protocols currently
implemented by community scientists at CPAR.
6
Finally, this review examines literature on the strengths and weaknesses of community
science data. The purpose of this thesis is to both evaluate the data and create a framework for
the Aquatic Reserve Program by which this kind of community science effort can be better
incorporated and used to inform management decisions.
Definitions of the Salish Sea and Puget Sound
The referenced literature refers to two main general locations: the Salish Sea and Puget
Sound. These are overlapping areas that encompass the inland waterways of Washington State
and British Columbia. Puget Sound is the southern portion of the Salish Sea. It includes
Washington’s inland waters from the opening of Admiralty Inlet including the Whidbey Basin to
the North and East, down to Olympia in the South. The Salish Sea includes Puget Sound as well
as the San Juan Islands and Straits of Juan de Fuca and Georgia.
7
Figure 1
Map of the Salish Sea and Puget Sound
Note. The Salish Sea is outlined in light blue and the dark blue fill denotes Puget Sound. Boundaries were
created using definitions from the Encyclopedia of Puget Sound (2015).
What is an indicator?
Indicators serve as quantitative proxies for ecological processes (e.g. energy flow) or
ecosystem state (e.g. biodiversity) (Kershner et al., 2011; Tam et al., 2017). Systems are
complex, interconnected, and hard to measure. Indicators are easier to measure and reflect the
function of the underlying system. Ecosystem managers select a portfolio of indicators specific
to their management goal(s) and use those indicators to assess management efficacy and inform
adaptive strategies.
Different management goals may require different indicators. Biological indicators are
applied to goals like increased biodiversity, habitat resilience and robust food webs. Social
8
indicators are better applied to goals like increased community involvement and participation
(Levin et al., 2009). Biological indicators can be diagnostic and track a few key attributes or be
nonspecific and track many attributes. They may respond quickly to perturbations and changes
and be early-warning indicators, or they may respond slowly and be retrospective.
The Puget Sound Partnership (PSP), the agency tasked with restoring and conserving
Puget Sound, is a local example of ecosystem-based management. PSP has identified five Puget
Sound Recovery Goals and 13 indicators that that they refer to as vital signs (McManus et al.,
2020). These 13 vital signs correspond to five main goals spanning physical, biological, and
human processes. The vital signs are a suite of complementary indicators identified by PSP to
best track Puget Sound recovery. Birds are just one vital sign used to track progress towards the
goal of thriving species and food web (McManus et al., 2020). Suites of complimentary
indicators may be the best way to approach complex ecosystems, but that effort is beyond the
scope of any single thesis or scientist.
Seabirds as ecosystem indicators
As higher trophic-level species, seabirds are controlled by bottom-up processes (McLeod
et al., 2009; Piatt, Sydeman, et al., 2007). Poor habitat leads to low prey availability which
results in fewer birds. Seabird biomass, or abundance, can reflect the biomass of lower trophiclevel organisms in Puget Sound (Harvey et al., 2012). Seabirds can thus indicate the
consequences of ecosystem trends related to climate change and other anthropogenic
disturbances (Piatt, Sydeman, et al., 2007). Warming waters, acidification and hypoxic zones
affect plankton and forage fish, which propagates up the food chain and is reflected by seabird
numbers. Between 2014 and 2016 a marine heat wave known as the ‘Blob’ hit the West Coast of
North America. It reduced the biomass of phytoplankton which altered the zooplankton
9
community to be less nutritional for forage fish (Piatt et al., 2020). Forage fish numbers
decreased and also became less nutrient rich. The scarcity and nutritional deficiency of forage
fish lead to a mass mortality of Common Murre. Between summer 2015 and spring 2016 over
60,000 Common Murre washed up on the beaches of Washington and Oregon in varying states
of starvation (Piatt et al., 2020). Thus, seabirds can provide valuable feedback on marine
ecosystem trends and anthropogenic activities that impact the environment.
Some seabird species are better indicators than others depending on their feeding
behavior, migratory habits, and other aspects of their life history (Harvey et al., 2012). Resident
diving birds such as cormorants and alcids; migratory diving birds such as grebes, mergansers,
and loons; some local and migratory gull species; and nearshore diving birds such as scoters,
goldeneye and bufflehead are all good lagging indicators. The term “lagging” refers to the delay
between the initial habitat change and a seabird response. Other species such as bald eagles and
dabbling ducks are poor indicators due to their lack of correlation with their prey groups (Harvey
et al., 2012). Feeding behavior also influences indicator quality. Aerial and surface feeders (i.e.,
gulls, terns, dabbling ducks) forage over large areas but rely on prey being available near the
surface and are therefore susceptible to vertical changes in prey density. Pursuit divers (i.e.,
alcids, cormorants, scoters etc.) are more able to cope with vertical changes but are vulnerable if
prey spread out over a wider horizontal area (Boyd et al., 2006). Seabirds that expend more
energy when foraging are more susceptible to changes in prey abundance, namely large-bodied
diving birds (Vilchis et al., 2015).
Seabird indicators used by the Puget Sound Partnership
In Puget Sound, researchers use seabirds as indicators of food web structure and
ecosystem resilience. The Puget Sound Partnership (PSP) is a state agency created in 2007 to
10
oversee the regional efforts to restore and conserve Puget Sound. PSP uses indicators to track
progress and restoration success. Kershner et al. (2011) focused on one goal of the Partnership
which is “healthy and sustaining populations of native species in Puget Sound, including a robust
food web” (p. 3). They identified a portfolio of indicators that could provide feedback to
managers across different temporal and spatial scales. Kershner et al. (2011) identified seven
indicators including non-breeding marine bird population size estimates. They identify seabird
abundance as a non-specific, retrospective indicator of ecosystem function.
Pearson et al. (2013) selected specific seabird indicator species for Puget Sound Vital
Signs, which is the monitoring component of the PSP. Since Kershner et al. (2011) established
that seabird abundance was a food web indicator, Pearson et al. (2013) selected specific seabird
species that have established monitoring efforts, are abundant and well distributed, and have
significant reliance on Puget Sound resources. ‘Significant reliance’ means that they consume
almost exclusively marine resources and spend most of their time in Puget Sound. This excludes
species that use both marine and freshwater ecosystems (i.e., Great Blue Heron, Double-crested
Cormorant, and loons). Pearson et al. (2013) recommends three resident species that breed in
Puget Sound (Pigeon Guillemot, Rhinoceros Auklet and Marbled Murrelet) and one overwintering species group (scoters).
For an indicator to be useful there must be a link between the population status and local
conditions. Overall, a considerable amount of migrating seabirds’ life is spent outside the Salish
Sea, therefore outside conditions, as opposed to local environmental conditions, may be driving
trends. Despite being migratory, scoter species also have an established link to the Salish Sea.
They have site fidelity, returning to the same molting locations year after year (de la Cruz et al.,
2009). Molting is energetically taxing, and scoters are particularly reliant on local habitat areas
11
and prey resources for successful primary feather regrowth. Additionally, some juveniles and
non-breeding individuals do remain in Puget Sound throughout the year. In this way, Puget
Sound has an exacerbated effect on scoter fitness compared to other migratory winter bird
species (Crewe et al., 2012). This increased reliance on Puget Sound habitats makes scoters good
indicators even though they are not year-round residents.
Limitations and a ‘coarse’ approach
There are limitations to any single indictor, especially a highly mobile top predator such
as seabirds. Abundance estimates can be highly variable, and it may take years of data to detect
trends (Boyd et al. 2006, Wahl et al., 1981). Therefore Pearson et al. (2013) selected only species
which already had long-term abundance data in Puget Sound. Even then, they refer to their
approach as “coarse-grained” (p. 3) with the intent to indicate trends in Puget Sound-dependent
bird populations which may reflect a long-term view of Puget Sound health. This ‘coarseness’
also makes it challenging to identify reasons for change in seabird abundance. A decreasing
abundance trend may reflect decreased habitat health but does not provide information on why
habitat degradation is occurring. Indicator species that have a foraging preference, will diversify
if their preferred prey becomes scarce (Boyd et al. 2006). Seabird abundance is unlikely to
respond to anything less than a large-scale change in many different prey options which requires
a sweeping impact to the ecosystem. That being said, seabird behavior is highly responsive to
ecosystem shifts (Montevecchi, 1993).
Seabirds as fisheries indicators
The idea of seabirds as fisheries indicators has been around since the 1980s (Cairns,
1987; Montevecchi, 1993; Piatt, Harding, et al., 2007; Piatt, Sydeman, et al., 2007). Seabirds
consume prey at multiple trophic levels and in areas that are otherwise challenging for fisheries
12
managers to survey. Incorporating seabird data into larger fisheries models allows managers to
see how fisheries impact the ecosystem at large. Seabird abundance, behavior and survival are all
indices that may reflect prey availability (Sydeman et al., 2017).
For fisheries indicators, the piscivorous (fish-eating) species are most appropriate. Many
of these species rely on forage fish, which are the group of small silver fish that form an essential
link between primary consumers like plankton and higher trophic-level consumers, such as larger
fish, birds, and marine mammals. Specific to herring, a 1999 seabird predation report (Bishop &
Green, 2001) created a bioenergetics model for spawn consumption that allowed Alaska
Department of Fish and Game to adjust their adult herring spawner biomass estimates. By
monitoring Glaucous-winged Gull aggregations, managers could better estimate the amount of
spawn consumed by avian predators and incorporate it into their model.
Similar to the application of seabird indicators to ecosystems, seabird indicators for
fisheries is a coarse approach and some species may be better suited than others (Sydeman et al.,
2017). A challenge that leads to this coarseness is that there are competing forces in play. There
may be direct competition between seabirds and fisheries that target forage fish species. Other
fisheries may remove competition since they often target higher trophic-level fish species that
would otherwise consume the same prey as seabirds. Removing these larger fish can alleviate
predation pressure on these prey sources allowing for more availability to seabird predators.
Seabirds and herring
Whether it is an ecosystem or fishery stock, seabird abundance reflects large changes, not
small events, or availability of specific prey sources. One exception to this may be Pacific
herring (Clupea pallasii) spawn. Herring are forage fish that mass reproduce by laying eggs on
submerged aquatic vegetation, particularly eelgrass and macroalgae, forming hotspots of
13
potential food for predators such as seabirds. Each spawning event lasts around three to five
weeks (Lewis et al., 2007) and, generally, the higher the latitude the later the spawn event (Lok
et al., 2012). Herring in Puget Sound spawn from January to June (Sandell et al., 2019).
Washington State Department of Fish and Wildlife (WDFW) customarily monitors the 21
distinct Pacific herring stocks in Puget Sound (Sandell et al., 2019). Each stock returns to the
same locality each year to spawn, which means that there is minimal individual movement
between localities. Stocks that are depleted receive few recruits from other populations, making
recovery more challenging. One struggling stock is the Cherry Point herring stock which is
genetically distinct and spawns later (typically late April to mid-June) than any other Pacific
herring stocks in Washington State. From 1973 to 2016, the Cherry Point herring stock declined
from 13,606 tonnes to 468 tonnes of stock biomass, a 96% decrease (Sandell et al., 2019). Figure
2 illustrates the decline of the Cherry Point herring spawn deposition. Until the 1990s, it was the
largest herring stock in Washington State and supported the only commercial roe fishery in Puget
Sound. That roe fishery may have caused the initial population plummet (Gustafson et al., 2006).
Despite extensive research, the cause for their continued decline is unknown. Climate change,
pollution, changes to predator/prey dynamics or disease are all potential culprits (Sandell et al.,
2019). In 2003 there was an estimated 1,461 tonnes of Cherry Point herring, which was only half
of the population size WDFW estimated was needed for the population to rebound.
14
Figure 2
Herring Spawn Deposition Maps from 1970 to 2021
Note. Reprinted from “Cherry Point Environmental Aquatic Reserve Management Plan 2022 Update,”
maps by A. Brownlee, Unpublished, using data from WDFW and prepared for DNR.
Herring spawning events provide ephemeral, high density food pulses that attract many
seabird species, especially sea ducks such as scoters (Crewe et al., 2012; Lewis et al., 2007;
Wahl et al., 1981). During the winter months, scoters feed primarily on benthic invertebrates like
mollusks (clams, snails etc.), marine worms and crustaceans (Lewis et al., 2007). When herring
spawn, scoters form massive aggregations at the spawn sites. Of all marine bird species, scoters
exhibited the strongest response to herring spawn in Holmes Harbor off Whidbey Island and
were observed on the spawning grounds in much higher densities (Cleaver & Frannet, 1946).
Wahl et al. (1981) describe a flock of 25,000 scoters off Point Whitehorn at Cherry Point in
1978. Other seabird species have also been documented to exhibit an aggregate response to
herring spawn, including gulls (Crewe et al., 2012; Wahl et al., 1981), Pacific Loon (Crewe et
al., 2012; Wahl et al., 1981), Harlequin Duck (Crewe et al., 2012; Rodway et al., 2003), and
several other species not as commonly seen at Cherry Point such as Common Murre, Marbled
15
Murrelet, Brandt’s Cormorant (Crewe et al., 2012; Wahl et al., 1981), Surfbird, and Black
Turnstone (Bishop & Green, 1999).
With the plummet of herring spawn at Cherry Point there has been a correspondingly
drastic decrease in scoter abundance. The population of scoters foraging on herring spawn at
Cherry Point has declined from 60,000 in the 1970s to 6,000 in the early 2000s (WDNR, 2010;
citing unpublished data from Nysewander)1. The scoters were likely able to shift their foraging
locations or migratory behavior (Lok et al., 2012). Crew et al. (2012) noted an increase in scoter
abundance at herring spawn locations in the Canadian portion of the Salish Sea over the last two
decades. This is the kind of territory shift that Montevecchi (1993) predicted as an indicator of
change in prey abundance. Healthy and persistent herring stocks are key to food web and
ecosystem health (Sandell et al., 2019) and their decline is concerning. Wahl et al. (1981)
designated Cherry Point as an area of particular importance and vulnerability due to the
importance of Cherry Point herring as a prey source for seabirds.
Cherry Point Aquatic Reserve
In 2010, DNR established Cherry Point Aquatic Reserve. It encompasses 3,050 acres of
aquatic habitats in the southeastern Strait of Georgia. These habitats include cobble beaches,
submerged aquatic vegetation, extensive tidal flats, and a steep subtidal gradient into natural
deep-water channels that support local industry. There are three “cut-outs” (as seen in Figure 3)
within the reserve boundary to accommodate the piers and shipping activities of the British
Petroleum and Phillips 66 refineries, and the Petrogas distribution terminal. The refineries were
constructed between 1954 and 1971.The designation of CPAR limits any future uses of the
1
Vessel traffic increased with the construction of a third refinery in 1971. This is one of many other environmental
factors that could have contributed to the decline in Surf Scoters at Cherry Point.
16
shoreline that might adversely impact habitats and species identified as important by the
management plan (WDNR, 2022).
DNR Aquatic Reserves Program
The Aquatic Reserves Program within DNR was established in 2002. DNR manages all
state-owned aquatic lands for five main goals, one of which is to ensure environmental
protection. Aquatic Reserves are areas of special ecological importance and the Aquatic
Reserves Program is responsible for their identification, establishment, and ongoing management
(Palazzi & Bloch, 2006). Aquatic Reserve designation does not change public access meaning
that fishing, harvesting, and recreation are still allowed within the reserve boundaries.
Designation does emphasize the restoration and conservation of natural ecosystems and therefore
provides extra protection by limiting future activities that threaten nearshore environments. One
way that Aquatic Reserve designation protects aquatic habitats is by removing aquatic land from
future leasing limiting the construction of over-water or in-water structures like marinas, docks,
or pipelines, within the reserve. Aquatic Reserve designation also creates opportunities for
monitoring, research, education, and public engagement which leads to enhanced agency and
community driven data sets specific to each Aquatic Reserve.
Currently, there are eight aquatic reserves, of which seven are marine. Figure 3 shows the
locations of the eight Aquatic Reserves. Each reserve has a management plan created by DNR
managers with input from Tribal managers and local stakeholders, including community
members. There are various scientific monitoring activities conducted by DNR staff, interns, and
community scientists. An overarching goal for the Aquatic Reserves Program is the conservation
of native ecosystems and ecosystem services (Palazzi & Bloch, 2006). The management plans
17
include ecosystem-based conservation goals like those that the Puget Sound Partnership tracks
via ecosystem indicators.
Figure 3
Map of all Aquatic Reserves and a Close-up of CPAR
Note. The map of all eight Washington State Department of Natural Resources Aquatic Reserves was
provided courtesy of the Aquatic Reserves Program. The close-up of CPAR on the right illustrates the
cut-outs for industrial piers.
Applying seabird indicators to CPAR
One major difference between PSP and DNR Aquatic Reserves is spatial scale. DNR
Aquatic Reserves are small areas within the larger Puget Sound and Salish Sea ecosystems.
Reserve boundaries are drawn based on habitat factors and stakeholder agreement within the
context of the larger Salish Sea ecosystems. Marine environments, however, do not tend to have
discreet environmental boundaries. Marine habitats and ecosystems are spatially and temporally
dynamic (Hooker & Gerber, 2004). When it comes to applying indicators such as seabird
18
abundance to CPAR, the size of the reserve and the overlap of marine environments is an
important consideration. Especially for mobile species, abundance may not reflect conditions
within the reserve. Nevertheless, protection provided by Aquatic Reserves benefits local species
even if that species is highly mobile and the reserve does not cover its entire range (Boyd et al.,
2006).
Seabird abundance can still provide valuable feedback to reserve managers, especially if
the data is incorporated into a monitoring network. Puget Sound Vital Signs includes a suite of
indicators used to track ecosystem function, resilience, and food webs. The Aquatic Reserves
Program has many monitoring activities that target some of these same indicators such as seabird
distribution and abundance. Trends in overall abundance, specific species and species groups are
various approaches to seabird data. One way to group seabirds is by prey source or resource
dependency called “foraging guilds” (Anderson et al., 2009, p20). When focusing on the Cherry
Point area, there are specific species and guilds that may provide more information than others.
Selecting species that reflect the spatial scale of interest, whether a reserve or a fishery
stock, is an important step (Einoder, 2009; Pearson & Hamel, 2013). Due to CPAR’s small area
within the southern Strait of Georgia and Salish Sea ecosystems, aerial divers that forage across
large horizontal spaces (i.e., gulls and terns) are not as appropriate as indicators. Seabirds that
roost or nest within the reserve boundaries or species that spend a substantial amount of time
within reserve habitats are more tied to the reserve resources and are better selections. This
excludes most migratory species apart from scoters that molt in the Strait of Georgia and have
historically relied on the Cherry Point herring spawn. CPAR is comprised predominantly of
nearshore marine habitats, so species that typically forage in open water or are unlikely to use the
nearshore are also less appropriate.
19
Theoretically, the best potential seabird ecosystem indicators for Cherry Point Aquatic
Reserve are Pigeon Guillemot, Pelagic Cormorant and Surf Scoter. Pigeon Guillemot are pursuit
divers that forage in the nearshore. While they nest in bluffs on many parts of the Salish Sea
shoreline, they have not been observed along the Cherry Point reach (L. Anderson, personal
communication, April 16, 2022). Pelagic Cormorants, unlike Double-crested Cormorants, rely
solely on marine habitats and they interact heavily with anthropogenic structures. There are
currently three refineries at CPAR with piers that extend into cutouts within the Aquatic Reserve
boundary. Pelagic Cormorants nest and roost on these structures. They also prey on benthic
species and Pacific sand lance (Crewe et al., 2012). Considering the Cherry Point herring
population, Surf Scoters are valuable indicators, as are benthivore and piscivore feeding guilds.
Salish Sea seabird trends
MESA historical baseline
The first comprehensive study of seabirds in the Salish Sea was conducted for the Marine
Ecosystem Analysis (MESA) Puget Sound Project in 1978/79. At the time it was the most
extensive study of marine bird populations in Washington State and remains the most commonly
used baseline for evaluating seabird trends in the southern Salish Sea (Anderson et al., 2009;
Wahl et al., 1981). Surveys spanned the Strait of Juan de Fuca, southern Strait of Georgia, and
San Juan Islands (Figure 4). MESA surveys included vessel-based, aerial, and shoreline surveys.
Shoreline surveys included beach walks, dead bird surveys, and point census counts. It is
important to recognize that these surveys are used as a baseline only because there were no
widespread and reliable seabird abundance surveys before the 1970s. Based off anecdotal
evidence, and the impacts caused to major rivers and waterways through the Salish Sea due to
20
industrialization and colonization, seabird numbers were likely higher before the 1970s (Bower,
2009).
MESA looked at bird density rather than abundance. Wahl et al. (1981) divided each
region into subregions and used nautical charts to calculate the subregion area. This helped
control for locations that may have similar abundances but vastly different spatial scales. Small,
shallow bays and inlets had much higher densities than larger areas of open water. In addition to
this spatial variation, there was temporal variation in density as some seabirds roosted in the
same area that they foraged within, and others returned to a different nesting location. Therefore,
density varied depending on seabird activity at that time of day. Seasonally the highest bird
densities were seen in winter as migratory waterfowl arrived in the Salish Sea and in the spring
as seabirds aggregated at herring spawn locations, including Cherry Point.
With only two years of data, MESA scientists saw large annual variation and suggested
that this is normal. Five to ten years of data at minimum are needed to observe meaningful
annual trends (Wahl et al., 1981). In total, MESA surveys documented 116 seabird species.
While aerial and vessel-based surveys allowed for greater spatial coverage, point census surveys
have the advantage of time. Vessel-based and aerial surveys require quicker scans since the boat
or plane is constantly moving forward. Seabirds that are smaller bodied and/or that dive to avoid
disturbance are the most likely to be missed. Point census surveys are conducted from a shoreline
and record all seabirds visible on the water, at the water’s edge, or that fly by during the survey
time. There is no time limit, but the survey locations are limited by accessibility and deeperwater species are less likely to be recorded. All surveys underestimated true seabird abundance
but point census surveys get the closest to enumerating the true numbers with minimal error.
21
Figure 4
MESA Vessel-based, Aerial, and Shore-based Survey Locations
Note. Map of MESA vessel-based and aerial surveys on the left and shore-based census survey locations
on the right. Reprinted from “Marine bird populations of the Strait of Juan de Fuca, Strait of Georgia and
adjacent waters in 1978 and 1979,” by T. Wahl, S. Speich, D. Manuwal, K.V. Hirsch and C. Miller, 1981,
Report prepared for MESA Puget Sound Project, 19-20. Copyright (1981) by the United States
Environmental Protection Agency.
Seabird monitoring efforts have continued in Puget Sound and the larger Salish Sea since
the 1970s. State agencies, Tribes, and community groups have pursued different types of seabird
survey including aerial, vessel-based, shoreline, and point census methods. The surveys that are
easiest to replicate and therefore the most utilized, especially by community scientists, are the
point census surveys. Additionally, consistent methodology makes it easier to compare recent
data to the MESA baseline.
Recent literature using the MESA baseline
Nysewander et al. (2005) compared Puget Sound Ambient Monitoring Program
(PSAMP) aerial survey data from 1992 to 1999 to MESA’s aerial surveys from 1978/79. Except
for Harlequin Duck, all species with significant changes in density were in decline. Figure 5
illustrates the overlapping survey areas from the two studies. The loud plane utilized by the
PSAMP surveys may have scared diving seabirds into submerging which would result in them
22
being missed by the observers and therefore underrepresented in the data. On the other hand,
PSAMP surveys were not limited to shorelines and did cover deeper-water habitats.
Figure 5
PSAMP Aerial Surveys that Overlapped with MESA Aerial Surveys
Note. Reprinted from “Report of Marine Bird and Marine Mammal Component, Puget Sound Ambient
Monitoring Program,” by D. Nysewander, B. Murphie, J. Evenson, and C. Thomas, 2005, report prepared
for WDFW, 106. Copyright (2005) by Washington State Department of Fish and Wildlife.
In the early 2000s, Bower (2009) instructed Western Washington University (WWU)
students to conduct point census and ferry-based seabird surveys within the southern Strait of
Georgia and adjoining inland waters (Figure 6). The WWU student surveys focused on nonbreeding seabird abundance and were conducted September through May in 2003, 2004, and
2005. Bower then compared WWU data to the MESA baseline and found a significant decrease
of 28.9% in overall seabird abundance. There was no pattern between feeding guilds and no
single reason these species are in decline. Bower found that surf scoters declined by 60%.
Although this was not a statistically significant finding, it is still notable because the historical
23
gathering at Cherry Point in response to herring spawn was so large, that when the area was
removed from the analyses, the magnitude of the decline was halved.
Figure 6
WWU Shoreline Census and Ferry Survey Locations
Note. Reprinted from “Changes in Marine Bird Abundance in the Slash Sea: 1957 to 2007,” J.L. Bower,
2009, Marine Ornithology, 37(1), 11. Copyright (2009) by Marine Ornithology.
Anderson et al. (2009) also used the WWU data but focused on Padilla Bay. Like Cherry
Point, Wahl (1981) identified Padilla Bay as an area of importance and vulnerability for marine
birds. At Padilla Bay, this recognition is due to extensive eelgrass beds. The authors found a
significant decrease in overall seabird density in Padilla Bay of 17% between the 1970s and early
2000s. Padilla Bay does not overlap with CPAR but is an example of how seabird density
analyses can be applied to a small management area. The authors documented a significant
24
decline in 13 taxa many of which were species that had formerly been the most abundant such as
Brant and Western Grebe. Declines were most prevalent during winter and spring migrationstimes when Wahl et al. (1981) had recorded the highest seabird densities across Puget Sound.
Many of the Padilla Bay declines were documented across the Salish Sea at large, leading the
authors to believe that the cause for species decline may be widespread. They were unable to
connect the declines with any habitat change in Padilla Bay; however, further declines or drastic
seabird changes could indicate habitat degradation.
Figure 7
Padilla Bay Shoreline Survey Points and Vessel-based Survey Routes
Notes. Reprinted from “Changes in avifaunal abundance in a heavily used wintering and migration site in
Puget Sound, Washington, during 1966-2007,” by E. Anderson, J. Bower, D. Nysewander, J. Evenson
and J. Lovvorn, 2009, Marine Ornithology, 37(1), 20. Copyright (2009) by Marine Ornithology.
25
Community science seabird monitoring efforts
Community science, also referred to as citizen science, refers to natural science data
collected by members of the public. Some of the longest running community science programs in
the United States are bird programs. The Christmas Bird Count has over 50 years of data
spanning terrestrial and marine ecosystems. Community science efforts are often developed to
establish baseline monitoring and address conservation questions, but there are limitations
inherent to community science data (Ward et al., 2015). Surveys usually have consistent
methodology and some form of quality control, but it is challenging to quantify the monitoring
effort. Additionally, there tends to be little auxiliary data (i.e., visibility, condition, behavior) and
there are often biases based on the location of the survey. Community science efforts are often
limited to non-random locations based on accessibility and volunteer availability. As with the
‘coarse’ approach taken by Pearson and Hamel (2013) these limitations mean that the data is
suitable for establishing trends, but perhaps not for analysis of specific management activities.
Nonetheless, community science is a growing field that supplies a bounty of data that is
increasingly incorporated into scientific models. Sipe (2019) incorporated Ebird data with
WDFW survey data to create occupancy models for common loons in Washington State. The
models used by Harvey et al. (2012), Pearson and Hamel (2013) and Kershner et al. (2011) for
Puget Sound Vital Signs all incorporate community science databases. In addition to comparing
WWU to MESA surveys, Bower (2009) compared Christmas Bird Counts from the Strait of
Georgia between 1975 and 1984, and 1998 and 2007. The results were mostly consistent with
regional changes identified within other literature for these species.
Puget Sound Seabird Surveys are run by the Seattle Audubon Society to monitor winter
seabird populations in Puget Sound, Strait of Juan de Fuca and San Juan Islands. Ward et al.
26
(2015) used Puget Sound Seabird Survey data to map seabird occupancy over space and time to
identify seabird hotspots for potential monitoring and conservation efforts. As indicators, shifts
in seabird habitat usage can reflect changes in the habitat itself. If areas that were historically
seabird hotspots no longer attract those populations, or do so during a different season, that area
can be identified for more targeted studies or more intensive monitoring to understand why that
shift has occurred. An example of a cooled hotspot is Cherry Point. Community science data is a
cost-effective way to identify areas of concern for seabird species on a scale beyond the scope of
most single programs.
The Canadian corollary to Puget Sound Seabird Surveys is the British Columbia Coastal
Waterbird Surveys (BCCWS). BCCWS is a community scientist effort that is providing seabird
information that otherwise would be unattainable to scientists due to cost and effort limitations.
This is the only survey focusing on winter non-breeding seabird populations in the Canadian
portion of the Salish Sea. Crewe et al. (2012) ran analyses across 12-years of BCCWS data
spanning 1999 to 2011. Ethier et al. (2020) continued these analyses using data from 1999 to
2019. Crewe et al. (2012) conducted a power analysis to evaluate data quality and found that the
survey is a credible data source capable of detecting annual changes of 3% or less. Both papers
saw declines in seabird abundance, especially forage fish dependent species. Many of these
species were long-distance migrants and feed at high trophic levels. Additionally, the populations
of these species on the outer coast have remained constant (Ethier et al., 2020). Brant, for which
Anderson et al. (2009) found a significant decline in Padilla Bay, had been increasing in the
Frasier Delta suggesting a territory shift (Crewe et al., 2012).
27
Interpreting Salish Sea seabird trends
Overall, seabird abundance is on the decline throughout the Salish Sea. Table 1 compiles
the results from Nysewander et al. (2005), Bower (2009), Anderson et al. (2009), Vilchis et al.
(2015), and Ethier et al. (2020). Temporal, spatial, and methodological variations in seabird
survey efforts leads to variability in the results and challenges when comparing data sets. Among
the five publications highlighted above, grebe species have consistently declined. However,
grebes nest on freshwater and their declines are likely due to degradation of freshwater
environments (Bower, 2009; Pearson & Hamel, 2013). This is where specific indicator species
suggest more meaningful results. I suggest the use of scoters, Pigeon Guillemot and Pelagic
Cormorant as seabird indicators for CPAR.
Scoters have likely been on a regional decline although there may be local redistribution
within the Salish Sea (Crewe et al., 2012). Pigeon Guillemot have increased according to
shoreline surveys (Bower, 2009; Crewe et al., 2012) although aerial surveys indicated a decline
(Nysewander et al., 2005). Shoreline surveys better represent Pigeon Guillemot since they nest in
shore-side bluffs and forage in near-shore habitats. The Salish Sea Guillemot Network is a
community science program monitoring Pigeon Guillemot breeding colonies throughout Puget
Sound with the intent to better understand their population dynamics and their role within
nearshore environments. Pelagic Cormorant trends are contradictory. According to Crewe et al.
(2012) their breeding population in the Strait of Georgia decreased by 50%, but their
nonbreeding presence in the winters appears to be increasing. This suggests that the Strait of
Georgia may be an increasingly important winter migratory stop for this species.
While no author can point to specific reasons for the observed population trends, the
trends of individual species can suggest areas or resources that may warrant further study.
28
Vilchis et al. (2015) attempted to characterize seabird trends across the Salish Sea by
incorporating WDFW aerial surveys and shoreline BCCWS and Christmas Bird Count (CBC)
data (note that the last two are both community science networks). In addition to the results
compiled in Table 1, they also looked for life history factors that may explain abundance trends.
These life history categories included feeding strategy, main prey source and breeding location
(resident vs. migratory). They found that diving species accounted for over 90% of the declines
and that diving birds that winter in the Salish Sea like alcids, grebes and loons were 11% more
likely to have declined than surface feeders like geese and dabbling ducks. Bird species that feed
on forage fish were 8% more likely to decline. On a positive note, species that breed within the
Salish Sea were less likely to have declined than non-local breeding species.
Due to the risk factors associated with life history, Vilchis et al. (2015) found that seabird
community structure in the Salish Sea shifted from 1990 to 2010. Where previously alcids and
sea ducks were common, 2010 saw more non-diving bird species and piscivores with diverse
diets (not specializing on forage fish). Overall, pursuit divers that specialize on forage fish and
do not breed locally were less likely to overwinter in the Salish Sea in 2010 compared to 1990.
Ethier et al. (2020) also found that migratory, forage-fish-specialized, diving birds were the most
likely to decline between 1999 and 2019. This is likely due to a shift in prey availability as
urbanization has decreased forage fish spawning habitats. Additionally, Ethier et al. (2020) found
that benthivores have declined in the Salish Sea while both piscivore and benthivore abundance
on the coast have remained stable. These Salish Sea specific decreasing trends reflect how
seabirds are choosing to winter elsewhere and indicate specific habitat concerns for nearshore
and benthic habitats within the Salish Sea.
29
Table 1
Percent Change to Seabird Abundance According to Five Relevant Studies for all Seabird Species
Currently Surveyed at CPAR
Species
Nysewander
et al. 2005
PSAMP/MES
A comparison
across the
southern
Salish Sea
1978-1990s
Bower 2009
WWU/MESA
comparison
across the
southern Strait
of Georgia
1978 to 2000s
Anderson et al.
2009
WWU/MESA
comparison in
Padilla Bay
1978 to 2000s
Vilchis et al.
2015
WDFW,
BCCWS and
CBC
comparison
across the
Salish Sea
1990-2010
-7.5 (-10.4 for
all cormorant)
Ethier et al.
2020
BCCWS trend
analysis in the
northern Strait
of Georgia
1999-2019
Double-crested
Cormorant
Phalacrocorax
auritus
-61.7
+97.7
Not significant
Pelagic
Cormorant
Urile pelagicus
-53.0 for all
cormorant
+87.7
Not significant
Not significant
(-10.4 for all
cormorant)
Not significant
Red-throated
Loon
Gavia stellata
-79.1 for all
Gavia spp.
-79.9
-11.7
-3 (-20.9 for all
Gavia spp.)
Not Significant
Pacific Loon
Gavia pacifica
-79.1 for all
Gavia spp.
Not significant
+20.9
-1.5 (-20.9 for
all Gavia spp.)
-6.01
Common Loon
Gavia immer
-64.3
+48.8
+9.0
-1.5 (-20.9 for
all Gavia spp.)
-2.96
Red-necked
Grebe
Podiceps
grisegena
Horned Grebe
Podiceps
auritus
-88.8
-45.9
-33.4
-3 and +6 (-23.9
and +6 for all
grebe)
Not significant
-82.4
-71.6
-59.1
-1.5 (-23.9 and
+6 for all grebe)
Not significant
Western Grebe
Aechmophorus
occidentalis
-95.2
-81.3
-82.6
-19.4 (-23.9 and
+6 for all grebe)
-12.72
Red-breasted
Merganser
Mergus
serrator
Common Murre
Uria aalge
Not significant
Not significant
Not significant
-1.5 and +3 for
all Mergus spp.
Not significant
Not reported
-92.4
Not reported
-22.4
Not significant
Pigeon
Guillemot
Cepphus
columba
-55.2
+108.9
Not reported
+4.5
Not significant
30
Not significant
Marbled
Murrelet
Brachyramphus
marmoratus
Rhinoceros
Auklet
Cerorhinca
monocerata
-96.3
-71.0
Not reported
-9 and +3
Not significant
Not reported
Not reported
Not reported
-9
Not reported
Caspian Tern
Hydroprogne
caspia
Not reported
Not reported
Not reported
Not reported
Not reported
Canada Goose
Branta
canadensis
Brant
Branta bernicla
Not reported
+10801.9
+9.9
+3
+4.92
-66.3
Not significant
-44.8
+6
Not significant
Mallard
Anas
platyrhynchos
Scaup
Aythya spp.
Harlequin Duck
Histrionicus
histrionicus
Not reported
Not significant
+18.6
+3
Not significant
-72.3
-64.8
-93.3
-3
-10.68
+188.6
Not significant
Not reported
Not significant
Not significant
Long-tailed
Duck
Clangula
hyemalis
Bufflehead
Bucephala
albeola
Common
Goldeneye
Bucephala
clangula
Barrow’s
Goldeneye
Bucephala
islandica
Ruddy Duck
Oxyura
jamaicensis
Surf Scoter
Melanitta
perspicillata
Not reported
Not significant
Not significant
Not significant
-5.07
Not significant
Not significant
-10.9
-1.5
Not significant
Not significant
-47.8
-10.7 for all
goldeneye
Not significant
for all goldeneye
Not significant
Not significant
Not significant
-10.7 for all
goldeneye
Not significant
for all goldeneye
Not significant
Not reported
-59.7
-47.6
-6
Not reported
-57.0 for all
Melanitta spp.
Not significant
+14.0 for all
Melanitta spp.
-9 for all
Melanitta spp.
-2.27
Black Scoter
Melanitta nigra
-57.0 for all
Melanitta spp.
-65.7
+14.0 for all
Melanitta spp.
-9 for all
Melanitta spp.
-14.96
White-winged
Scoter
Melanitta fusca
-57.0 for all
Melanitta spp.
Not significant
+14.0 for all
Melanitta spp.
-9 for all
Melanitta spp.
-4.3
31
Bald Eagle
Haliaeetus
leucocephalus
Not significant
+187.0
Not reported
+1.5
Not significant
Great Blue
Heron
Ardea herodias
Not significant
Not significant
Not reported
Not significant
Not significant
Note. Statistical significance was evaluated with an α < 0.05 for all except Vilchis et al. (2015) who used
α < 0.10. Vilchis et al. (2015) used a depth-based analysis that sometimes resulted in both positive and
negative trends depending on the species-depth combination.
CPAR marine bird survey effort
As previously mentioned, it can take years of data for seabird trends to become apparent.
The resources required to conduct these long-term monitoring efforts are substantial. The
Aquatic Reserves Program is small, and the efforts of community scientists are invaluable. The
Cherry Point Aquatic Reserve Citizen Stewardship Committee (CSC) is a group of volunteers
who meet monthly to help promote the protection and monitoring of CPAR. RE Sources for
Sustainable Communities, a non-profit in Bellingham, WA, coordinates and sponsors the Cherry
Point and Fidalgo Bay CSCs. These volunteer teams have several research projects they
developed and undertake within the Aquatic Reserves to benefit Aquatic Reserve management.
The CSC seabird surveys are one of these efforts. Surveys began at CPAR in spring 2013 and
were modeled after the WWU seabird surveys and therefore the MESA point census surveys.
Both the WWU and current CPAR surveys are winter surveys designed to capture the influx of
migratory species that arrive in the Salish Sea in late fall and depart in the spring. Winter surveys
capture non-breeding species and activities. Birds that are not tied to a breeding location are
more able to move and follow prey resources which means their presence and abundance may
better reflect resource availability (Vilchis et al., 2015).
The CPAR surveys deviate from MESA protocols in two ways: teams of trained
observers conduct the surveys instead of a single scientist and CSC members only count
32
individuals of certain species. CPAR surveys started with seven species in 2013 and have
increased to 29 species and species groups as volunteers have become more comfortable with the
methods and capable of executing them.
With any point census survey, the numbers recorded likely underestimate the true number
of seabirds present, but by having unlimited time, the errors are minimized. These surveys are a
snapshot of all birds within surveyed species groups at that area at that time. Based off MESA
methods, observers record all seabirds on the water, shoreline, or in flight. In a spatially
constrained area such as a reserve, birds flying by may not be using the area in question. In fact,
as prey options deplete, seabirds may have to travel farther to forage. An area with poorer habitat
will see more birds flying by and fewer stopping to use the habitat (Piatt, Harding, et al., 2007).
By recording flying birds, this may be artificially increasing the number of seabirds recorded as
present within the management area.
In addition to species counts, observers record condition information at each survey site
including glare, Beaufort Sea State Code, human disturbance, and visibility. The visibility metric
is a value judgment made by volunteers and recorded as poor, fair, good, or excellent. In optimal
conditions, seabirds up to about 2000 m are visible. There is currently no way to relate the
visibility estimate to a maximum observation distance. BCCWS use a similar methodology with
their volunteers but have them record whether the observation as made within 500 m or beyond
500 m. Adding a distance estimate could be an addition to increase the quality of the CSC
surveys.
Purpose and value of this thesis
Birds can indicate environmental health, biodiversity, condition of habitats and climate
change (Pearson & Hamel, 2013). For the Aquatic Reserves Program, the goal is to promote
33
diverse and resilient ecosystems; seabirds can indicate progress towards this goal. Trends in
seabird abundance and distribution can reflect desirability of habitat, prey availability and
general ecosystem health. CPAR is a small portion of the larger Salish Sea. Depending on the
species, seabirds may not spend much time in the Aquatic Reserve itself and are therefore
impacted by things that happen elsewhere in their life history. Therefore, Pigeon Guillemot, Surf
Scoter, and Pelagic Cormorant are the three indicator species specifically suggested for CPAR.
All three species are documented to spend time on the reserves and are closely tied to a localized
marine resource.
Regional seabird abundance studies have varied results depending on methodology and
location. The data used in this thesis was collected by community scientists using a methodology
adapted from historical MESA surveys and more recent WWU efforts. At the time the CPAR
surveys were developed, the intent was to compare to these two baselines. However, these
historical datasets are not open-access, and it was meaningful from a management perspective to
focus on the status and trends of seabird indicators. I took a similar approach to BCCWS
analyses and focused on this dataset alone to identify shifts that occurred within the last decade.
Changes to seabird abundance within CPAR may highlight habitats that warrant further
monitoring or reflect changes to resource availability like herring spawn.
DNR’s Aquatic Reserves Program manages seven marine Aquatic Reserves. Finding a
way to track and assess effectiveness of management decisions is crucial for the program to
improve management strategies. Ecosystem indicators, like seabirds, are used in the Salish Sea to
track ecosystem-based management efforts and this thesis applies them to a single reserve at
Cherry Point. This review and subsequent analyses will evaluate the current methods, create a
status update on CPAR seabird density, and provide an example to DNR and other agencies on
34
how to support and incorporate this kind of community science effort. Through this thesis,
CPAR community scientists will help DNR better manage and protect our Aquatic Reserve
ecosystems.
35
METHODS
Site description and survey methods
Cherry Point Aquatic Reserve covers 3,050 acres of aquatic habitat in the eastern Strait of
Georgia. It is bordered to the north by Birch Bay State Park and to the south by Lummi
Reservation. Point census seabird surveys are conducted at three shore-based locations along the
Cherry Point shoreline. Figure 8 shows the reserve boundary, survey locations and survey areas.
Figure 8
Map of Current CPAR Seabird Survey Locations and Survey Areas
36
Six or more volunteers conduct each survey forming two teams of three or more. Each
team has a spotter who uses a spotting scope to identify distant birds, a counter who used
binoculars and identifies nearer birds and a recorder who reports the data on the data sheet. The
two teams stand on the same shoreline location, identify a reference point in the middle of the
survey area and begin their surveys moving out from that center point with each team covering
half of the total area. Figure 9 has an example diagram of the survey design. Binoculars and
spotting scopes were approximately equal in quality to Eagle Optics Ranger 10×40 binoculars
and 20-40× scopes (Eagle possible Optics, Middleton, WV, USA) which were the equipment
strengths used by historical MESA surveys (Bower, 2009; Hines & Jaeren, 2018; Wahl et al.,
1981). Surveys continue without time or distance constraint (beside those imposed by the
equipment) until all the birds within the targeted species categories have been recorded. Survey
times ranged from four to 65 minutes and lasted an average of 19.2 ± 8.3 minutes. All spotters
and counters were trained on seabird identification.
37
Figure 9
Diagram of CPAR Seabird Field Survey Operating Procedure
Point Census Surveys were conducted once a month from September to May starting in
April 2013. Rather than identify and count all species present, the surveys target only certain
species that are known to be the most numerous and frequently seen in the region. Initially, the
targeted species list included seven species. It was expanded to 15 species in March 2015 and
type groups were added in September of 2016. Type groups are genus or family level groups
meant to capture individuals that could not be identified to species. This means that a mystery
loon would be recorded under a loon category, or an unidentifiable duck is recorded as duck
rather than being omitted from the data collection as was the previous practice. The final species
additions were made in December 2016. Since December 2016, there are 29 species and nine
type groups included in the CPAR surveys.
38
Table 2
All Seabird Species Targeted by the CPAR Survey Effort Including Common Name, Scientific Name,
Four-letter Species Code, Feeding Guild, and the Month/Year of Addition to the Targeted Species List.
Type
Common name
Scientific name
Species
Code
DCCO
Guild
cormorant
cormorant
Double-crested
Cormorant
Pelagic Cormorant
Phalacrocorax
auritus
Urile pelagicus
PECO
Piscivore
cormorant
cormorant species
Phalacrocoracidae
CORM
Piscivore
dabbling duck
Mallard
Anus
platyrhynchos
Aythya marila
MALL
Herbivore
diving duck
Greater Scaup
GRSC
Omnivore
diving duck
Harlequin Duck
HARD
Benthivore
Long-tailed Duck
Histrionicus
histrionicus
Clangula hyemalis
diving duck
LTDU
Benthivore
diving duck
Bufflehead
Bucephala albeola
BUFF
Benthivore
diving duck
Ruddy duck
RUDU
Benthivore
diving duck
duck species
Oxyura
jamaicensis
NA
DUCK
Other/all
goose
Canada Goose
Branta canadensis
CAGO
Herbivore
goose
Brant
Branta bernicla
BRAN
Herbivore
goose
Black goose species
Branta spp.
GOOS
Herbivore
goldeneye
Common Goldeneye
COGO
Benthivore
goldeneye
Barrow's Goldeneye
BAGO
Benthivore
goldeneye
goldeneye species
Bucephala
clangula
Bucephala
isandica
Bucephala spp.
GOLD
Benthivore
grebe
Red-necked Grebe
RNGR
Piscivore
grebe
Horned Grebe
Podiceps
grisegena
Podiceps auritus
HOGR
Piscivore
grebe
Western Grebe
Aechmophorus
occidentalis
WEGR
Piscivore
39
Piscivore
Date
added
Mar
2015
Mar
2015
Sept
2016
Dec
2016
Dec
2016
Apr
2013
Dec
2016
Dec
2016
Dec
2016
Dec
2016
Dec
2016
Apr
2013
Dec
2016
Apr
2013
Mar
2015
Sep
2016
Mar
2015
Mar
2015
Apr
2013
grebe
grebe species
Podicipeformes
GREB
Piscivore
loon
Red-throated Loon
Gavia stellata
RTLO
Piscivore
loon
Pacific Loon
Gavia pacifica
PALO
Piscivore
loon
Common Loon
Gavia immer
COLO
Piscivore
loon
loon species
Gavia spp.
LOON
Piscivore
merganser
Mergus serrator
RBME
Piscivore
merganser
Red-breasted
Merganser
merganser species
MERG
Piscivore
scoter
Surf Scoter
SUSC
Benthivore
scoter
Black Scoter
BLSC
Benthivore
scoter
White-winged Scoter
Mergus spp. and
Lophodytes spp.
Melanitta
perspicillata
Melanitta
americana
Melanitta deglandi
WWSC
Benthivore
scoter
scoter species
Melanitta spp.
SCOT
Benthivore
alcid
Common Murre
Uria aalge
COMU
Piscivore
alcid
Pigeon Guillemot
Cepphus columba
PIGU
Piscivore
alcid
Marbled Murrelet
MAMU Piscivore
alcid
Rhinoceros Auklet
alcid
alcid sp.
Brachyramphus
marmoratus
Cerorhinca
monocerata
Alcidae
heron
Great Blue Heron
eagle
Bald Eagle
tern
Caspian Tern
RHAU
Piscivore
ALCI
Piscivore
Ardea Herodias
GBHE
Other
Haliaeetus
leucocephalus
Hydroprogne
caspia
BAEA
Other
CATE
Piscivore
Sep
2016
Mar
2015
Mar
2015
Apr
2013
Sep
2016
Dec
2016
Dec
2016
Apr
2013
Dec
2016
Mar
2015
Sep
2016
Dec
2016
Dec
2016
Dec
2016
Dec
2016
Dec
2016
Dec
2016
Apr
2013
Dec
2016
Data management
Since the surveys began in 2013 there has been no comprehensive database review.
Previously, both team’s data sheets were combined before being entered into a spreadsheet. I
went through scans of the data sheets filled out by CPAR community scientist during their
40
surveys. I reworked the database so that raw data is being entered, not compiled data. I then
created a pass/fail test to remove data that was low quality or missing values.
Table 3
Criteria for the Pass/Fail Test Used to Assess Data Quality Before Analysis
Pass: Criteria for data to be included in analyses
-
Few or no missing values. Clear scan and no quality concerns.
-
90% of the data passed and was included in the analyses
Fail: Criteria used to remove low quality data
-
Major errors. The scans where the species counts were compromised, and the original
data sheet could not be found.
-
Poor visibility. Observers stated that all surveys were only conducted when the full survey
area was visible, and the poor rating was due to glare or high Beaufort rather than
decreased range of vision (personal communication, March 12, 2022). Even though they
felt confident that they were able to document all birds in the survey area during poor
visibility, the conditions likely led to a greater than normal underestimation of seabirds
present, and I opted to exclude this data from analyses.
-
Surveys that started after 3 pm. Seabirds behavior changes based on time of day. For
consistency, surveys conducted in the late afternoon/early evening were excluded.
-
10% of the data failed the quality test
I also flagged surveys for additional quality assurance/quality control (QA/QC) due to
minor errors and missing values that could be extrapolated from the other team. The errors did
41
not impact the seabird count data, and these surveys were included in the analyses, but further
review will improve the database.
Some surveys like those conducted in April and May 2020 had only one team of data
collectors due to the Covid 19 pandemic. Surveys conducted with reduced numbers were
accepted for analysis and their density calculations were altered to account for a single team
covering the entire survey area.
Calculating seabird density
To control for the omitted data, I looked at density instead of abundance. Since I am
interested in trends across the entire reserve, I did not compare between sites, but combined them
to look at trends over the total area. I estimated maximum survey distance to be around 2 km
based on personal observation and conversations with the data collectors. The Gulf Road and
Neptune Beach locations are on long stretches of beach partially bounded by industrial piers. The
survey location at Sandy Point is on a curved promontory and includes a human-made inlet next
to the survey location. Views to the south at Sandy Point are bounded by a house and to the north
by an additional point of land. To estimate survey area, I mapped the survey locations in ArcGIS
Pro and created 2 km buffers. I edited the resulting polygons to account for the visual barriers
described above. Estimated survey areas rounded to the nearest tenth of a kilometer are as
follows: Gulf Road: 5.5 km2, Neptune Beach: 5.2 km2, and Sandy Point: 6.9 km2. If a single
team surveyed, their totals were calculated across the entire survey area.
Data exploration and analyses
To identify species for further analysis, I looked at encounter rate and average density for
each of the 29 targeted species. I focused further analyses on the seven species with the highest
probability of encounter and average density (Surf Scoter, Common Loon, Horned Grebe,
42
Pelagic Cormorant, Bufflehead, Western Grebe, and Brant). In addition to these seven species, I
also looked at total seabird density and feeding guild density. Due to the non-normal
distributions of the density data, all analyses that compared between survey seasons and months
were conducted using a Kruskal-Wallis rank sums test using the R package dyplyr (R Core
Team, 2019) and post hoc Dunn’s test using the R package CRAN (Dinno, 2017) with a
Bonferroni p-value correction. The independent variable was either season or month and the
dependent was density.
In addition to being non-normal, the data was heavily zero skewed. A negative binomial
distribution is best suited to data with many zero-observations (Crewe et al., 2012; O’Hara &
Kotze, 2010). I utilized a negative binomial regression model using R packages foreign (v0.0-71,
R Core Team, 2018) and MASS (Venables & Ripley, 2002) to look for change in density as a
function of season and month [glm.nb(Density ~ season_num+month)]. In this way I could
assess whether density has been increasing or decreasing since 2013 while accounting for the
effect of month. Further analyses should focus on model selection within negative binomial
regression or Poisson analyses. A pseudo-R squared calculation or log likelihood significance
test could be done to assess the strength of the model. Other factors like visibility, tide, time, and
human interaction could be incorporated into this model in the future. I also recommend a follow
up pair-wise test using the emmeans R package (v1.7.3, Lenth, 2022) to identify which seasons
and months had different densities and compare these results to the Kruskal-Wallis output. All
the data wrangling, visualization and analyses were done in the statistical program R 3.6.0 (R
Core Team, 2021). R scripts are retained by the author and DNR Aquatic Reserves Program.
43
RESULTS AND DISCUSSION
Originally, the Cherry Point seabird data was developed primarily for comparison to an
historical dataset. Using the current data as a stand-alone data source required extensive data
exploration to identify potential patterns and directions for analysis. I explored both seabird
encounter rate and density to identify seven species that have been recorded most often and/or in
the largest average densities. I then compared the individual density for each of the seven species
between and across seasons and months. Each species has a subsection below with the plots,
results of the analyses, and discussion with potential reasons behind the trends. Additionally, I
looked at total bird density and density of targeted species grouped by feeding guilds.
This data was collected by community scientists at CPAR. In total, 498 surveys had been
conducted constituting over 150 hours of survey time and counting over 37,000 birds. These
surveys start April 2013 and are ongoing, although the last data I included was from February
2022. Of the total 498 surveys conducted, 426 passed the QA/QC process outlined in the
methods and were included in the analyses and visualizations below. Figure 10 below illustrates
the hours of survey time, and the total number of birds counted each season. Each column in the
figure corresponds to a single season and each color band within that column represents a month.
The taller the color band, the more time was spent, or the more birds were seen depending on the
plot. Typically, longer surveys are expected when the number of birds is greater, but the survey
length was also influenced by environmental factors like Beaufort Sea State, glare, and
precipitation that decrease visibility and make it more challenging to census all targeted bird
species. As more species were added to the target list, survey time is expected to increase,
however this coincided with the birders becoming more practiced.
44
Figure 10
Total CPAR Survey Time and Total Seabird Counts Per Season
Note. On the top is total survey time per season with the total time per month as fill. On the bottom is the
total number of birds recorded per season with the birds per month as fill. Species were added to the
target list in March 2015, September 2016, and December 2016. The 2012-2013 season consisted of a
single survey and the 2021-2022 season did not include March, April, or May data at the time of analysis.
The number of species targeted by the CPAR surveys has expanded over the total survey
effort. During each monthly survey event, all individuals within the targeted species list were
counted by trained observers. This was either 73, 61 or 46 months of data depending on when the
species was added to the targeted list. Each month, two teams visited three survey sites. Figure
11 illustrates the presence and absence of each targeted species since the surveys first began in
May 2013. If the species was marked present it means that it was recorded by at least one survey
team at one or more of the three sites on that survey day.
45
Figure 11
Presence and Absence of All Species Targeted by the CPAR Seabird Surveys
Note. Species encountered most often are at the top and species encountered least often are on the bottom.
The figure also illustrates when the survey effort was expanded from 7 to 15 then to 29 species. Type
groups were excluded from this plot.
Encounter rate was used to compare how often targeted seabird species were seen at
CPAR. To standardize the encounter opportunity across all species, I used only surveys
conducted December 2016 or to February 2022 since those were all the surveys in the dataset
that were conducted with the most expanded target species list. Encounter rate was calculated by:
sum (# of surveys with that species)
total # of surveys
The average probability of encounter across all species was 23 ± 26 percent. Figure 12 plots each
species in order of overall encounter probability. I selected the five species with a 50% or higher
probability of encounter for further analysis: Surf Scoter, Common Loon, Horned Grebe, Pelagic
Cormorant and Bufflehead.
46
Figure 12
Encounter Rate for All Species Targeted by the CPAR Seabird Surveys
I compared mean density to identify which species were often recorded in groups or
flocks (>5 individuals or 1.5 birds/km2). Unless surveys were excluded in the QA/QC process,
each survey event consisted of six density calculations for the two teams at three different sites. I
compared density across the four seasons from 2017/2018 to 2020/2021 since these were the
most complete seasons with only one survey missed in October 2018. Across all species, mean
density was 1bird per km2 with a standard deviation of 8 (rounded to the nearest bird). Figure 13
shows which species were seen in above average densities. I selected the four species with
average densities above 1.5 birds per square kilometer for further analysis: Brant, Surf Scoter,
Western Grebe, and Horned Grebe.
47
Figure 13
Mean Density for All Species Targeted by the CPAR Seabird Surveys
Note. The dashed vertical line represents average density across all surveys (1.02 birds/ square km). Each
point is mean density for that species and the error bars represent standard deviation truncated at zero.
Between encounter rate and average density, I identified seven species that were often in
the CPAR area (probability of encounter > 50%) and/or appear in larger numbers when they
were present (mean density > 1.5 birds/km2): Surf Scoter, Common Loon, Horned Grebe,
Pelagic Cormorant, Bufflehead, Brant, and Western Grebe. For each of these seven species I
looked at change between seasons and survey months. Additionally, I compared total bird
density and the density of birds grouped by feeding guild across months and seasons. Through
these analyses, I identified times when these seabirds are most present at CPAR. Knowing when
seabirds are present is important for management decisions. It can identify times when habitat
resources may be more available at CPAR or when certain species may be more susceptible to
disturbance. Feeding guilds may indicate the presence of certain prey types (i.e., higher piscivore
density may mean there are more fish prey available).
48
The species-level visualizations are violin plots which reflect the data distribution as well
as summary statistics. Each shape represents the density distribution for that month or season.
Wider sections of the shape correspond to bird density values that were recorded more often –
the wider the shape, the higher the probability that birds will be seen in that density that
season/month. Short, wide shapes mean that all the data points were similar in value i.e., density
was consistent during that season/month with little variation. Tall, skinny shapes represent more
data variation i.e., density was more variable during that time period. Within each violin shape is
a box with the mean (dark horizontal line) between the upper and lower quartiles. Multi-species
groups had greater levels of variation and, subsequently, are represented with a different plot
style. Feeding guild, overall density and type group plots include mean density as a point with
standard deviation as error bars truncated at zero.
49
Surf Scoter
Figure 14
Surf Scoter Density by Season and Month
Note. Surf Scoter density shown by season on top plot and by month in the bottom plot. This plot
visualizes data distribution including aggregations, or peaks, in certain seasons or months
50
Figure 15
Surf Scoter Density by Season and Month Focusing on Summary Statistics
Note. Surf Scoter density by season on the top and month on the bottom with the y-axis bounded at 20
birds/km2.
There was no significant difference in Surf Scoter density between seasons of observation
(Kruskal-Wallis, H9=8.98, p = 0.44). There were some notable aggregations in the 2014-2015
season that can be seen in Figure 14. While not statistically significant, these instances of high
density represented large flocks or aggregations of Surf Scoter and points towards an area of
future inquiry. Historically, Surf Scoter formed massive aggregations in response to Cherry Point
herring spawn. The 2014-2015 seabird season overlapped with a 2015 peak in Semiahmoo
herring spawn deposition (Sandell et al., 2019) which is the herring stock just to the north of the
Cherry Point area.
51
Figure 15 magnifies the summary statistics (means and quartiles) within the violin plots.
According to the negative binomial regression model, there has been a significant (p < 0.01)
decrease in surf scoter density since 2013. The decrease is still significant (p < 0.01) even if only
data after the peak 2014-2015 season is included in the model.
Across survey months, September had significantly lower density than any other month
(Kruskal-Wallis, H8 = 60.27, p < 0.01, Dunn’s post hoc, p < 0 .01 for all pairs except May-Sep
p=.01). This is likely due to migration timing as Surf Scoter enter the area. Additionally, there is
a significant difference between April and May (Dunn’s post hoc p = 0.02). Analysis shown in
Figure 14 shows that there were non-significant density peaks in April and May. This timing is
notable because this was when the Cherry Point herring were spawning. Other north Puget
Sound herring stocks spawn in February and March (Sandell et al., 2016).
Future applications of this data could include exploring this relationship between Surf
Scoter density and local herring spawn deposition. Figure 16 visualizes Surf Scoter density
trends across each season and that spring density peak has not been seen since 2017. Whatever
resource, whether it be herring spawn or something else, that was keeping Surf Scoter at Cherry
Point into the Spring, may no longer be present. Cherry Point herring have been in decline since
the 1970s so it will take additional monitoring and research to understand the mechanisms
behind this shift in Surf Scoter migration timing.
52
Figure 16
Surf Scoter Density Across Each Season
Note. Mean Surf Scoter density each month is represented by a point. The points are connected by best fit
line using a loess formula. Each line estimates the density trend for that season.
53
Common Loon
Figure 17
Common Loon Density by Season and Month
Note. Common Loon density shown by season on top plot and by month in the bottom plot.
There was a significant difference in Common Loon density between the 2018-2019 and
2021-2022 seasons (Kruskal-Wallis, H9 = 22.89, p = 0.01, Dunn’s post hoc, p = 0.02). However,
the 2021-2022 season was incomplete at the time of analysis so this test should be recreated
when the full season of data becomes available. Additionally, there was a significant difference
in Common Loon density between months (Kruskal-Wallis, H8=38.01, p < 0.01). Common Loon
density was significantly higher in October than September (Dunn’s post hoc p < 0.01),
November (Dunn’s post hoc p < 0.01), December (Dunn’s post hoc p < 0.01), January (Dunn’s
54
post hoc p = 0.01), March (Dunn’s post hoc p < 0.01) and May (Dunn’s post hoc p < 0.01). May
densities were also significantly lower than April (Dunn’s post hoc p = 0.04). Common Loon are
winter migrants that breed on freshwater lakes and ponds. Degradation of their lake breeding
areas has led to a regional decrease and their listing as a sensitive species in Washington State
(Richards et al., 2000). They were seen often and were relatively numerous in the CPAR surveys
and there was no change over time in their density (negative binomial regression, p = 0.72).
Loons form mixed species aggregations, and it can be challenging to distinguish to
species when they are near the edge of the visible survey area. When all loon densities are
grouped and analyzed, overall loon density has significantly decreased since 2013 (negative
binomial regression, p < 0.01). A Kruskal-Wallis rank test (H9=95.33, p <0.01) with Dunn’s test
post hoc found that the 2013-2014 and 2014-2015 seasons had significantly higher loon densities
than the 2017-2018 through 2020-2021 seasons (p < 0.01) and the 2014-2015 season was also
higher than the 2021-2022 season (p = 0.02). Additionally, the 2015-2016 season had higher
loon densities than 2017-2018 (p = 0.02) and 2018-2019 (p < 0.01). However, loon density does
appear to be increasing again as the 2021-2022 season has higher densities than the 2018-2019 (p
< 0.01).
Loons are piscivores and Pacific loons have shown an aggregate response to herring. The
2014-2015 season was the same year as the Semiahmoo herring peak. The genus-level loon
group was not added until December of 2016. This means that all loon individuals were not
recorded prior to that time and there may have been more loons than were recorded in the
surveys. This also means there may have been fewer zeros in the data which would artificially
increase mean density. If only dates from the 2017-2018 season or later are used, then loon
density has a significant increasing trend (negative binomial regression, p < 0.01).
55
Figure 18
Mean Loon Density by Season
Note. The points represent the mean density for all loon species, and the error bars are standard deviation
truncated at zero. The 2017-2018 season is the first complete season where all loon individuals were
recorded even if they could not be identified to species.
56
Horned Grebe
Figure 19
Horned Grebe Density by Season and Month
Note. Horned Grebe density shown by season on the top plot and by month on the bottom plot.
There was no significant difference in Horned Grebe density across (negative binomial
regression, p = 0.11) or between seasons (Kruskal-Wallis, H7 = 7.35, p = 0.39). Horned Grebe
density is significantly (Kruskal-Wallis, H8=122.23, p< 0.01) lower in May (Dunn’s post hoc, p
< 0.01 for all pairwise except September) and September (Dunn’s post hoc, p < 0.01 for all
pairwise except May) and all other months. Horned Grebe are a winter migrant that appear to
have consistent times that they arrive and depart the Cherry Point area.
57
Pelagic Cormorant
Figure 20
Pelagic Cormorant Density by Season and Month
Note. Pelagic Cormorant density shown by season on top plot and by month in the bottom plot.
Pelagic Cormorant density was significantly higher in the 2020-2021 season than 20172018 (Kruskal-Wallis, H7=17.33, p = 0.02, Dunn’s post hoc, p=0.03). Additionally, there were
significant differences between months (Kruskal-Wallis, H8 = 80.56, p < 0.01). Pelagic
Cormorant density was higher in September than the months of December (Dunn’s post hoc, p =
0.01), January (p < 0.01), February (p < 0.01), March (p < 0.01), and April (p = 0.01). October
densities were also significantly higher than February (p< 0.01) and March (, p < 0.01). February
and March both had lower densities than May and November (p < 0.01) and March was also
58
significantly lower than April ((Dunn’s post hoc, p = 0.03) and December (Dunn’s post hoc, p =
0.02).
Pelagic Cormorant were one of the few resident species that were seen often and in large
numbers. They nest on bluffs and anthropogenic structures in the summer which may explain
why they were significantly denser in the fall before their numbers appeared to decrease at
CPAR over the winter months and into the early spring. This is the opposite pattern to many of
the winter migrant species targeted by these surveys. Pelagic Cormorant density did significantly
increase over the seasons (negative binomial regression, p < 0.01) which is consistent with the
findings from Crewe et al. (2012).
59
Bufflehead
Figure 21
Bufflehead Density by Season and Month
Note. Bufflehead density shown by season on top plot and by month in the bottom plot.
There was no significant difference in Bufflehead density across (negative binomial
regression, p = 0.43) or between seasons (Kruskal-Wallis, H5=6.61, p = 0.25). Bufflehead are
another migratory species. September, October, and May were all significantly lower (Dunn’s
post hoc p < .01 for all pairwise) than November, December, January, February, March, and
April. Bufflehead are a winter duck species often seen across Puget Sound and the Salish Sea.
The monthly data illustrates their migration timeline as they enter the area in the late fall and
leave in the spring.
60
Brant
Figure 22
Brant Density by Season and Month
Note. Brant density shown by season on top plot and by month in the bottom plot.
Brant densities have been particularly variable as seen by the narrow plots. There was a
significant difference in Brant density between the 2013-2014 and 2018-2019 seasons (KruskalWallis, H9=23.03, p = 0.01, Dunn’s post hoc p = 0.02); however, there was no change across
seasons (negative binomial regression, p = 0.95). There was a significant difference (KruskalWallis, H8=57.58 p< 0.01) in Brant density between September and the months of December (p
= 0.02), January (p < 0.01), February (p < 0.01), and April (p < 0.01); and October and the
months of January (p< 0.01), February (p = 0.01), and April (p=0.01). January had significantly
higher densities than November (p < 0.01) and May (p < 0.01).
61
Brant form large flocks and the appearance of those flocks likely contributes to the large
amount of variability in this data. Brant are herbivores with a preferred diet of seagrass. Like
other goose species, they travel in flocks. There are seagrass beds to the north of the survey sites
in Birch Bay and to the south near Lummi Bay, but the aggregations were likely due to migratory
flocks resting in the CPAR area rather than foraging within it.
Western Grebe
Figure 23
Western Grebe Density by Season and Month
Note. Western Grebe density shown by season on top plot and by month in the bottom plot. Western
Grebe have a lower encounter rate as reflected by the diminished size of the violin shapes in the plots.
There was no significant difference in Western Grebe density across (negative binomial
regression, p = 0.16) or between seasons (Kruskal-Wallis, H9=6.48, p = 0.69). There was a
62
significant difference (Kruskal-Wallis, H8=61.55, p<0.01) in Western Grebe density between
months. November was significantly higher than all other months (p < .01) except October.
September was significantly lower than October (p = 0.01), November (p <0.01) and April
(p=0.04).
Western Grebe have been declining regionally. Bower (2009) and Crewe et al. (2012)
conducted analyses showing that Western Grebe have declined over 80% since the MESA
baseline and over 16% in the last 12 year. These declines have led to their candidate status in
Washington State and their placement on the red list in British Columbia. Due to these
conservation concerns, continued monitoring of this species is advised. More than other grebe
species, they form large rafts on the water which may account for the records of aggregation at
CPAR.
63
All targeted species
Figure 24
Total Seabird Density by Season and Month
Note. Mean density of all targeted birds shown by season on top plot and by month in the bottom plot.
Density by month uses only data collected after December 2016 when the effort expanded to include all
29 species currently being targeted. The points represent the overall mean density, and the error bars are
standard deviation truncated at zero.
I looked at all the targeted species including the genus or family level groups (type
groups) where the data collectors were unable to identify a bird to species. This analysis included
Bald Eagle and Great Blue Heron which are not seabirds but do interact with the marine
environment. Prior to December 2016, the few species being targeted included were those that
have a higher encounter rate and are seen in higher densities like Surf Scoter, Western Grebe,
and Brant. With fewer zeros and more high-density encounters, this artificially biases the data
towards high mean densities in those early years. I included those seasons in the density by
64
season plot, but excluded all data collected prior to December 2016 for the density by month plot
and analyses. Figure 24 illustrates how density decreases as more species were added in the
2015-2016 and 2016-2017 seasons.
Across all seasons (2012-2013 to 2021-2022) there was a significant decrease in overall
seabird density (negative binomial regression, p < 0.01). However, if I include only seasons
where all 29 species were targeted (2017-2018 to 2021-2022), there was a significant increase in
overall seabird density (negative binomial regression, p < 0.01). Across the last five seasons,
there was a significant difference (Kruskal-Wallis, H4=14.69, p < 0.01) in bird density between
the 2018-2019 season which had lower bird densities than the 2019-2020 (p = 0.03) and 20212022 (p < 0.01) seasons. These differences and trends will likely change as there are more
surveys conducted that target all 29 current species.
There is a significant difference in bird density (Kruskal-Wallis, H8=146.82, p < 0.01)
between the months. September and May both were significantly lower than all other months
(Dunn’s post hoc p< 0.01 for all pair-wise tests). This mirrors migration timelines which makes
sense since over 70% of the targeted species are winter migrants and again include those species
that are encountered most often and recorded in the highest density.
If I look at only the seven species originally targeted by the CPAR surveys (Surf Scoter,
Western Grebe, Bald Eagle, Common Loon, Common Goldeneye, Brant, and Harlequin Duck), I
can compare across all seasons without reservation. Figure 25 illustrates the density of the
original seven species. There was a significant decrease in compiled density for these seven
species (negative binomial regression, p = 0.01). This trend is driven by decreasing Surf Scoter
density. If Surf Scoter are removed from the model, then there was no significant change to
combined density of the other six species (negative binomial regression, p = 0.91). This supports
65
the need for long-term data sets and the importance of continuing this survey effort. Patterns in
seabird density may not be visible without five or more seasons of data.
Figure 25
Mean Density of Original Seven Species by Season
Note. Mean density of the original seven species plotted across the seasons. The points represent the
overall mean density, and the error bars are standard deviation truncated at zero.
Where individual species may be more specific indicators of a prey source or habitat
type, overall seabird density is also an indicator of habitat quality. However, these surveys target
only certain species. These results may have statistical significance, but the practical significance
is limited because they do not capture the entire seabird population. This limitation is also
illustrated by the conflicting negative binomial regression results depending on which species are
included in the analysis. Until the CPAR surveys include all bird species present, this combined
data presents a partial picture, and this data set may be most rigorous when applied to individual
species.
Feeding guilds
Feeding guilds were assigned according to Bower (2009). There were four feeding guilds
represented by the CPAR data, but I focused on only two for further analysis. The omnivore (eats
66
everything) feeding guild was represented by a single species (Greater Scaup) and the herbivore
(eats vegetation and marine algae) guild was composed of four species groups (Mallard Duck
and Brant, Canada, and other geese) which, except for Brant, were rarely recorded at CPAR. In
contrast, the benthivore (eats benthic invertebrates) and piscivore (eats fish) guilds had 11 and 19
species groups, respectively. Benthivores include diving ducks like scoters, and piscivores
include the loons, alcids, mergansers, cormorants, and grebes.
Figure 26
Species Composition of the Four Feeding Guilds
Note. This visualization only includes species currently targeted by the CPAR seabird effort. The species
labeled other in the legend is Caspian Tern.
Similar concerns arise when interpreting the data grouped by feeding guild as when
looking at overall density. Because the CPAR survey efforts targeted a subset of the species
present at CPAR, these guild groupings do not include all potential species that meet the group
criteria. For example, there are likely more benthivore seabird species at CPAR than were
included in the survey effort. That being said, the surveyed species were targeted because they
67
are known to be the most frequent and abundant seabird visitors in the region and would likely
drive any guild related changes even if more species were incorporated into the surveys. I only
included data collected December 2016 or later when the survey effort expanded to include the
genus and family level groups. I did exclude Bald Eagle and Great Blue Heron since they are not
seabirds and therefore do not have an assigned guild.
Figure 27
Mean Piscivore Density by Season and Month
Note. Mean density of all piscivorous seabirds shown by season on top plot and by month in the bottom
plot. The points represent the overall mean density, and the error bars are standard deviation truncated at
zero.
The only significant difference between seasons was between the 2018-2019 and 20212022 season (Kruskal-Wallis, H4=13.83, p < 0.01, Dunn’s post hoc p = 0.02). There was no
significant change in piscivore density across seasons (negative binomial regression, p = 0.08).
There were significant differences across months (Kruskal-Wallis, H8=49.54, p < 0.01) with May
having lower densities than October (p < 0.01), November (p < 0.01), December (p < 0.01),
68
January (p = 0.04), February (p < 0.01), and April (p < 0.01). October had higher densities than
September (p = 0.03), and November has higher densities than March (p = 0.01) and September
(p < 0.01). This were relatively consistent with migratory patterns although this guild does have
the resident species like cormorants and alcids.
Figure 28
Mean Benthivore Density by Season and Month
Note. Mean density of all benthivore seabirds shown by season on top plot and by month in the bottom
plot. The points represent the overall mean density, and the error bars are standard deviation truncated at
zero.
There was no significant difference in benthivore density between (Kruskal-Wallis,
H4=3.34, p= 0.50) or across (negative binomial regression, p = 0.51) seasons. There were
significant differences in benthivore density across months (Kruskal-Wallis, H8=152.11, p <
0.01). Due to the migratory life histories of most species in this guild, benthivore density was
lowest in the fall and spring. May and September both had significantly lower densities (Dunn’s
post hoc, p < 0.01) than November, December, January, February, March, and April. October
69
was significantly lower than December (p < 0.01), January (p< 0.01), February (p < 0.01) and
March (p = 0.04). The piscivore and benthivore guilds warrant continued monitoring as they are
the two guilds found to be decreasing in the Salish Sea (Bower, 2009; Ethier et al., 2020; Vilchis
et al., 2015).
Knowing what months seabirds are most present may be meaningful to answer specific
questions or to grant permit applications, but the value of a seabird indicator is most apparent
when looking for charge over time. However, shifts to migration timing and duration of stay, like
those seen with Surf Scoter, are also indicative of resource availability. Seabirds will choose to
visit and stay in areas with healthy habitats and abundant prey. At Cherry Point, and across the
Salish Sea, positive progress towards marine conservation should be indicated by increasing
seabird numbers.
70
CONCLUSIONS
The Cherry Point marine bird monitoring project is an ongoing community science effort
that is nearing 500 surveys and has documented over 37,000 birds. The CPAR Birders are a
largely self-organized group of community members whose passion for birding has produced an
incredible data source for Aquatic Reserves managers. The CPAR marine bird monitoring
project was originally designed to replicate historical MESA surveys and therefore be compared
to the MESA seabird baseline. My approach did not include this historical comparison. Rather
than designing a project to answer a specific question, I approached an ongoing project and
identified questions it may be able to answer. This required extensive data exploration and
visualization. My thesis created a database, quality control process and replicable analyses for
the Aquatic Reserves Program that makes the data set available to Washington State scientists.
Of the total 29 marine bird species surveyed, seven species were encountered the most
often (> 50% of the time) and/or in higher numbers (> 1.5 birds per km2 on average): Surf
Scoter, Common Loon, Horned Grebe, Pelagic Cormorant, Bufflehead, Western Grebe and
Brant. Of these species, all except Pelagic Cormorant are migratory. Four are piscivores, two are
benthivores and one is an herbivore. This reflected the overall guild makeup of the targeted
species and implied that no single feeding guild dominated the area. Pelagic Cormorant density
was found to increase over the seasons while Surf Scoter density decreased. The other five
species had non-significant change across the seasons. Guild and overall densities were analyzed
across only the five most recent seasons due to the addition of more species in 2015 and 2016.
There were no significant changes in piscivore, benthivore or overall seabird density over this
time. Monthly fluctuations seemed to be driven predominantly by migratory patterns with the
possible exception of Surf Scoter.
71
I had proposed three specific seabird indicators: scoters, Pelagic Cormorant and Pigeon
Guillemot. Pigeon Guillemot, like all the alcids, were seen infrequently and that paucity of data
makes them a poor indicator. Scoters and Pelagic Cormorant are viable candidates due to their
high number of encounters. Due to the skill of the community scientists, the scoter indicator can
be made species specific and become Surf Scoter. Pelagic Cormorant and Surf Scoter use the
area differently—Pelagic Cormorant are residents that nest in the area over the summers and Surf
Scoter are winter migrants that molt in the area before continuing to nesting grounds farther
north. They also forage at different trophic levels and within differing parts of the habitat—
Pelagic Cormorant are piscivorous pursuit divers and Surf Scoter are benthivores. Incorporating
species that represent different aspects of the habitat is important when selecting an indicator
portfolio. Thus, Pelagic Cormorant and Surf Scoter are complementary indicators for CPAR.
It may be too early to tell which species are the most responsive to CPAR habitat shifts,
but I recommend that future analyses include Surf Scoter, Pelagic Cormorant and the piscivore
and benthivore feeding guilds. I also recommend that a diving seabird that predominantly preys
on forage fish be added to a CPAR indicator portfolio. Because alcids are seen so infrequently, a
loon species like Red-throated Loon or Pacific Loon—both of which have a documented
aggregate response to herring—may be an excellent choice.
Future inquiries and applications
As data collection continues, other directions for future analyses include exploring
migration timing, looking for correlations between human impact and changes to seabird density,
and further monitoring the relationship between seabirds and Cherry Point herring spawn. Other
statistical models may produce more accurate results. As mentioned in methods, assessing other
variables may increase the fit of the negative binomial model and, in some cases, a Poisson
72
model may be a better fit. A power analysis like that conducted by Crewe et al. (2012) to assess
the rigor of the CSC data set would also be incredibly valuable.
Seabirds may be entering and leaving the CPAR area earlier each year. Shifts to
migration timing may be explained by impacts elsewhere on the migratory route but may also
reflect habitat resources becoming available earlier in the CPAR region. Additionally, the
duration of time seabirds spend at CPAR may reflect the availability of resources provide by the
area.
The connection between seabirds and herring spawn is well studied, including several
current projects being conducted in the CPAR area (e.g., exclusion studies conducted by
WDFW). A better understanding of Surf Scoter, or other species, response to herring spawn in
the area may help predict spawning or contribute to herring spawn estimates. In the past two
years, herring spawning has been recorded near the industrial piers in the southern part of the
reserve. Genetic tests revealed that the spawning fish were not from the Cherry Point stock
(Sandell pers. Comm., April 2022), but tracking the response of seabird density to these new
spawning locations could provide insight into this relationship.
The value of this data is manifold. It promotes community engagement in Aquatic
Reserve monitoring and management efforts. The data provides a baseline for future comparison.
If there is a substantial change to human activity or habitat, this data forms a picture of the
seabird community prior to that change. If there was a permit application that would impact the
reserve habitats, managers could reference this data to see when seabirds are in the area,
including species of concern like those listed as threatened or endangered. Within the seven most
frequent and numerous species, Western Grebe and Common Loon are listed as candidate and
sensitive respectively by the State of Washington.
73
Recommendations for community science seabird efforts
Through my data management and analyses I identified common errors and suggested
protocol updates to address them. I also suggested updates to the data sheet to minimize future
errors. Appendix A includes examples of the old and appended data sheets. Table 4 compiles
suggestions to improve the pre-analysis data collection and review processes.
Specific to this effort, I recommend that an additional survey location be added at the
north end of CPAR. Currently only one of the three survey locations is within CPAR boundaries.
Adding a northern location at Point Whitehorn would better represent the overall reserve seabird
community, better replicate historical MESA survey locations, and include the area closer to the
current Cherry Point herring spawning location off Birch Head.
Table 4
Common errors, recommended solutions and actions taken for the data collection and review processes
associated with the CPAR marine bird surveys.
Error
Confusing/unclear tallies
and totals
Teams have non-matching
information. Some
conditions like glare (and
therefore visibility) may
differ but others like cloud
cover should be consistent
Different start times. Stop
times may differ depending
on the number of birds
encountered but start time
should always be the same.
Missing values and
incomplete data sheets
Solution
Before leaving the survey
location the recorder should
write the total count for each
species next to the tally and
circle it. Each team should
have their own data sheet.
Bold or highlight the data
sheet fields that should
match and have teams
double check before leaving
the field
My action
I created a new space on the
data sheet for the total count
of each species. This space
should never be blank as it
will either be zero or a total
count.
I bolded the sections of the
data sheet that should match
between both teams.
Add to the protocol that
teams must start at the same
time. Recorders
communicate and match
start time and end time on
the data sheet.
Before leaving the field,
inspect the data sheet for
Consistent start times and
communication are in the
protocol. I brought this error
to the attention of the CP
birders.
74
This is already in the
protocol but was discussed
with the birders.
Lack of data review
Incomplete or poor-quality
scans
completeness. No field
should be left blank.
All data should be entered
and scanned within one
week of collection. The
review will catch any
additional errors.
All scans should be
inspected during data entry,
and hard copies retained.
I created a QA/QC script and
system for the Aquatic
Reserves Program to help
review the data in addition to
RE Sources staff.
Early hard copies were the
main missing items. RE
Sources does retain hard
copies of the data sheets and
will continue to do so or
hand them off to DNR staff
if they no longer have
capacity.
For the future of this project, or for other entities who may wish to establish or refine
their own marine bird monitoring efforts, I compiled a list of recommendations in Table 5. These
points, while not necessary, will increase the potential applications of the dataset.
Table 5
Recommendations to improve the scale and application of marine bird monitoring efforts including the
pros and cons associated with each recommendation.
Include all bird species on the water and shoreline at the time of survey.
Pros
Cons
Focusing on only a few key species is a great Survey effort and time will increase and some
way to maximize volunteer time and
species like shorebirds can be challenging to
capability but including all species will
identify which may require additional
provide a more complete picture of the marine training.
bird community. Additionally, shorebirds that
forage in the intertidal habitat are a valuable a
seabird indicator portfolio (Pearson & Hamel,
2013).
Include a distance metric within the data either as a cutoff for observations or a
minimum distance.
Pros
Cons
The farther the bird is from the survey
We do not want to limit the field of
location, the hard it is to see and identify,
observation, but a good survey to emulate is
especially smaller diving birds. There are
the British Columbia Coastal Waterbird
algorithms that can account for the added
Surveys which include all individuals but note
challenge of distance and would strengthen
those within a 500m radius. We can
the analyses.
75
confidently assume that all seabirds within
that 500 m radius are seen and counted. This
would require a range finder and distance
calibration for the surveyors.
Quantify visibility (i.e., poor visibility = maximum 500 m, Good = maximum 1000m etc.).
This can tie in with the 500 m notation above.
Pros
Cons
Having visibility directly correlate with a
Distance estimations would require a range
maximum distance of observation will allow
finder and regular distance estimate
flexibility within the analysis of the data.
calibration for the volunteers.
Rather than excluding data with limited
visibility, analyses can account for changes to
the survey radius when calculating density
and still incorporate that data.
Quantify human disturbance
Pros
Cons
In addition to noting human activity, quantify There is an initial effort to create a Likert
scale and decide what activities qualify as
the level of disturbance you estimate that
high, medium, or low disturbance. Once the
activity to cause. This will make it easier to
scale has been established, this should require
explore the relationship between seabird
minimal effort to implement.
presence and anthropogenic disturbance.
Note: I created a four-point Likert scale for CPAR like below.
0 = no human activity/no disturbance
1 = minimal activity/no to little disturbance
2 = some activity/some disturbance observed
3 = lots of activity/birds seem to be avoiding the area
Note bird behavior
Pros
Cons
Birds in flight may not be interacting with or
For the CPAR surveys, this would require a
using the management area. This is especially rework of the current datasheet and may make
pertinent to small management areas like an
data collection and entry more time
Aquatic Reserve. By noting basic behavior
consuming.
(i.e., flying, foraging, resting, etc.) the
analyses can focus on birds actively
interacting with the managed habitats.
Survey at a consistent time of day
Pros
Cons
Birds exhibit different behaviors at different
There would be less flexibility around survey
times of day. Survey time consistency will
timing.
control for any impact that time of day may
have on seabird presence and abundance.
76
Final summary
The CPAR seabird data is a valuable resource for the Aquatic Reserves Program and my
thesis sets the groundwork for it to be referenced and utilized. This cumulative effort contributes
to the literature around regional seabird trends and patterns, explores the application of seabird
indicators to small marine management areas, and provides an example of incorporating
community-driven science into an agency’s decision-making framework. Community science
fills data gaps and maintains data continuity, especially for small agencies and programs.
Conservation efforts increasingly inhabit the space where resource managers, community
scientists and academic researchers overlap. By focusing on this nexus, my thesis promotes
collaborative monitoring and management for the purpose of marine conservation.
77
NOTES
If readers are interested in the Aquatic Reserves Program or have questions about this thesis,
please visit https://www.dnr.wa.gov/aquatic-reserves or contact the author at
erinstehr@gmail.com
78
REFERENCES
Anderson, E. M., Bower, J. L., Nysewander, D. R., Evenson, J. R., & Lovvorn, J. R. (2009).
Changes in avifaunal abundance in a heavily used wintering and migration site in puget
sound, Washington, During 1966-2007. Marine Ornithology, 37(1), 19–27.
Bishop, M. A., & Green, S. P. (2001). Predation on Pacific herring (Clupea pallasi) spawn by
birds in Prince William Sound, Alaska. Fisheries Oceanography, 10, 149–158.
https://doi.org/10.1046/j.1054-6006.2001.00038.x
Bower, J. L. (2009). Changes in marine bird abundance in the Salish Sea: 1975 to 2007. Marine
Ornithology, 37(1), 9–17.
Boyd, I., Wanless, S., & Camphyusen, C. F. (Eds.). (2006). Top predators in marine ecosystems:
their role in monitoring and management. Cambridge: Cambridge University Press.
Cairns, D. K. (1987). Seabirds as indicators of marine food supplies. Biological Oceanography,
5, 261–271. https://doi.org/10.3354/meps07078
Cleaver, F. C., & Frannet, D. M. (1946). The Predation by Sea Birds Upon the Eggs of the
Pacific Herring (Cupea pallasi) at Holmes Harbor During 1945. Biological Report No.
46B. State of Washington Department of Fisheries
Crewe, T., Barry, K., Davidson, P., & Lepage, D. (2012). Coastal waterbird population trends in
the Strait of Georgia 1999 – 2011 : Results from the first 12 years of the British
Columbia Coastal Waterbird Survey. 22(May), 8–35.
de la Cruz, S. E. W., Takekawa, J. Y., Wilson, M. T., Nysewander, D. R., Evenson, J. R., Esler,
D., Boyd, W. S., & Ward, D. H. (2009). Spring migration routes and chronology of surf
scoters (Melanitta perspicillata): A synthesis of Pacific coast studies. Canadian Journal
of Zoology, 87(11), 1069–1086. https://doi.org/10.1139/Z09-099
Einoder, L. D. (2009). A review of the use of seabirds as indicators in fisheries and ecosystem
management. Fisheries Research, 95(1), 6–13.
https://doi.org/10.1016/j.fishres.2008.09.024
Ethier, D., Davidson, P., Sorenson, G. H., Barry, K. L., Devitt, K., Jardine, C. B., Lepage, D., &
Bradley, D. W. (2020). Twenty years of coastal waterbird trends suggest regional patterns
of environmental pressure in British Columbia, Canada. Avian Conservation and
Ecology, 15(2), 1–24. https://doi.org/10.5751/ACE-01711-150220
Gaydos, J. K., & Pearson, S. F. (2011). Birds and Mammals that Depend on the Salish Sea: A
79
Compilation. Northwestern Naturalist, 92(2), 79–94. https://doi.org/10.1898/10-04.1
Geographic boundaries of Puget Sound and the Salish Sea. Encyclopedia of Puget Sound. (2015,
October 1). Retrieved April 16, 2022, from
https://www.eopugetsound.org/articles/geographic-boundaries-puget-sound-and-salishsea
Gustafson, R. G., Drake, J., Ford, M. J., Myers, J. M., Holmes, E. E., & Waples, R. S. (2006).
Status review of Cherry Point Pacific herring (Clupea pallasii) and updated status review
of the Georgia Basin Pacific herring distinct population segment under the Endangered
Species Act. In NOAA technical memorandum NMFSNWFSC 76 (Issue June).
http://ezproxy.library.ubc.ca/login?url=http://search.proquest.com/docview/904482068?a
ccountid=14656
Harvey, C. J., Williams, G. D., & Levin, P. S. (2012). Food Web Structure and Trophic Control
in Central Puget Sound. Estuaries and Coasts, 35(3), 821–838.
https://doi.org/10.1007/s12237-012-9483-1
Hines, E., & Jaeren, L. (2018). Marine Bird Abundance in the Cherry Point and Fidalgo Bay
Aquatic Reserves.
Hooker, S. K., & Gerber, L. R. (2004). Marine Reserves as a Tool for Ecosystem-Based
Management: The Potential Importance of Megafauna. BioScience, 54(1), 27–39.
https://doi.org/10.1641/0006-3568(2004)054[0027:mraatf]2.0.co;2
Kershner, J., Samhouri, J. F., James, C. A., & Levin, P. S. (2011). Selecting indicator portfolios
for marine species and food webs: A Puget Sound case study. PLoS ONE, 6(10).
https://doi.org/10.1371/journal.pone.0025248
Levin, P. S., Fogarty, M. J., Murawski, S. A., & Fluharty, D. (2009). Integrated Ecosystem
Assessments: Developing the Scientific Basis for Ecosystem-Based Management of the
Ocean. PLoS Biology, 7(1), e1000014. https://doi.org/10.1371/journal.pbio.1000014
Lewis, T. L., Esler, D., & Boyd, W. S. (2007). Foraging behaviors of Surf Scoters and Whitewinged Scoters during spawning of Pacific herring. Condor, 109(1), 216–222.
https://doi.org/10.1650/0010-5422(2007)109[216:FBOSSA]2.0.CO;2
Lok, E. K., Esler, D., Takekawa, J. Y., De La Cruz, S. W., Boyd, W. S., Nysewander, D. R.,
Evenson, J. R., & Ward, D. H. (2012). Spatiotemporal associations between Pacific
herring spawn and surf scoter spring migration: Evaluating a “silver wave” hypothesis.
80
Marine Ecology Progress Series, 457, 139–150. https://doi.org/10.3354/meps09692
Mcleod, K. & Leslie, H. (Eds.) (2009). Ecosystem-Based Management for the Oceans. Island
Press, 2009.
McManus, E., Durance, K., Khan, S., Ross Strategic, & Puget Sound Partnership. (2020).
Revisions to Puget Sound Vital Signs and Indicators. December, 90.
Montevecchi, W. A. (1993). Birds as Monitors of Environmental Change. Birds as Monitors of
Environmental Change, January 1993. https://doi.org/10.1007/978-94-015-1322-7
Nysewander, D. R., Evenson, J. R., Murphie, B. L., & Cyra, T. A. (2005). Report of Marine Bird
and Marine Mammal Component, Puget Sound Ambient Monitoring Program.
O’Hara, R. B., & Kotze, D. J. (2010). Do not log-transform count data. Methods in Ecology and
Evolution, 1(2), 118–122. https://doi.org/10.1111/j.2041-210x.2010.00021.x\
Palazzi, D., & Bloch, P. (2006). Priority Marine Sites for Conservation in the Puget Sound.
Washington State Department of Natural Resources Aquatic Resources Division
Pearson, S. F., & Hamel, N. J. (2013). Marine and Terrestrial Bird Indicators for Puget Sound.
Washington Department of Fish and Wildlife and Puget Sound Partnership, Olympia,
WA, 55 pp.
Piatt, J. F., Harding, A. M. A., Shultz, M., Speckman, S. G., Van Pelt, T. I., Drew, G. S., &
Kettle, A. B. (2007). Seabirds as indicators of marine food supplies: Cairns revisited.
Marine Ecology Progress Series, 352(1987), 221–234.
https://doi.org/10.3354/meps07078
Piatt, J. F., Parrish, J. K., Renner, H. M., Schoen, S. K., Jones, T. T., Arimitsu, M. L., Kuletz, K.
J., Bodenstein, B., García-Reyes, M., Duerr, R. S., Corcoran, R. M., Kaler, R. S. A.,
McChesney, G. J., Golightly, R. T., Coletti, H. A., Suryan, R. M., Burgess, H. K.,
Lindsey, J., Lindquist, K., … Sydeman, W. J. (2020). Extreme mortality and reproductive
failure of common murres resulting from the northeast Pacific marine heatwave of 20142016. PLoS ONE, 15(1). https://doi.org/10.1371/JOURNAL.PONE.0226087
Piatt, J. F., Sydeman, W. J., & Wiese, F. (2007). Introduction: A modern role for seabirds as
indicators. Marine Ecology Progress Series, 352, 199–204.
https://doi.org/10.3354/meps07070
R Core Team (2021). R: A language and environment for statistical computing. R Foundation for
Statistical Computing, Vienna, Austria. https://www.R-project.org/.
81
Richardson, S., D. Hays, R. Spencer, and J. Stofel. 2000. Washington state status report for the
common loon. Washington Department of Fish and Wildlife, Olympia. 53 pp.
Rodway, M. S., Regehr, H. M., Ashley, J., Clarkson, P. V., Goudie, R. I., Hay, D. E., Smith, C.
M., & Wright, K. G. (2003). Aggregative response of Harlequin Ducks to herring
spawning in the Strait of Georgia, British Columbia. Canadian Journal of Zoology, 81(3),
504–514. https://doi.org/10.1139/z03-032
Sandell, T., Lindquist, A., Dionne, P., & Lowry, D. (2019). 2016 Washington State Herring
Stock Status Report. September, 1–90. Washington State Department of Fish and Wildlife
Sipe, H. (2019). Multi-state occupancy modeling and optimal allocation of survey resources for
Common Loons in Washington State. University of Washington.
Sydeman, W. J., Thompson, S. A., Anker-Nilssen, T., Arimitsu, M., Bennison, A., Bertrand, S.,
Boersch-Supan, P., Boyd, C., Bransome, N. C., Crawford, R. J. M., Daunt, F., Furness, R.
W., Gianuca, D., Gladics, A., Koehn, L., Lang, J. W., Logerwell, E., Morris, T. L.,
Phillips, E. M., … Zador, S. (2017). Best practices for assessing forage fish fisheriesseabird resource competition. Fisheries Research, 194, 209–221.
Tam, J. C., Link, J. S., Rossberg, A. G., Rogers, S. I., Levin, P. S., Rochet, M.-J., Bundy, A.,
Belgrano, A., Libralato, S., Tomczak, M., van de Wolfshaar, K., Pranovi, F., Gorokhova,
E., Large, S. I., Niquil, N., Greenstreet, S. P. R., Druon, J.-N., Lesutiene, J., Johansen,
M., … Rindorf, A. (2017). Towards ecosystem-based management: identifying
operational food-web indicators for marine ecosystems. ICES Journal of Marine Science,
74(7), 2040–2052. https://doi.org/10.1093/icesjms/fsw230
Toft, J., Fore, L., Hass, T., Bennett, B., Brubaker, L., Brubaker, D., Rice, C., & Beach Watchers,
I. C. (2017). A Framework to Analyze Citizen Science Data for Volunteers, Managers,
and Scientists. Citizen Science: Theory and Practice, 2(1), 9.
https://doi.org/10.5334/cstp.100
Vilchis, L. I., Johnson, C. K., Evenson, J. R., Pearson, S. F., Barry, K. L., Davidson, P., Raphael,
M. G., & Gaydos, J. K. (2015). Assessing ecological correlates of marine bird declines to
inform marine conservation. Conservation Biology, 29(1), 154–163.
https://doi.org/10.1111/cobi.12378
Wahl, T. R., Speich, S. M., Manuwal, D. A., Hirsch, K. V., & Miller, C. (1981). Marine bird
populations of the Strait of Juan de Fuca, Strait of Georgia, and adjacent waters in 1978
82
and 1979. In EPA: Vols. 600/7-81-.
Ward, E. J., Marshall, K. N., Ross, T., Sedgley, A., Hass, T., Pearson, S. F., Joyce, G., Hamel, N.
J., Hodum, P. J., & Faucett, R. (2015). Using citizen-science data to identify local
hotspots of seabird occurrence. PeerJ, 3, e704. https://doi.org/10.7717/peerj.704
WDNR. (2010). Cherry Point Environmental Aquatic Reserve Management Plan. Washington
State Department of Natural Resources Aquatic Reserves Program
WDNR. (2022). Cherry Point Environmental Aquatic Reserve Management Plan - DRAFT.
Washington State Department of Natural Resources Aquatic Reserves Program
R Packages
Auguie, B. (2017). gridExtra: Miscellaneous Functions for "Grid" Graphics. R package version
2.3. https://CRAN.R-project.org/package=gridExtra
Clayden, J. (2019). shades: Simple Colour Manipulation. R package version 1.4.0.
https://CRAN.R-project.org/package=shades
Firke, S. (2021). janitor: Simple Tools for Examining and Cleaning Dirty Data. R package
version 2.1.0. https://CRAN.R-project.org/package=janitor
Garnier, S., Ross, N., Rudis, R., Camargo, A.P., Sciaini, M. & Scherer, C. (2021). Rvision Colorblind-Friendly Color Maps for R. R package version 0.4.0.
Grolemund,G. & Wickham, H. (2011). Dates and Times Made Easy with lubridate. Journal of
Statistical Software, 40(3), 1-25. URL https://www.jstatsoft.org/v40/i03/.
Henry, L. & Wickham, H. (2020). purrr: Functional Programming Tools. R package version
0.3.4. https://CRAN.R-project.org/package=purrr
Hothorn, T. (2022). TH.data: TH's Data Archive. R package version 1.1-1. https://CRAN.Rproject.org/package=TH.data
Kassambara, A. (2020). ggpubr: 'ggplot2' Based Publication Ready Plots. R package version
0.4.0. https://CRAN.R-project.org/package=ggpubr
Lenth, R.V. (2022). emmeans: Estimated Marginal Means, aka Least-Squares Means. R package
version 1.7.3. https://CRAN.R-project.org/package=emmeans
Makowski, D., Ben-Shachar, M.S., Patil, I. & Lüdecke, D. (2020). Automated Results Reporting
as a Practical Tool to Improve Reproducibility and Methodological Best Practices
Adoption. CRAN. Available from https://github.com/easystats/report. doi: .
83
R Core Team (2018). foreign: Read Data Stored by 'Minitab', 'S', 'SAS', 'SPSS', 'Stata', 'Systat',
'Weka', 'dBase', .... R package version 0.8-71. https://CRAN.Rproject.org/package=foreign
R Core Team (2019). R: A language and environment for statistical computing. R Foundation for
Statistical Computing, Vienna, Austria. URL https://www.R-project.org/.
Ripley, B. & Lapsley, M. (2019). RODBC: ODBC Database Access. R package version 1.3-16.
https://CRAN.R-project.org/package=RODBC
Sarkar, D. (2008) Lattice: Multivariate Data Visualization with R. Springer, New York. ISBN
978-0-387-75968-5
Therneau, T. (2021). _A Package for Survival Analysis in R_. Rpackage version 3.2-11,
<URL:https://CRAN.R-project.org/package=survival>.
Venables, W. N. & Ripley, B. D. (2002) Modern Applied Statistics with S. Fourth Edition.
Springer, New York. ISBN 0-387-95457-0
Wickham, H. (2016). ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag New York,
2016.
Wickham, H. (2021). tidyr: Tidy Messy Data. R package version 1.1.3. https://CRAN.Rproject.org/package=tidyr
Wickham, H. & Bryan, J. (2019). readxl: Read Excel Files. R package version 1.3.1.
https://CRAN.R-project.org/package=readxl
Wickham, H., François, R., Henry, L. & Müller, K. (2021). dplyr: A Grammar of Data
Manipulation. R package version 1.0.6. https://CRAN.R-project.org/package=dplyr
Wilke, C.O. (2021). ggridges: Ridgeline Plots in 'ggplot2'. R package version 0.5.3.
https://CRAN.R-project.org/package=ggridges
84
APPENDICES
Appendix A - Data Sheets
Data Sheet A-1: CPAR marine bird survey data sheets developed by the community scientists
and used until May 2022.
85
86
Data sheet A-2: Updated data sheets with recommended updates for the CPAR marine bird
survey effort.
87
88
