Formal Sampling Protocol

Young-of-the-year rockfish monitoring plan
for the Southern Salish Sea

Prepared by Adam K. Obaza1, Dayv Lowry2, James Selleck2, Kelly S. Andrews3, and Andrew O. Shelton3

Paua Marine Research Group, 4745 Del Monte Avenue, San Diego, CA, USA 92107
Protected Resources Division, West Coast Regional Office, National Marine Fisheries Service, National Oceanic & Atmospheric
Administration, 7600 Sand Point Way, Seattle, WA, USA 98115
3 Conservation Biology Division, Northwest Fisheries Science Center, National Marine Fisheries Service, National Oceanic &
Atmospheric Administration, 2725 Montlake Blvd E., Seattle, WA, USA 98112
1
2

A collaboration among:

April, 2023

Recommended citation:
Obaza, AK, Lowry, D, Selleck, J, Andrews, KS, and AO Shelton. 2023. Young-of-the-year rockfish
monitoring plan for the southern Salish Sea. National Marine Fisheries Service. Seattle, WA. 29 pp. +
App.

Program Contact:
Adam Obaza - Adam@pauamarineresearch.com

Table of Contents
Introduction .................................................................................................................................................. 1
Policy and Conservation Basis for YOY Survey Effort .................................................................................... 2
Current and Past YOY Survey Efforts ............................................................................................................ 4
National Oceanic & Atmospheric Administration (NOAA) ....................................................................... 4
Northwest Fisheries Science Center (NWFSC) ...................................................................................... 4
West Coast Region Office (WCR) .......................................................................................................... 4
The Ocean Wise Research Institute .......................................................................................................... 5
Reef Environmental Education Foundation (REEF)................................................................................... 5
Point Defiance Zoo and Aquarium (PDZA) ................................................................................................ 5
Seattle Aquarium ...................................................................................................................................... 6
Washington Department of Fish & Wildlife (WDFW) ............................................................................... 6
Reef Check Foundation ............................................................................................................................. 8
Components of a Robust Monitoring Plan ................................................................................................... 8
Accessible and Adaptive Surveys .............................................................................................................. 8
Complementary Survey Approaches ........................................................................................................ 9
Consistency ............................................................................................................................................. 10
Monitoring Survey Design........................................................................................................................... 10
Index Site Selection and Utilization ........................................................................................................ 10
Current and/or Historic Index Sites ........................................................................................................ 11
Central Sound...................................................................................................................................... 12
North Edmonds (kelp) ..................................................................................................................... 12
Blake Island ..................................................................................................................................... 12
Saltwater State Park........................................................................................................................ 13
Alki Cove 2 ....................................................................................................................................... 13
Norrander’s Reef/Sunrise Beach State Park ................................................................................... 13
Hood Canal .......................................................................................................................................... 13
Sund Rock ........................................................................................................................................ 13
Octopus Hole................................................................................................................................... 14
Flagpole Point ................................................................................................................................. 14
Sisters Rock ..................................................................................................................................... 14
San Juan Islands .................................................................................................................................. 15
Bell Island East ................................................................................................................................ 15
Broken Point.................................................................................................................................... 15
Skyline Wall ..................................................................................................................................... 15
South Sound ........................................................................................................................................ 15
Fox Island West Wall ....................................................................................................................... 15
Sunnyside Beach ............................................................................................................................. 15
Z’s Reef ............................................................................................................................................ 15
Whidbey Basin .................................................................................................................................... 16
Big Gulch (seagrass) ........................................................................................................................ 16
Mukilteo Lighthouse ....................................................................................................................... 16
Possession Point fingers.................................................................................................................. 16
i

Admiralty Inlet .................................................................................................................................... 16
Keystone Jetty ................................................................................................................................. 16
Additional Considerations for Index Site Selection............................................................................. 16
Random Sites .......................................................................................................................................... 17
Standardized YOY Survey Methodology ..................................................................................................... 17
Rockfish Identification ............................................................................................................................ 17
Survey Mode ........................................................................................................................................... 18
Habitat Data ............................................................................................................................................ 19
Example Datasheet/Slate Organization .................................................................................................. 20
Data Quality Assurance and Quality Control .......................................................................................... 21
Liability and Diver Safety............................................................................................................................. 21
Statistical Methodology .............................................................................................................................. 22
Sources of data ....................................................................................................................................... 22
Model framework ................................................................................................................................... 23
Version 1 model ...................................................................................................................................... 24
Estimation ............................................................................................................................................... 25
Future iterations ..................................................................................................................................... 26
Literature Cited ........................................................................................................................................... 27
Appendix A. Results from Version 1 Model ...............................................................................................A-1
Four-basins management framework ...................................................................................................A-1
Two-basins management framework ....................................................................................................A-4
Example of future model ...........................................................................................................................A-7

ii

List of Figures
Figure 1. Rockfish survey index sites sampled by the Seattle Aquarium in Puget Sound. The year in which
surveys began is provided to the right of each site name. ........................................................................... 6
Figure 2. Location of six index sites surveyed by the WDFW for bottomfish, including all size classes of
rockfish, from 1995 through 2010 (from LeClair et al. 2018). ...................................................................... 7
Figure 3. Example image from a web-enabled map tool for visualizing YOY survey effort across time and
space through adaptive querying. ................................................................................................................ 9
Figure 4. A diver surveying for rockfish at Saltwater State park. ................ Error! Bookmark not defined.3
Figure 5. Map of proposed index sites with associated coordinates………………………………………………………14
Figure 6. YOY yellowtail rockfish among boulders at Keystone Jetty......................................................... 16
Figure 7. Morphology and physical attributes of a generalized YOY rockfish, showing the dorsal spot, and
the two morphological categories used to classify fish of unknown species. ............................................ 17
Figure 8. Key to YOY rockfish species identification applying the dorsal spot and body shape criteria
described in text and shown in Figure 6. .................................................................................................... 18
Figure 9. Schematic depiction of timed roving YOY survey showing major sampling considerations and
data elements that warrant attention. ....................................................................................................... 20
Figure 10. Example of data sheet/slate layout currently used to collect YOY survey data by NOAA
fisheries. ...................................................................................................................................................... 20
Figure A-1. Mean (+/-SD) counts of young-of-the-year rockfish per minute surveyed across the (a) four
and (b) two management basin frameworks............................................................................................ A-1
Figure A-2. Trace plot of iterations 5000 – 15000 for each of the primary parameters of the model across
five chains after a 5000-iteration warm-up period................................................................................... A-2
Figure A-3. Comparison of observed counts of YOY rockfish to the corresponding posterior predictive
mean across 10,000 iterations for (a) professional (n = 535) and (b) citizen-science (n = 473) sampling
event. The red line is the one-to-one line and the blue lines are the 95% confidence interval for which
we would expect 95% of the data points to be included. ........................................................................ A-2
Figure A-4. Index of YOY rockfish abundance in each management basin from the four management
basin framework from 2015 to 2020 on (a) log scale and (b) normal scale. ............................................ A-3
Figure A-5. Year-to-year deviations in the mean density of YOY rockfish across all sampled basins in the
Puget Sound/Georgia Basin DPSs. The deviations are multiplicatively scaled to a value of 1 (e.g., YOY
densities were ~1.8 times more abundant in 2017 than the long-term mean across years. ................... A-3
Figure A-6. Variation in (a) mean density (count/min) of YOY rockfish across all sites in all basins (𝛾𝛾𝛾𝛾 +
𝜅𝜅𝜅𝜅𝜅𝜅) and (b) mean deviation among sites within individual basins (𝜅𝜅𝜅𝜅𝜅𝜅) in the Puget Sound/Georgia
Basin DPSs. Site-to-site deviations are multiplicatively scaled to the mean density (equal to 1) within
each basin (e.g., YOY densities at Edmonds Underwater Park were 6.4 times more abundant than the
grand mean of sites within the Main Basin). ............................................................................................ A-4
Figure A-7. Trace plot of iterations 5000 – 15000 for each of the primary parameters of the model across
five chains after a 5000-iteration warm-up period................................................................................... A-5
Figure A-8. Comparison of observed counts of YOY rockfish to the corresponding posterior predictive
mean across 10,000 iterations for (a) professional (n = 535) and (b) citizen-science (n = 473) sampling
event. The red line is the one-to-one line and the blue lines are the 95% confidence interval for which
we would expect 95% of the data points to be included. ........................................................................ A-5

iii

Figure A-9. Index of YOY rockfish abundance in each management basin from 2015 to 2020 on (a) log
scale and (b) normal scale......................................................................................................................... A-6
Figure A-10. Year-to-year deviations in the mean density of YOY rockfish across all sampled basins in the
Puget Sound/Georgia Basin DPSs. The deviations are multiplicatively scaled to a value of 1 (e.g., YOY
densities were ~1.6 times more abundant in 2017 than the long-term mean across years. ................... A-6
Figure A-11. Variation in (a) mean density (count/min) of YOY rockfish across all sites in all basins (𝛾𝛾𝛾𝛾 +
𝜅𝜅𝜅𝜅𝜅𝜅) and (b) mean deviation among sites within individual basins (𝜅𝜅𝜅𝜅𝜅𝜅) in the Puget Sound/Georgia
Basin DPSs. Site-to-site deviations are multiplicatively scaled to the mean density (equal to 1) within
each basin. For example, YOY densities at Edmonds Underwater Park were 17.2 times more abundant
than the grand mean of sites across the ‘Rest of DPSs’ basin. ................................................................. A-7

List of Tables
Table 1. Sites visited as part of the citizen science survey efforts at least five times between 2015 and
2020. ........................................................................................................................................................... 12

iv

Introduction
Rockfish (Sebastes spp.) are a diverse clade of long-lived, ovoviviparous fishes that are integral
components of food webs in the northeast Pacific Ocean, serving as both mid-level predators and an
important prey source for a myriad of consumers. In U.S. waters of the Salish Sea (often referred to as
Puget Sound or greater Puget Sound) 28 species of rockfish are known to occur (Palsson et al. 2009;
Pietsch and Orr 2015), though over 60 species occupy waters of the North American West Coast (Love et
al. 2002). Various aspects of their life history, including slow growth, late age of maturity, and episodic
recruitment success, allow rockfish populations to persist despite broad fluctuations in oceanic
conditions. These same attributes, however, can result in low population growth rates and long
generation times, making rockfishes highly susceptible to sustained fishery exploitation because
extraction can easily outpace recruitment potential. As rockfish species have been targeted by both
recreational and commercial fisheries throughout the 20th century, many species have experienced
population declines since at least the 1970s, and two species are listed in the Salish Sea under the
Endangered Species Act (ESA): yelloweye rockfish (Sebastes ruberrimus) as Threatened and bocaccio (S.
paucispinis) as Endangered (NMFS 2010). Several other rockfish species are considered Species of
Greatest Conservation Need under the Washington Department of Fish and Wildlife’s (WDFW) State
Wildlife Action Plan (WDFW 2015) and retention of all species of rockfish in U.S. waters of the Salish Sea,
except the westernmost portion of the Strait of Juan de Fuca, has been prohibited since May of 2010
(WDFW 2010; WAC 220-314-010 and -020). For a review of the history of rockfish exploitation and
changes in management in the Salish Sea, see Williams et al. 2010.
Conservation efforts have increased in response to population declines; however, significant data gaps
still exist with regard to fundamental biology of rockfishes in the Salish Sea. As rockfish develop from
planktonic larvae to benthic or semipelagic adults, their habitat associations change from open water to
nearshore vegetation beds and, eventually, the use of high-relief, complex bottom features such as
boulder piles and rock walls. In addition to spatial variation across habitat types and regions, rockfish
reproduction and recruitment varies within and among years in ways that are not well understood (Love
et al. 2002; Dauble et al. 2012; Haggarty et al. 2017; Markel et al. 2017). Oceanic and climatic drivers are
primarily responsible for changes in abundance during the planktonic stage (Field et al. 2021), though
the primacy of those drivers may change across large spatial scales (Caselle et al. 2010). These oceanic
conditions are believed to influence rockfish year class strength via prey availability and growing
conditions during early developmental phases (Laidig et al. 2007), but other factors closer to settlement
may also be determinants (Markel et al. 2020). Efforts to monitor recently settled young-of-the-year
(YOY) and juvenile rockfish are complicated by these highly variable patterns of recruitment, cryptic
appearance, elusive behavior, and changing habitat needs. Recovery and conservation efforts require a
broad understanding of individual species and life history needs beyond what is currently known. For
ESA-listed and other rare rockfish species, using more common species as surrogates to infer likely
settlement patterns may continue to be necessary for some time. This approach is supported by surveys
conducted elsewhere that indicate recruitment of numerous rockfish species fluctuate together (Ralston
et al. 2013; Stachura et al. 2014; Schroeder et al. 2019). By monitoring recruitment of all YOY rockfish in
the southern Salish Sea, policy makers may infer year class strength of ESA listed species, and apply this
information to numerous management issues outlined in the section below.

1

On September 18, 2017, a group of regional experts from regulatory agencies, conservation
organizations, and citizen science groups convened a workshop to coordinate YOY rockfish monitoring
efforts and work towards developing statistical analyses capable of integrating YOY rockfish data
collected under different survey methodologies. This document represents a synthesis of these
contributions to craft a unified path forward for YOY rockfish dive surveys in Puget Sound and
integration of YOY rockfish data from other surveys under a single modeling framework. This approach
of unifying disparate data sources for synergistic monitoring of a marine resource has precedent with
the recent monitoring plan for floating kelp in Washington (Berry et al. 2022). The utilization of relatively
shallow and nearshore habitats by YOY rockfish makes surveys on SCUBA possible. A visual census on
SCUBA allows for direct observation of fishes in vegetated, high-relief, and/or shallow habitats that may
be challenging for other sampling approaches. This plan will be used to guide dive data collection efforts
by a wide constituency and build a robust database of these observations capable of calculating an index
of recruitment across and/or within each of the management units that make up the yelloweye rockfish
and bocaccio distinct population segments (DPSs) in Puget Sound and the Georgia Basin (i.e., the Salish
Sea). As noted above, information for non-listed rockfish is also collected so that these species can be
used as surrogates in regions where data for ESA-listed species are sparse or missing. Though there is
substantial justification in the literature for recruitment synchronicity among rockfish species (Ralston et
al. 2013; Stachura et al. 2014; Schroeder et al. 2019), these data will still need to be interpreted carefully
to avoid any potential pitfalls from using indicator species. This index will be prudently used in
conjunction with other population status and threats-based data to inform management decisions
related to recovery of ESA-listed species (e.g. increased knowledge regarding the distribution of ESAlisted rockfishes at every stage of life will help focus consultation efforts on biologically significant
locations and/or habitat types). Future efforts may include regional partners from British Columbia, and
other monitoring organizations not incorporated here.

Policy and Conservation Basis for YOY Survey Effort
The need for a consistent, long-term YOY survey effort in the Salish Sea is emphasized by numerous
state and federal resource management documents. In 2009, the WDFW produced a summary report of
rockfish biology and population status for greater Puget Sound (i.e., all U.S. waters of the Salish Sea)
(Palsson et al. 2009). This document listed the need to identify juvenile and adult habitats, and to better
understand habitat associations at critical life stages, as crucial to future protection of rockfish at large
(Research and Data needs 8.1.1 and 8.1.2.). Since that time, WDFW has directed effort toward
evaluation of subadult and adult habitat associations using a remotely operated vehicle (Pacunski et al.
2013; 2020; Lowry et al. 2022; WDFW unpublished data), but a commensurate effort for YOYs and
juveniles has not occurred at the same scale. In 2010, the National Marine Fisheries Service (NMFS) of
the National Oceanic and Atmospheric Agency (NOAA) listed yelloweye rockfish and bocaccio under the
federal Endangered Species Act (75 FR 22276), drawing heavily on Palsson et al. (2009) for fundamental
aspects of biology and management to date in their status review (Drake et al. 2010). In response to
broader conservation concerns among rockfish species, the WDFW developed a fishery conservation
and recovery plan for all rockfish species in Puget Sound, recognizing that shared biology and habitat
requirements across rockfishes would result in conservation actions benefiting the suite of species
rather than just those that were listed (WDFW 2011). In this plan, the WDFW again identified the need
for better information about habitat use and natural recruitment in their Habitat Protection and
2

Restoration, and Research, sections. Drawing on new information from research and recovery actions
that the WDFW and NMFS partnered closely to develop, in 2017 NMFS released a recovery plan for
Puget Sound/Georgia Basin yelloweye rockfish and bocaccio, providing an explicit roadmap to restoring
populations of both species. Among other recommendations, this plan highlights a need for annual YOY
surveys throughout Puget Sound (NMFS 2017, Recovery Action 1.5) and engagement with citizen dive
groups to further rockfish recovery (NMFS 2017, Recovery Action 4.5). Furthermore, dive groups
collecting data for use in federal management fits within NOAA’s broader strategy of utilizing citizens to
advance conservation actions (NOAA 2021). Though disparate efforts to monitor YOY rockfish have
sprung up throughout the region to address this long-identified need, a lack of consistent funding and
sampling methods has prevented development of a cohesive sampling plan that can be broadly used to
inform management actions. By coordinating the research efforts described here and proposing an
integrated index development method, this approach can better inform important aspects of state and
federal resource agency recovery goals.
In addition to strategic planning for long-term recovery, data collected as part of this YOY survey
partnership can inform regulatory conservation actions in the short term. Projects conducted, funded,
or authorized by the federal government that may adversely affect a species listed under the ESA must
undergo a consultation process, per Section 7 of the ESA. These consultations use the best available
science (e.g. habitat use, geographic trends) to assess likely project impacts and any mitigation
measures that may alleviate them (NOAA 2022). YOY data collected from this program will provide best
available science for use during such consultations. Essential Fish Habitat consultations, which apply to
federally managed fisheries, including those for rockfish, and are required under the Magnuson-Stevens
Fishery Conservation and Management Act, will also use YOY survey data to inform impact assessments
for this crucial life stage. NMFS is also responsible for conducting five-year reviews of rockfish recovery
efforts in accordance with the ESA to document progress on conservation measures and evaluate
species status. In the most recent rockfish five-year review, scientists ran a multivariate autoregressive
state-space (MARSS) model with various data sources to estimate changes in rockfish abundance since
the 1970s (Tonnes et al. 2016; Tolimieri et al. 2017). The next five-year review in this series is expected
in the near future and will comprehensively synthesize new data available since 2017. As the citizenscience YOY dive survey database grows, these data will be incorporated into future models, potentially
improving their performance which ultimately would assist managers in better tracking population
trends.
Applications of this YOY sampling program may also inform a broad array of nearshore and offshore
restoration projects in the region. Successful habitat restoration activities incorporate a thorough
analysis of available data to maximize benefits to species of interest. Long-term data collected
throughout the Salish Sea would be useful in designing habitat restoration projects that seek to benefit
rockfish and other benthic, structure-associated species. For example, restoration of kelps, particularly
bull kelp (Nereocystis luetkeana), is currently receiving heightened attention in the region (Calloway et
al. 2020). Because these habitats are considered vital for rockfish recruitment, restoration projects
located in areas closer to known YOY hotspots may be given priority for funding or other support. The
scientific, restoration, and conservation benefits of YOY surveys are diverse, and can inform projects
with both short- and long-term time scales.

3

Current and Past YOY Survey Efforts
While a great deal of effort and coordination is required to create a robust YOY monitoring system, a
number of separate fish survey programs already exist that collect data useful to this effort.
Understanding the data collected by these programs will reduce duplication of effort during
development of new protocols, and allow for more robust analysis and long-term coordination where
data are comparable. There are six organizations currently coordinating and collecting YOY rockfish data
through dive surveys in greater Puget Sound, including NOAA, the Ocean Wise Research Institute, Reef
Environmental Education Foundation (REEF), the Point Defiance Zoo and Aquarium (PDZA), Reef Check,
and the Seattle Aquarium. The WDFW has extensive historical dive data collected from surveys
conducted between 1991-2010. Although WDFW currently has limited capacity to conduct YOY-focused
dive work, they may re-engage in the future as resources allow.

National Oceanic & Atmospheric Administration (NOAA)
Northwest Fisheries Science Center (NWFSC)
NOAA’s NWFSC has conducted quarterly SCUBA surveys in six eelgrass meadows since 2013. These
surveys record numbers of all fish (including YOY rockfish) and numerous macroinvertebrate species, as
well as eelgrass characteristics. The SCUBA survey methods employed by the NWFSC include a twoperson dive team swimming three, 30-m transects (2 m wide and 2 m high), making two passes on each
transect. On the outbound leg, the first diver swims a measuring tape out and records all fish
observations while the second diver swims behind and records macroinvertebrates. When large enough
to differentiate via morphological traits, fishes were identified to species. Sizes of each fish were also
recorded. Transects are separated by ~5 m and are located at ~5 m depth. On the inbound leg, both
divers record data on habitat characteristics (e.g., percent cover of vegetation species, density and
height of eelgrass shoots). Divers do not disturb the vegetation during the survey; only readily
observable fish are counted.
In addition to SCUBA-based surveys, Standard Monitoring Units for the Recruitment of Fishes (SMURFs)
and minnow traps have been deployed during specific research projects (2015-19) targeting peak
settlement periods to collect YOY and juvenile (ages 1-2) individuals, respectively. Both SMURFs and
minnow traps are deployed on one date and retrieved 2-7 days later, at which time fish are identified to
species and counted. An advantage of these survey tools is that fish can be closely examined by hand,
and genetic samples can be collected when species identification is not immediately possible.
West Coast Region Office (WCR)
Building on YOY sampling protocols of the WDFW, Seattle Aquarium, and others, NOAA’s WCR
developed a citizen science YOY dive survey protocol and sampling program with participation from local
dive clubs and other non-government organizations. Abundance data, by morphological group, are
collected using a timed roving SCUBA survey, with qualitative data taken on habitat and depth. Only fish
within 1 m to either side of the diver, and within 1 m of the bottom, are counted. Data are coded into
depth bins and by general habitat type and, as with the NWFSC protocol, divers do not disturb the
sediment or vegetation during surveys. This effort started in 2015, and over 2,000 transects have been
completed through 2022. Sixteen groups collaborate in data collection with surveys coming from over
4

100 participants. Sampling regularly occurs at 21 index sites and at least 86 haphazardly selected target
dive sites annually. Sampling effort has steadily grown since inception of the program in 2015, with
more than 2,296 survey minutes across 203 transects in January and February 2022 alone. This is
currently the most robust and spatiotemporally comprehensive YOY-focused sampling effort in the
region (Obaza et al. 2021) and serves as the foundation of the monitoring dive program proposed
herein.

The Ocean Wise Research Institute
Since 2005, the Ocean Wise Research Institute (ocean.org) has led an annual citizen science rockfish
survey effort in the Strait of Georgia. Data are collected from August to October by SCUBA divers
swimming timed roving surveys. All recorded rockfish are identified to species and broken into age
classes of adult (> 20 cm), juvenile (10-20 cm), and YOY (< 10 cm). Results are split regionally; a total of
twenty-eight survey regions stretch from the British Columbia’s central coast to Puget Sound and
Washington’s outer coast. Within the Salish Sea, twenty regions are identified on the Canadian side and
four on the US side. Distinct habitat types and depth bins are not included as part of the survey, though
surveys on soft bottom habitat are discouraged. Encounter rates are recorded as rockfish observed per
survey hour.

Reef Environmental Education Foundation (REEF)
REEF is an international marine conservation organization that has conducted over 16,800 citizen
science SCUBA surveys in Washington since 1998. Fish data are collected using a roving diver technique,
where divers swim through a site and record species identification and their abundance by tens and
hundreds. Metadata for each survey including total dive time, surveyor ID, date and location may also
be included. REEF provides divers with training and associated testing such that each participant can be
assigned a skill level rating. The higher testing level achieved; the greater weight is given to data
submitted by an individual. Diver data is reported through an online database where it is analyzed by
REEF staff. YOY rockfish are not the exclusive target of REEF surveys, which seek to more broadly
characterize the fish community, but they are an integral component.

Point Defiance Zoo and Aquarium (PDZA)
The PDZA is located in Tacoma and has a lengthy history of exhibiting marine specimens from local
Pacific Northwest waters. PDZA has contributed considerable magnification of survey effort by
committing staff and volunteer divers to Seattle Aquarium and State-run monitoring. Since 2015, the
PDZA has coordinated their sampling with NOAA’s WCR program, engaging local SCUBA divers in citizen
science survey efforts throughout the region and providing institutional support, such as dive vessels, a
lead safety officer, and coordinated data management. In late 2019, the PDZA began once more
coordinating their survey efforts with the Puget Sound Marine Fish Science Unit of the WDFW, but
momentum for this partnership was delayed by the global COVID-19 pandemic and a lack of available
WDFW divers.

5

Figure 1. Rockfish survey index sites sampled by the Seattle Aquarium in Puget
Sound. The year in which surveys began is provided to the right of each site name.

Seattle Aquarium
The Seattle Aquarium has a research conservation mission that parallels that of the PDZA. Since 2009,
however, the Seattle Aquarium has independently conducted dive surveys of the fish community,
including YOY rockfish, at 11 sites in Puget Sound, ranging from Point Hudson to Sund Rock in Hood
Canal, and Z’s Reef in southern Puget Sound (Figure 1). Surveys include quarterly counts of fish
(including YOY rockfish) along 100-m permanent transects at depths ranging from ~10-25 m. Divers
collect data by recording underwater video and calling out species using a tethered communications
system during the dive. Recordings are made while swimming one direction along the transect deploying
a tape measure, as well as while swimming the opposite direction during tape retrieval. Fish are
identified and counted during post-processing, and only those fish within 1 m of the tape are included.
The goal of this work is to document changes in species diversity across seasons and over years. Puget
Sound observations of YOY have been consistently low, but these surveys provide crucial nonoccurrence data as video recordings provide long-term records of verified absence.

Washington Department of Fish & Wildlife (WDFW)
WDFW staff have conducted SCUBA-based surveys that collected YOY rockfish data from 1991-93, 19952010, and 2015-17 (Frierson et al. 2018; LeClair et al. 2018; WDFW unpublished data). During the
6

earliest efforts, divers swam tape transects and collected presence/absence data on YOY rockfish, but
not density. These surveys were initially largely focused on identifying habitat characteristics associated
with YOY and juvenile rockfish presence to improve the design of artificial reefs (e.g., West et al. 1994;
Buckley 1997). A more formalized effort to estimate the density and abundance of marine fishes on
complex rocky habitats began in 1995 with the establishment of a series of fixed index stations (Figure 2;
LeClair et al. 2018). Though methods and sampling frequency varied over the years, surveys generally
consisted of quarterly to biannual sampling at three, 30-m long transects at shallow, mid, and deep
depths at each site. Actual transect depth differed by site, but most transects occurred on isobaths
between 5 and 20 m in depth. Similar to other protocols described above, only fish within 1 m to either
side of the diver and within 1 m of the bottom are included in counts. Contrary to other protocols
presented above, WDFW surveys included the use of PVC poles to flush fish from vegetation and the use
of lights to inspect beneath shelves and in crevices. The 1-m PVC poles also had 10 cm marks to improve
accuracy of fish length estimates. Surveys in 2015-17 were exclusively focused on waters around several
U.S. Navy installations but employed the same sampling protocol (e.g., Frierson et al. 2018). Associated
with these final few years of surveys, WDFW staff also tested the utility of modified shrimp traps with
fine mesh liners to monitor YOY rockfish.
The WDFW also conducts regular monitoring of fish populations with other tools that sometimes
encounter YOY rockfish, though they are not the primary focus. The most consistent survey methods
employed are benthic trawling and exploration with remotely operated vehicles (ROVs). Details on these
methods are available in Blaine et al. (2020), Pacunski et al. (2020), and Lowry et al. (2022). Encounter
rates of YOY rockfish are low in these surveys as neither method is designed to capture small-bodied
fishes – the bottom trawl primary net mesh is 10 cm2 and the ROV used through 2021 is equipped with a
standard definition video camera that does not record video at high enough resolution to regularly
detect or identify fish smaller than approximately 10 cm. Additionally, fish length data is not explicitly
collected during ROV surveys due to technological limitations.

Figure 2. Location of six index sites surveyed by the WDFW for bottomfish, including
all size classes of rockfish, from 1995 through 2010 (from LeClair et al. 2018).

7

Reef Check Foundation
Reef Check Kelp Forest Monitoring program is a citizen science-based program that has conducted
transect based density surveys in California since 2006. In 2022 Reef Check completed the first year of
surveying in Washington, training 50 citizen science divers and surveying 30 sites across the Salish Sea.
Each site consists of 6 core transects and an additional 12 fish only transects, and each transect is 30m
(2m wide x 2m high). A core transect includes a fish swath, invertebrate swath, a kelp swath and
universal point contact transect to characterize the reef (substrate, cover and relief). Divers do not
disturb the substrate or vegetation during surveys. Volunteer divers are trained to ID and count a set of
indicator species on each swath type; 30 fish species, 26 invertebrate species, and 13 species of kelp. Of
the 30 species counted on fish transects approximately half are rockfish species including YOY. The goal
of this long-term monitoring is to produce data that can be used for the management and conservation
of kelp forests and rocky reefs, and to involve the public in the scientific process to foster an educated
public, supportive of science–based management and ocean stewardship.

Components of a Robust Monitoring Plan
Biologists and managers from the 2017 YOY workshop identified two primary objectives to make use of
existing fish survey efforts in Puget Sound and expand on current monitoring programs: 1) use existing
data to model historic trends in YOY rockfish presence, absence, abundance, and density, where
possible; and 2) identify an appropriate monitoring approach to fill data gaps and improve future
assessments of rockfish recruitment trends. To address the second task, workshop attendees provided
guidance on components of a sufficient dive monitoring plan, determining that such a plan must be
accessible for many surveyors; adaptive to dynamic staffing, budget, and environmental conditions;
incorporate multiple complementary methods; and remain consistent over time.

Accessible and Adaptive Surveys
YOY rockfish presence can be highly variable across space and time as a result of a broad suite of factors,
some of which are poorly understood, making large sample sizes necessary to draw valid conclusions
about abundance and distribution trends. Sufficient effort will require participation from a consortium
of organizations and a method to facilitate data sharing among them. To serve this purpose, an online
portal has been created and will be made accessible for qualified surveyors to submit data. A surveyor is
considered qualified when they have demonstrated proficiency in species identification and survey
protocol. This tool is currently a simple shared spreadsheet (i.e., Google Sheet) but, if funding allows,
could evolve into a secure, dedicated data portal (e.g., Amazon Web Services Aurora database). The
platform selected for sharing survey data must have the capacity to store version-controlled
documentation of survey protocols and other metadata (e.g., Google Drive). This repository will be

8

secure, but openly accessible to monitoring
partners, and will also contain meeting notes,
presentations, published results, and other
content for contextualizing new project data.
While archiving survey data in such a manner for
use by analysts is crucial, information also needs
to be made available to surveyors so that if a
recruitment event or listed species is observed
additional surveys can be rapidly performed. Being
able to visualize the spatial extent of site coverage
and review frequently updated survey activity
would also reveal areas that may need increased
survey effort. Such needs can be met by a public
spatial database and web-enabled mapping tool to
display near real-time information on survey
Figure 3. Example image from a web-enabled map tool
for visualizing YOY survey effort across time and space
findings, and an existing pilot tool is undergoing
through adaptive querying.
evaluation and refinement
(https://public.tableau.com/app/profile/adam.obaza/viz/YOYRockfishSurveySites/YOY_Rockfish_Survey
Sites) (Figure 3). By providing timely information to participating surveyors, the monitoring approach
can be adapted to changing circumstances and leverage collaborative opportunities to collect more
comprehensive and useful data about ephemeral events. Partners can also easily scale and prioritize
survey effort to match staffing and funding resources, and coordinate with other partners to maximize
overall sampling efficiency across time and space.

Complementary Survey Approaches
Researchers in the southern Salish Sea utilize various approaches to survey YOY rockfish, as documented
above, depending on habitat type, site accessibility, depth, overall project goals, and various
administrative factors (e.g., funding, staffing). The most common method currently used for small,
cryptic fishes such as YOY rockfish is direct observation by SCUBA, and guidelines for that approach are
presented below. Together, these multiple procedures can provide a robust dataset that more fully
captures the spatiotemporal heterogeneity of the region. Options available include:
• direct observation through SCUBA surveys,
• SMURFs suspended in select areas of Puget Sound,
• remotely operated vehicle (ROV) and/or underwater video,
• and benthic trawling.
Within these broader procedures, methodological consistency is important so that time series of
observations can be used for trend analysis. If data collected use methods with similar selectivity and
bias, and produce comparable metrics (i.e., encounter rate, density) they may be analyzed together. If
not directly comparable, but internally consistent, there are statistical methodologies that allow trend
data from disparate methods to be incorporated into a single modeling framework (e.g., multivariate
autoregressive state-space [MARSS] model). For example, Tolimieri et al. (2017) assessed adult rockfish
status in Puget Sound using MARSS models to incorporate multiple data sets with different sampling
9

methods that crossed multiple management time periods; this analysis was able to successfully describe
shared, underlying trends in population status that were not readily apparent in the component
datasets in isolation (Tonnes et al. 2016; Tolimieri et al. 2017)

Consistency
Surveyors have varying professional backgrounds and training, so establishing consistent characteristics
of a SCUBA-based sampling protocol is necessary to achieve repeatable and comparable results. This
consistency includes use of terminology and methodology. For example, a YOY rockfish must be well
defined such that field identification is clear and all participants are recording the same size class. NonSCUBA survey methods employed by the WDFW and NWFSC already involve consistent protocols
described in previous reports (e.g., Frierson et al. 2018; Blaine et al. 2020; Pacunski et al. 2020; and
Lowry et al. 2022) and provide a valuable example to the present effort. Given that these surveys are
conducted almost exclusively by singular teams over time, with relatively little turn-over, there is less
concern about the need for inter-partner consistency.
Sampling effort, geographic coverage, and temporal variation must also contain elements of consistency
for trend analysis to be successful. Establishing a series of index stations that are sampled at least
quarterly around a similar time to create a comparable time series is fundamental to detecting change in
YOY abundance through time and across geography. Data from these core sites may then be put into a
broader context by the addition of adaptive efforts throughout the region. Having consistent survey
guidelines and spatiotemporal coverage allows data from a diverse suite of participants to be more
readily integrated and synthesized when evaluating recruitment trends.

Monitoring Survey Design
Index Site Selection and Utilization
Index sites are often surveyed as a representative subset of locations that accurately depict trends
within a larger area. For the purposes of this monitoring plan, their primary role will be to calibrate data
collection from different survey groups to ensure any corrections may be made for the most accurate
model output. Therefore, index sites will be used to ensure data from all sites (i.e. index and random)
surveyed for this effort may be prudently applied to the model presented below. They may also be used
for their traditional application, but the existing lack of data from these index sites, and a preference for
sites that frequently contain YOY to improve calibration, will require careful examination of biases. The
list presented below should be considered living, and open to revision as the YOY rockfish database
grows.
Currently, active index sites are in each of the six sub-basins of greater Puget Sound (Admiralty Inlet,
Central Sound, South Sound, Whidbey Basin, San Juan Islands, and Hood Canal), facilitating description
of both local trends in recruitment and variation in inter-basin dynamics. Site selection has been based
on the presence of habitat known to support YOY rockfish (kelp forests, eelgrass beds, rocky and
artificial reefs), other beneficial geological features (e.g., rugose soft bottom, or hardpan), ease of
access, and consistency of YOY encounter. Not all sites and habitats are equally used by rockfish,
10

depending on factors such as species preference, life-history stage, and prey availability (Buckley 1997;
Dauble et al. 2012; Palsson et al. 2009; LeClair et al. 2018; Pacunski et al. 2020; Andrews et al. 2021;
Lowry et al. 2022); however, it is important to include these sites to evaluate if YOY rockfish occasionally
appear during major recruitment events, providing opportunities to document previously unknown sites
or changes in species distribution. If a habitat type offers lower value to YOY but has extensive coverage
in the region, it may still be a major contributor to population size. Without at least nominal effort spent
sampling what are assumed to be less suitable habitats, adequately describing utilization of the habitat
mosaic of the southern Salish Sea by YOY rockfish will not be possible.
Monitoring frequency should be high enough to reasonably identify when a major recruitment event
occurs (~ 6 weeks in duration; Moser and Boehlert 1991), though may be at lower frequencies to assess
year class strength (Doherty and Fowler 1994). Frequent monitoring events help to track potentially
large cohorts over time, and to explore survival relative to specific biological and physical variables.
However, mobilizing frequent field efforts is resource intensive and must be balanced with the expense
of spatial replication. A benefit of this collaborative effort will be the sharing of monitoring index sites
among multiple organizations, allowing for higher temporal and spatial replication without the burden
falling to any one entity. Having a diverse partnership also benefits the team at large by diversifying
options for financial support to maintain sampling effort into the future.

Current and/or Historic Index Sites
Research observations from the 1990s through 2020, along with existing citizen and professional YOY
survey data from 2015-21, were used to compile a preliminary list of valid index sites. To refine this list,
an exercise was conducted on the recent survey data to identify locations with high frequencies of YOY
encounter. These results, along with information on ease of access and habitat type, generated the
subset of sites with best opportunity to calibrate different survey approaches.
Data for this exercise were taken from YOY surveys conducted using a timed roving diver survey in
discrete habitat types and depth bins at 107 sites throughout Puget Sound. Three hundred eighty-nine
survey events, made up of 1,874 roving transects were completed across these sites. Effort was not
evenly distributed across sites, as citizen participants often repeat surveys in the areas they regularly
dive, and access to certain sites is limited by boat use, tidal exchanges, and other factors.
Recorded rockfish for all transects within a sampling event (i.e., all surveys at a site on a single day) were
aggregated to a single measure of presence/absence. The number of sampling events in which a YOY
was located were divided by total sampling events within each year to return an annual site-specific
frequency of YOY reporting. Those values were then averaged across all years a survey took place at a
given index site to return a mean frequency of YOY presence. This approach was preferable to other
metrics, such as YOY encounter rate, because reliability of YOY presence at a site is of greater concern
than the number of rockfish a surveyor may encounter. Habitat characteristics of each site were then
reviewed in the context of YOY presence to ensure a variety of habitat types were included.

11

Of the 107 sites surveyed, 86 were surveyed once or twice while only 20 were surveyed five or more
times (Table 1). YOY were recorded at least once at 34 sites. These observations identify sites with high
likelihood of repeated YOY rockfish encounter but do not represent the diversity of sites and habitats in
Puget Sound. Using this approach, index sites were weighted towards artificial reefs in Central Sound, so
additional index sites are needed to diversify surveyed habitat types. Selection of an index site for the
purposes of comparison might
require specific survey guides for a
Table 1. Sites visited as part of the citizen science survey efforts
site. For example, Edmonds
at least five times between 2015 and 2020.
Underwater Park in Central Sound
has been surveyed 29 times
Site Name
Survey Events 2015-2021
Basin
throughout this effort, with a 91%
Keystone Jetty
15
Admiralty Inlet
mean YOY encounter frequency,
Edmonds Underwater Park
29
Central
Les
Davis
Marine
Park
28
Central
making it a prime index candidate.
Alki - Cove 2
23
Central
Edmonds is a large site, though, and
Point
Ruston
Ferry
21
Central
it would be possible for two dive
Saltwater State Park
18
Central
groups to conduct long surveys and
Dickman Mill
12
Central
never overlap. Without information
Lobster Shop Wall
11
Central
on how evenly YOY are distributed
Redondo Beach
10
Central
on the site, disparate surveys from
KVI Tower
7
Central
each group may both be accurate. A
Sunrise Beach State Park
7
Central
specific survey guide for this site
Sund Rock
7
Hood Canal
would make data more comparable.
Flagpole Point
5
Hood Canal
Bell
Island
East
6
San Juan
The below list provides a variety of
Rosario Beach
5
San Juan
sites with individual strengths and
Fox
Island
West
Wall
6
South
weaknesses relative to providing
Day Island Wall
5
South
adequate survey information for YOY
Sunnyside Beach State Park
5
South
rockfish and serve as a preliminary
Mukilteo Lighthouse
14
Whidbey
list of index sites. Figure 5 displays a
Mukilteo T-Dock
12
Whidbey
map with associated coordinates.

Central Sound

North Edmonds (kelp)
This site was visited by NWFSC divers from 2017-19 while collecting YOY rockfish for stable isotope and
growth studies. Densities in 2017 were very high, and in 2018 and 2019 there were large numbers of
YOY located in the understory kelp Laminaria saccharina. This relative consistency suggests this site
receives larval supply in most years and is likely one of the best locations to monitor relative changes in
abundance of YOY rockfish in Puget Sound. This site is also monitored for kelp canopy (Nereocystis
leutkeana) by the Snohomish County Marine Resources Committee.
Blake Island
The Blake Island Artificial Reef was constructed in 1980, and was a regular survey site for the WDFW for
15 years from 1995-2010 (LeClair et al. 2018). The reef consists of concrete rubble and other materials,
with a natural substrate of sand and gravel around it. The site is only accessible by boat, and can be
12

tidally variable because of its location along the Kitsap Peninsula. Surveys conducted from 1995-2010
found that brown rockfish were the dominant rockfish species.
Saltwater State Park
Located partway between Tacoma and Seattle, this
artificial reef is uniquely structured with discrete
fingers leading from ~50 feet to ~80 feet in depth,
depending on the tide. The fingers abut a cobble ramp
leading towards shallow eelgrass habitat. This site has
been surveyed 12 times with a 71% YOY encounter
frequency. Reef fingers and cobble ramps are surveyed
in a single dive (Figure 4) and would make for discrete
habitat units that could be replicated across survey
groups.

Figure 4. A diver surveying for rockfish at Saltwater
State park.

Alki Cove 2
Similar to Saltwater State Park, this site is very close to Seattle, reducing the effort required to conduct a
dive. The reef, composed of a series of artificial structures, is not feasible to cover in a single dive.
However, anchored lines connect the structures and it would be possible to create a repeatable survey
path. After 17 surveys at this site, YOY encounter frequency is 98%.
Norrander’s Reef/Sunrise Beach State Park
Though the majority of dive sites in Central Sound are artificial reefs, rockfish recruitment at these sites
may not be representative of many areas in the Sound. In fact, many artificial reefs in Central Sound
appear to have higher mean encounter frequencies than natural reefs. Inclusion of at least one natural
reef for this region is prudent, though each of the two likely candidate sites, Norrander’s Reef and
Sunrise Beach State Park, have drawbacks. YOY encounter frequency at both sites is low (Norrander’s –
50%, Sunrise – 17%), though these values are derived from only seven total surveys conducted across
the sites. Norrander’s Reef is on Bainbridge Island, requiring a ferry ride for most divers and reducing
access. Sunrise Reef is north of Gig Harbor, which is farther from many population centers, has difficult
shore access, and experiences strong tidal currents. Both sites have discrete natural reefs that provide
high likelihood of survey area overlap, making more reliable replicate surveys.
Hood Canal

Sund Rock
A natural reef with high and low relief sections that has received the greatest survey effort of anywhere
in Hood Canal, with a mean encounter frequency of 58%. It is the only site discussed here at which a
YOY yelloweye rockfish has been encountered on a survey. There is a cost to access this site from shore
and coordination with the operating group must be completed beforehand, likely reducing interest.
However, Hood Canal is some distance from population centers and if participants are planning a full
day dive trip, there are multiple sites in the area to survey.

13

Octopus Hole
A natural reef with shore access located north of Sund Rock. It was designated as a WDFW Conservation
Area in 1999 (WAC 220-303-040). It is a relatively small site, but YOY have been encountered as part of
the NOAA Citizen Science Diving program. There are two reef walls that extend down to -80 feet, and it
is not subject to strong tidal currents, making it a popular site for recreational divers.
Flagpole Point
A natural reef with a distinct deep (> 80 feet) section and intermediate/shallow (< 50 feet) sections on
the west side of Hood Canal. Over five surveys were conducted, but no YOY rockfish have been
recorded. Despite the paucity of YOY, high encounters with adult rockfish and the shore-accessible
natural reef across multiple depth bins make this site worth continued visitation. Note that access is
paid through Mike’s Beach Resort, limiting access for some citizen science divers.
Sisters Rock
Sisters Rock is situated just southwest of the Hood Canal Bridge. It is a tidally influenced site, mostly only
accessible by boat. Various rockfish species are present at the site, and while YOY have been captured in
WDFW shrimp pot traps, no YOY have yet been observed on roaming dive surveys.

Figure 5. Map of all proposed index sites color coded by basin. Numbers on map correspond with table.

14

San Juan Islands

Bell Island East
This site has high relief rocky reef with heavy algae in the shallows, along with interspersed kelp. Only
one YOY has been located in four survey attempts, and encounters are generally low in this basin. As
with many sites in the San Juan Islands, tidal current is a limiting factor. This site is located in narrow
Wasp Passage between Orcas and Shaw Islands, making weather-influenced access more reliable by
virtue of substantial shielding.
Broken Point
This site is similar to Bell Island, with high relief rocky reef and dense algae in the shallows, but no kelp is
present on this site. YOY have been located on both the east and western side of the point and no
preference was observed to one side or the other. The site is larger than Bell Island East and is unlikely
to be completely surveyed in a single dive. This site is located between Orcas and San Juan Islands,
making access more reliable.
Skyline Wall
This natural reef southwest of Anacortes has the benefit of being located on the mainland, while still
counting as within the San Juan basin, obviating the need for participants to purchase and plan a ferry
ride. The habitat at this site includes natural reef, eelgrass and a seasonal Nereocystis bed. Through four
survey events, YOY have been recorded 50% of the time. Careful tidal planning is a must on this site, as
it should only be surveyed at slack before an ebb current.
South Sound

Fox Island West Wall
This ledge habitat in ~50 feet of water is located a short drive from Gig Harbor, minimally affected by
current, and has a mean YOY encounter frequency of 75%. The ledge is relatively easy to locate and not
very long, making replicated shore-based surveys feasible. Additional sites around the island are also
accessible by boat, making selection of random sites to pair with this index site straightforward.
Sunnyside Beach
Rock surrounding a pipeline creates a discrete habitat that enables replicated surveys by various survey
groups. This site has beach access and is among the most frequently surveyed sites in South Sound. YOY
encounter frequency is 33%.
Z’s Reef
This natural reef on the north side of Fox Island is only accessible by boat. The habitat of interest is a
discrete, 5-15-foot-high rocky feature in 40-60 feet of water that runs for almost 200 yards. This site has
been surveyed twice and YOY were encountered on both occasions.

15

Whidbey Basin

Big Gulch (seagrass)
This site is one of the sites that NWFSC divers have surveyed for
fish, invertebrates, and seagrass characteristics on a quarterly
basis since 2015. YOY rockfish were routinely observed in the
eelgrass (Zostera marina) beds in 2017-19 during stable isotope
and growth study collections.
Mukilteo Lighthouse
A short clay wall approximately 200 feet long, in 55-60 feet
(MLLW) contains many juvenile and YOY rockfish, and mean YOY
Figure 6. YOY yellowtail rockfish among
encounter frequency is 93%. The site is easy to access from the
boulders at Keystone Jetty
beach, with available close parking, but can experience strong
tidal currents. No other site in the Whidbey Basin has been surveyed as frequently. The discrete wall
habitat allows for comparison of results across survey groups. The wall is also natural habitat and may
provide an accurate assessment of overall rockfish trends in Puget Sound.
Possession Point fingers
On the eastern edge of the southern tip of Whidbey Island is a shore-accessible reef feature that
descends to greater than 80 feet. The high-relief habitat that continues to approximately 30 feet has
numerous locations for YOY to shelter. This site has been surveyed twice and YOY were encountered on
both occasions. Coordination with state parks is highly recommended to obtain a gate code so that
vehicles may be brought close to the entry point.
Admiralty Inlet

Keystone Jetty
This site is among the most popular in Puget Sound, though it requires a ferry ride from Port Townsend
or a long drive from Seattle and careful dive planning to avoid currents. The site was designated as a
Conservation Area in 2002, and the WDFW conducted regular transect dive surveys from 1995-2010
(LeClair et al. 2018). The habitat is a mix of artificial reef (a jetty with a dilapidated pier slightly to the
south) and kelp forest that frequently has schools of adult yellowtail and black rockfish. Mean YOY
encounter frequency is 75%, and multiple species have been observed (Figure 6). The WDFW found that
the dominant species is Puget Sound Rockfish, followed by black and copper rockfish (LeClair et al.
2018).
Additional Considerations for Index Site Selection

The goal of this exercise was to select potential YOY index sites using the best available data. However,
there are many possible sites in Puget Sound that have not been surveyed for a variety of reasons.
Seasonality was not included in this exercise and may have had an impact on accessibility and
observations at sites. Additional index sites will be added to this list as new information becomes
available.

16

Random Sites
Numerous factors influence rockfish settlement location, including tidal currents, wind drift, food
availability, and intricacies of habitat preference (West et al. 1994; Buckley 1997; Kashef et al. 2014).
Many of these factors are complex, and their interactions relevant to recruitment are not fully
understood in the Salish Sea. To capture these spatial dynamics, surveys must be conducted throughout
the region at locations that encompass the broadest degree of variability for these presumed forcing
factors. Surveying in as wide an area as possible will fill data gaps on spatial recruitment and may reveal
novel sites of high value to rockfish recruits. More sites will also inform knowledge gaps on physical and
biological characteristics important to settlement, as described below in the modeling section, allowing
for more robust and comprehensive data analysis. The spatial data may inform understanding of
rockfish recruitment in the region, presenting an opportunity to adaptively adjust the monitoring plan to
accommodate expanded sampling. Selection of these sites will be at the surveyor’s discretion, but
should include features presently known to support rockfish recruits, including macroalgae, seagrass, or
reef habitat. These sites will be surveyed opportunistically and may eventually become index sites.

Standardized YOY Survey Methodology
This plan documents recent use of multiple methods for recording YOY rockfish occurrence, abundance,
and density. In some cases, these methods are employed to sample the fish community at large, rather
than focusing specifically on YOY rockfish and often require specialized equipment and/or permits (e.g.,
SMURFs, ROVs, trawling). While these methods provide valuable contextual data to supplement focused
collection of YOY rockfish distribution and abundance data,
they typically cannot be implemented by citizen scientists.
Here, we present methodological guidelines for SCUBA surveys
as the most common and practical method for use by a broad
user group to survey YOY rockfish. By using this standardized,
focused sampling protocol, organizations and individuals may
collect data that are readily integrated with existing datasets
for long-term trend analysis. This protocol represents the
minimum standards for inclusion of data in the aggregate
monitoring database and partners are encouraged to collect
additional data as necessary to meet organizational goals and
maximize dive time efficiency. Additional support in
implementation of this protocol may be obtained by
contacting Adam Obaza at adam@pauamarineresearch.com.

Rockfish Identification
Consistent with multiple existing survey programs, a YOY is
defined as any rockfish under 10 cm in total length (Palsson et
al. 2009; LeClair et al. 2018). An exception to this length-based
definition is made for the smallest rockfish species regularly
encountered in the Salish Sea, Puget Sound rockfish (Sebastes
emphaeus). If surveyors can identify an individual Puget Sound
17

Figure 7. Morphology and physical
attributes of a generalized YOY rockfish,
showing the dorsal spot, and the two
morphological categories used to classify
fish of unknown species.

rockfish to the species level, it should only be included in the dataset as a YOY if it is under 6 cm.
Identification of YOY rockfish to species level can be difficult due to their small body size and limited
morphological differentiation between groups of closely related species at small body sizes. If individuals
cannot be identified to the species level, counts should be grouped using two morphological attributes:
presence or absence of a dark spot on the spinous portion of the dorsal fin (i.e., dorsal spot vs. no spot),
and the overall body shape in profile (i.e., deep body vs. elongate body) (Figure 7).
Figure 8 provides a key to guide species identification. Species-specific data are preferred when they can
be accurately obtained. Ideally, photo or video documentation should be provided, but is not
mandatory. When photo/video evidence is available, it may be used to assess surveyor bias, identify
difficult species to identify, determine age classes, or indicate a need for additional training.

Figure 8. Key to YOY rockfish species identification applying the dorsal spot and body shape criteria
described in text and shown in Figure 6.

Survey Mode
There are two SCUBA-based survey modes defined for the YOY rockfish monitoring plan to standardize
survey effort: timed roving and band transect.
•

Timed Roving Survey - A surveyor swims freely, recording all YOY rockfishes encountered within
their swimming path, and documenting the total survey duration. Surveys predominantly target the
area within 1 m of the substrate, though exceptions may be made for surveys in mid-water or in
kelp canopy, provided that deviation is accurately noted. This method provides an indicator of effort
(i.e., counts of YOY rockfish encountered per unit time) that enables comparison within and
between locations over time during analysis.
18

•

Band Transect – A surveyor deploys a transect tape of known length, or otherwise validates
swimming distance using fixed artificial or natural structures, and surveys for YOY rockfish within a
box bounded by the tape and a predetermined width (typically 1 m, based on reliable visibility) on
either side of it. An advantage of band transects is it minimizes the likelihood of repeat observations
of the same fish as the diver records observations in only one direction. This method provides a
density estimate (i.e., counts of YOY rockfish per unit of area). Time of survey along the transect may
also be collected to allow for direct comparison with data from timed roving surveys.
Roving surveys where time is not recorded are useful for identifying potential new index site
locations; however, this survey type is not standardized and the data collected would not be
included in this monitoring plan.

Habitat Data
Rockfish in Puget Sound are known to inhabit different depth zones and habitat types (Hallacher and
Roberts 1985; Love et al. 2002; Drake et al. 2010; Blaine et al. 2020; Pacunski et al. 2020; Lowry et al.
2022). In addition to recording the number of YOY rockfish encountered and time spent surveying, data
on depth and habitat type should be recorded for each survey. For the purposes of this survey protocol,
habitat can be characterized using broad categories (e.g., rocky reef, seagrass, kelp forest), though
description of the degree of vegetative coverage (e.g., 25% eelgrass in patches around rock piles) is also
helpful. Details of common example habitat types are provided in Figure 9. Depth may also be described
as within a certain range and habitat as the primary type encountered during survey (e.g., 25-28 m over
exposed bedrock). Attributes of the habitat should be recorded with as much specificity as possible
without compromising the ability to accurately count and identify YOY rockfish as outlined in Figure 8.
For this reason, having one diver record fish while their partner evaluates habitat may be preferable. A
helpful depiction of major elements of the survey protocol is provided in Figure 9 for reference.

19

Figure 9. Schematic depiction of timed roving YOY survey showing major sampling considerations and data
elements that warrant attention.

Example Datasheet/Slate Organization
Based on the methodological guidelines presented here, each survey will include information on depth,
habitat type, number of YOY rockfish counted, and either duration of survey or survey dimensions.
Because depth and habitat types may change multiple times at a given site, multiple surveys (e.g.
different habitat types or depth bins) may be recorded during a single dive. An example datasheet or
slate organization is provided below (Figure 10) in accordance with NOAA WCR’s YOY survey program.
Information on site name, basin, and date is necessary to spatially and temporally track surveys and
standard names should be applied systematically to sites that are revisited multiple times (e.g., Alki
Beach Site 1a). Groups using this standard survey method are welcome to modify the proposed slate
organization to incorporate species-specific count data or otherwise meet their needs, but care should
be taken to include all core data elements described above.

Figure 10. Example of data sheet/slate layout currently used to collect YOY survey data by NOAA fisheries.

20

Data Quality Assurance and Quality Control
This plan is intended to encourage participation from data collectors with diverse experience levels,
ranging from beginner citizen scientists to expert marine biology professionals. Given this range, a level
of quality assurance must be applied to the data to ensure accuracy and facilitate inclusion in trend
analysis. The following guidelines are suggested for individual participating programs:
●

●

●

●

Meaningful documentation on survey protocol, data collection, and species identification should
be sought out and provided to all participants. These materials may include survey aides, such as
the slate depicted in Figure 10 or a reference guide (Figures 7, 8, and 9). These documents are
readily available through project leads in print and pdf forms. Regular presentations, in-water
training, and regular feedback to participants will also improve participant performance.
Index site creation outlined in this plan provides an opportunity for multiple survey groups to
collect data at the same site. Comparison among these data, provided they are close enough in
time, will enhance understanding of inherent variability in observation efficiency and other
sources of bias. This could lead to correction factors applied to data from specific divers or
groups to make data more comparable. Participating programs are also welcome to identify new
index sites with adequate encounter rates and survey effort such that site-specific trends can be
generated for comparison with other locations. Establishment of new sites should be
coordinated with the existing NOAA citizen science program for maximum utility.
Project leads should develop dedicated, systematic methods for error checking and otherwise
validating data collected by their program before submitting it to the shared database. At a
minimum, this should involve: immediate post-dive error-checking to flag spurious or conflicting
observations and questions about fish identification; a double-entry or spot-check procedure for
catching data entry errors; and consideration of a data confidence metric to be included with
their submission based on diver experience, survey conditions, or other factors that may affect
data accuracy. If data do not meet standards, they will be omitted from analysis.
More affordable photography equipment, including GoPro and other waterproof compact
cameras, have made image collection more accessible than ever. Participants may share their
YOY rockfish identifications with project leaders for feedback (rockfishID@noaa.gov).
Participants benefit through increased confidence and improving rockfish identification, regular
communication with project leaders, and shared images represent a dataset on identification
accuracy. Review of these data may facilitate development of correction factors, flag spurious
observations worth additional investigation, and/or highlight identification aspects worthy of
additional attention in outreach materials.

Liability and Diver Safety
All divers conducting surveys for this program are doing so under their own liability or that of their home
institution. Unless maintaining active diving authorization with an institution (e.g., government agency,
university, or NGO), participants should assume they are diving of their own free will and liability. It is
recommended that participants be healthy enough to dive, do so with a buddy, and make informed
decisions regarding conditions and equipment. Insurance through the Divers Alert Network (DAN) or
another carrier is also recommended.
21

Statistical Methodology
The primary objective of this monitoring plan is to quantify estimates and uncertainty of recruitment
across spatial management units for ESA-listed rockfish in the Puget Sound/Georgia Basin DPSs. In this
context, recruitment refers to the annual abundance of YOY that are observed shortly (weeks-tomonths) after they settle from the pelagic environment as larvae into benthic habitats. Estimates of ESAlisted rockfish recruitment provides managers that are evaluating downlisting and delisting criteria with
an indication of whether the current adult population is likely to increase, decrease, or remain the same
in the near future. Estimates of other species of YOY rockfish in the Puget Sound region may be useful to
other agencies and organizations tasked with monitoring and management of these species within their
respective jurisdictions.
One of the main hurdles in estimating YOY rockfish is the extremely high variability observed over space
and time associated with life-history characteristics highlighted above. To adequately quantify this
variation and estimate an index of abundance with any confidence, a large number of sampling events is
required annually. The ability to use data collected from as many sources as possible, including data
from a combination of professional and citizen science surveys, is critical. There are numerous analytical
challenges in using data collected across multiple survey programs using different survey methods with
disparate levels of detectability; however, there are statistical methods available to address these
challenges and provide estimates with appropriate levels of uncertainty.
There are multiple sources of variation that contribute to the number of YOY rockfish observed during
surveys or other sampling efforts. These include spatial, temporal, environmental, demographic, and
methodological elements. To the extent possible, each of these sources of variation should be
accounted for in our modeling framework and estimates of recruitment. Spatially, the abundance of YOY
rockfish can vary across multiple geographic or management boundaries (e.g., biogeographic and
oceanographic basins; state, federal and international boundaries), habitat types (e.g., kelp forests,
seagrass meadows, rocky reefs, unstructured), and depths. The Rockfish Recovery Plan (NMFS 2017)
identified two populations (Hood Canal and non-Hood Canal) and five management units (four in U.S.
waters and one in Canadian) for yelloweye rockfish and bocaccio that we consider here. Temporally, the
abundance of YOY rockfish varies across years and seasonally, in part due to species-specific dispersal
timing. Environmentally, the growth, survival, and abundance of YOY rockfish may vary with the
productivity within regions during larval dispersal (e.g., temperature, prey availability) and settlement
(e.g., habitat quality, competition, and predator-prey dynamics). The abundance of YOY rockfish will also
vary at spatiotemporal scales relevant to the abundance and size of mature adults. Finally, the number
of YOY rockfish observed will vary across sampling groups (e.g., professional, citizen science) and
sampling methods (e.g., SCUBA strip transects, SCUBA timed-roving, capture techniques). All of these
considerations make it important to develop a flexible modeling framework that can integrate multiple
data sources and account for the varying levels and hierarchy of variation.

Sources of data
There were five sources of data identified to help estimate an index of recruitment for ESA-listed
rockfish in the PSGB DPS. These sources were classified by survey group and survey method. Survey
22

groups were categorized into two classifications: professional and citizen science. Survey methods were
categorized into roving SCUBA survey, timed-roving SCUBA survey, and band-transect SCUBA surveys.
Roving SCUBA surveys: These surveys are primarily conducted by volunteer citizen science SCUBA divers
and are associated with data available from the Reef Environmental Education Foundation (REEF).
Volunteer divers “swim freely throughout a site and record every observed fish species that can be
positively identified” (REEF Survey Protocols 2022). Divers also record one of four abundance categories
based on how many fish of each species were observed: Single (1 individual), Few (2-10), Many (11-100),
or Abundant (>100). Divers collecting these data span a wide range of expertise levels and produce
qualitative measures of abundance at a site.
Timed-roving surveys: These surveys are conducted by both professional (e.g., NOAA Western Regional
Office, SeaDoc Society) and citizen science (e.g., Harbor WildWatch, Emerald Dive Club, Ocean Wise
Research Institute) groups. The design of these surveys is described above and quantitative counts of
YOY per unit of time surveyed are produced.
Band-transect surveys: These surveys are conducted by both professional (e.g., NWFSC, WDFW, SA) and
citizen science (e.g., Reef Check) groups. Specific survey design varies among survey teams, but all
consist of SCUBA divers swimming a known distance (e.g., 30 or 100 m) along a measuring tape or
otherwise defined transect and counting only fish that are observed within a known width (e.g., 0.5 or 1
m) and height (e.g., 2 m) of the measurement tape. Some surveys count and record the numbers of fish
in real time, while some surveys are recorded with video and fish are subsequently counted in the
laboratory. These surveys result in quantitative counts of YOY per unit volume (e.g., 120 m3) for each
transect surveyed.

Model framework
Here, we developed a flexible, state-space hierarchical statistical modeling framework that can:
(1) incorporate and share information across each of the sources of data and variables,
(2) address each of the main sources of uncertainty, and
(3) quantify an index of abundance for YOY rockfish across spatial management basins in the
PSGB DPSs.
We constructed two models that account for variation across years, management basins, sites and
survey groups and produce an annual index of YOY abundance across different spatial management
frameworks. First, we estimated an annual index across and for each of the four U.S. waters’
management units as outlined in the Rockfish Recovery Plan for yelloweye rockfish and bocaccio (NMFS
2017):
Puget Sound/Georgia Basin management units
(1) The San Juan Islands/Strait of Juan de Fuca Basin
(2) Main Basin – includes Central Puget Sound, Admiralty Inlet, and Whidbey Basin
(3) South Puget Sound
(4) Hood Canal
(5) The Canadian portion of the DPS (excluded due to lack of data)
23

Secondly, we estimated an annual YOY index for all rockfish species across and for each of the two
spatial populations identified in the Rockfish Recovery Plan for yelloweye rockfish and bocaccio (NMFS
2017). For yelloweye rockfish only, the Plan separates the Hood Canal basin from the rest of the
management units within the PSGB DPS:
Puget Sound/Georgia Basin DPS yelloweye rockfish populations
(1) non-Hood Canal (includes management units 1-3 and 5 from above, but 5 was not included
here due to lack of data)
(2) Hood Canal
As more data become available over time, we will continue to produce indices across the two spatial
management frameworks and will increase the complexity of the model to incorporate multiple
sampling methods and progressively account for each of the remaining sources of uncertainty (e.g.,
seasonal, habitat type, environmental, demographic, and methodological).

Version 1 model
The statistical model is based on counts of YOY rockfish for each sampling event, where an event is
either a single roving or timed-roving survey or a single transect. These sampling events directly observe
YOY rockfish, so we model the observed count 𝑍𝑍𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 of YOY rockfish per sampling event 𝑖𝑖 at site 𝑠𝑠
within basin 𝑏𝑏 (where basin describes the spatial management units of each of the two spatial
management frameworks) in year 𝑦𝑦 for each survey method 𝑚𝑚 as
Observation model:

𝑍𝑍𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ~ 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁(𝑒𝑒𝑒𝑒𝑒𝑒(𝜆𝜆𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ), 𝜈𝜈𝑚𝑚 )

where 𝜆𝜆𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 represents the natural logarithm of mean abundance for each sampling event, and 𝑣𝑣
controls the amount of overdispersion (greater variability) for each survey method. The mean and
variance of the Negative Binomial is calculated as

and

Mean[Z] = 𝑒𝑒𝑒𝑒𝑒𝑒�𝜆𝜆𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 �

Var[Z] = 𝑒𝑒𝑒𝑒𝑒𝑒�𝜆𝜆𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 � +

𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 �𝜆𝜆𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 � 2
𝜈𝜈𝑚𝑚

,

respectively. We estimate a single, shared overdispersion parameter 𝜈𝜈 in the Version 1 model, but as
more data from more survey methods become available, each method will have independent priors for
the overdispersion parameter. We used the Negative Binomial distribution to capture the highly skewed
distribution of counts observed in YOY surveys, including a large number of zero counts. This type of
dataset does not meet the assumptions required for normally distributed statistical methods, even after
various transformations. The negative binomial distribution accounts for the relatively discrete nature of
these observations and uses an appropriate residual distribution.

24

In Version 1 of the model, we used data from professional and citizen science timed-roving sampling
events, and modeled the expected log density 𝜆𝜆 for each sampling event as
𝜆𝜆𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑋𝑋𝑠𝑠𝑠𝑠𝑠𝑠 + 𝑙𝑙𝑙𝑙𝑙𝑙 𝛦𝛦𝑖𝑖𝑖𝑖 + 𝛼𝛼𝑚𝑚

where 𝑋𝑋𝑠𝑠𝑠𝑠𝑠𝑠 is the expected log density for each site-basin-year combination, 𝛦𝛦𝑖𝑖𝑖𝑖 is the survey effort
value (i.e., number of minutes) for each sampling event-survey method combination, and 𝛼𝛼𝑚𝑚 is an offset
for differences between survey methods, where 𝛼𝛼𝑚𝑚=1 = 0 for identifiability, 𝑚𝑚1 represents professional
timed roving surveys, and 𝑚𝑚2 represents citizen science timed roving surveys.

The Version 1 model focuses primarily on understanding the variability in, and quantifying an annual
index of, YOY abundance across and for each basin-year combination, while accounting for variation of
sites within each basin. We model these processes as
Process model:

Fixed effects:

Random effects:

𝑋𝑋𝑠𝑠𝑠𝑠𝑠𝑠 = 𝛾𝛾𝑏𝑏 + 𝛿𝛿𝑦𝑦 + 𝜓𝜓𝑏𝑏𝑏𝑏 + 𝜅𝜅𝑠𝑠𝑠𝑠
𝛾𝛾𝑏𝑏 ~ 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁(−4, 8)
𝛿𝛿𝑦𝑦 ~ 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑎𝑎𝑎𝑎(0, 𝜎𝜎 2 )
𝜓𝜓𝑏𝑏𝑏𝑏 ~ 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁(0, 𝜔𝜔2 )
𝜅𝜅𝑠𝑠𝑠𝑠 ~ 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁(0, 𝜁𝜁 2 )

where the expected mean density 𝑋𝑋𝑠𝑠𝑠𝑠𝑠𝑠 is in log space, 𝛾𝛾𝑏𝑏 is the only fixed effects term and is the mean
density for each basin (spatial management framework) across all years sampled, 𝛿𝛿𝑦𝑦 represents year-toyear variation measured across all basins, 𝜓𝜓𝑏𝑏𝑏𝑏 represents variation associated with each basin-year
combination, and 𝜅𝜅𝑠𝑠𝑠𝑠 represents the variation of spatial nesting of sites within each basin.

Estimation

We estimate the parameters of the statistical model using Stan, a Hamiltonian Markov Chain Monte
Carlo (MCMC) sampler for Bayesian statistical models (Gelman et al. 2015; Carpenter et al. 2017), as
implemented with the rstan package (v.2.21.2) in the R environment (Stan Development Team 2020; R
Core Team 2020). We use 5 parallel chains with diffuse starting locations and examine Gelman-Rubin
diagnostics to ensure convergence and adequate mixing among chains. We use diffuse prior
distributions for all parameters and will refine priors as additional data are collected and analyzed. In
Version 1 of the model the priors are defined as
𝛼𝛼 ~ 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁(−0.5, 3)
𝜈𝜈𝑚𝑚 ~ 𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺(5, 5)
𝜎𝜎 2 ~ 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁(0, 1)
𝜔𝜔2 ~ 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁(0, 1)
𝜁𝜁 2 ~ 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁(0, 1)
25

We present the results of this version of the model in Appendix A.

Future iterations
As more data become available, there are two additional components that we will add to the model
framework. First, when strip transect data are available, these survey data will enter into the model as
additional categories of 𝑚𝑚. Over the course of the next five years, we expect two additional survey
method categories: professional strip transects and citizen science strip transects. Because these survey
methods will use a different survey effort metric (counts/m3 as opposed to counts/min) than the timedroving surveys, the model needs a way to convert to common units. In order to be consistent with
surveys of adult populations in the PS/GB DPSs, the goal will be to quantify expected density estimates
in counts/m2. This will be implemented in the model as
𝜆𝜆𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑋𝑋𝑠𝑠𝑠𝑠𝑠𝑠 + 𝑙𝑙𝑙𝑙𝑙𝑙 𝛦𝛦𝑖𝑖𝑖𝑖 + 𝛼𝛼𝑚𝑚 + 𝛽𝛽𝑚𝑚

where 𝛽𝛽 is an offset that converts counts/min to counts/m3. In this more complex model, the survey
methods will be 𝑚𝑚1 = professional strip transect surveys, 𝑚𝑚2 = citizen science strip transect surveys, 𝑚𝑚3
= professional timed-roving surveys, and 𝑚𝑚4 = citizen science timed-roving surveys where 𝛼𝛼𝑚𝑚=1,3 = 0
and 𝛽𝛽𝑚𝑚=1,2 = 0 for identifiability and the prior on 𝛽𝛽𝑚𝑚=3,4 will be
𝛽𝛽𝑚𝑚=3,4 ~ 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁(−3, 0.5)

Second, there is considerable variation in YOY rockfish abundance across habitat types throughout their
ranges. To account for this variation, we will add habitat type as a fixed effect covariate 𝜌𝜌ℎ to this
framework as
𝑋𝑋𝑠𝑠𝑠𝑠𝑠𝑠 = 𝛾𝛾𝑏𝑏 + 𝛿𝛿𝑦𝑦 + 𝜓𝜓𝑏𝑏𝑏𝑏 + 𝜅𝜅𝑠𝑠𝑠𝑠 + 𝜌𝜌ℎ

As more data becomes available, it is within the capability of the framework to add and examine
environmental and demographic parameters as well.

26

Literature Cited
Andrews, K., B. Bartos, C.J. Harvey, D. Tonnes, M. Bhuthimethee, and P. MacCready. 2021. Testing the
potential for larval dispersal to explain connectivity and population structure of threatened rockfish
species in Puget Sound. Marine Ecology Progress Series 677: 95-113.
Berry, H., W.W. Raymond, D. Claar, P. Dowty, E. Spaulding, B. Christiansen, L. Ferrier, J. Ledbetter, N.
Naar, T. Woodard, C. Palmer-McGee, T. Cowdrey, D. Oster, S. Shull, T. Mumford, and M. Dethier.
2022. Monitoring program design and data assessment protocols for floating kelp monitoring in
Washington State. 52 pp.
Blaine, J., D. Lowry, and R. Pacunski. 2020. 2002-2007 WDFW scientific bottom trawl surveys in the
southern Salish Sea: species distributions, abundance, and population trends. Washington
Department of Fish and Wildlife, Olympia, WA. FPT 20-01. 90 pp. + App.
Buckley, R.M. 1997. Substrate associated recruitment of juvenile Sebastes in artificial reef and natural
habitats in Puget Sound and the San Juan Archipelago, Washington. University of Washington School
of Fisheries, Doctoral Dissertation. 320 pp. + App.
Calloway, M., D. Oster, H. Berry, T. Mumford, N. Naar, B. Peabody, L. Hart, D. Tonnes, S. Copps, J.
Selleck, B. Allen and J. Toft. 2020. Puget Sound kelp conservation and recovery plan. Prepared for
NOAA-NMFS, Seattle, WA. 52 pp.
Carpenter, B., A. Gelman, M.D. Hoffman, D. Lee, B. Goodrich, M. Betancourt, M. Brubaker, J. Guo, P. Li,
and A. Riddell. 2017. Stan: A Probabilistic Programming Language. Journal of Statistical Software
76(1): 1–32. https://doi.org/10.18637/jss.v076.i01
Casselle, J.E., M.H. Carr, D.P. Malone, J.R. Wilson and D.E. Wendt. 2010. Can we predict interannual and
regional variation in delivery of pelagic juveniles to nearshore populations of rockfishes (Genus
Sebastes) using simple proxies of ocean conditions? Reports of California Cooperative Oceanic
Fisheries Investigations 51: 91-105.
Dauble, A.D., S.A. Heppell, and M.L. Johansson. 2012. Settlement patterns of young-of-the-year rockfish
among six Oregon estuaries experiencing different levels of human development. Marine Ecology
Progress Series 448: 143-154. https://doi.org/10.3354/meps09504.
Doherty, P., and T. Fowler. 1994. An empirical test of recruitment limitation in a coral reef fish. Science
263: 935-939.
Drake, J.S., E.A. Berntson, J.M. Cope, R.G. Gustafson, E.E. Holmes, P.S. Levin, N. Tolimieri, R.S. Waples,
S.M. Sogard, and G.D. Williams. 2010. Status review of five rockfish species in Puget Sound,
Washington: bocaccio (Sebastes paucispinis), canary rockfish (S. pinniger), yelloweye rockfish (S.
ruberrimus), greenstriped rockfish (S. elongatus), and redstripe rockfish (S. proriger). U.S. Dept.
Commer., NOAA Tech. Memo. NMFS-NWFSC-108, 234 pp.
Field, J.C., R.R. Miller, J.A. Santora, N. Tolimieri, M.A. Haltuch, R.D. Brodeur, T.D. Auth, E.J. Dick, M.H.
Monk, K.M. Sakuma and B.K. Wells. 2021. Spatiotemporal patterns of variability in the abundance
and distribution of winter-spawned pelagic juvenile rockfish in the California Current. PloS one
16(5): e0251638.
Frierson, T., D. Lowry, L. LeClair, L. Hillier, R. Pacunski, J. Blaine, A. Hennings, A. Phillips, and M. Millard.
2018. Final assessment of Threatened and Endangered juvenile rockfish presence and occurrence of
their nearshore critical habitat adjacent to the NAVBASE Kitsap Bangor & NAVMAG Indian Island:
2017 Survey Results. For Cooperative Agreements N44255-16-2-0003. WDFW, Marine Fish Science
Unit, Olympia, WA. 22 pp.
27

Gelman, A., D. Lee, and J. Guo. 2015. Stan: a probabilistic programming language for Bayesian inference
and optimization. Journal of Educational and Behavioral Statistics 40 (5): 530–543.
Haggarty, D.R., K.E. Lotterhos, and J.B. Shurin. 2017. Young-of-the-year recruitment does not predict the
abundance of older age classes in black rockfish in Barkley Sound, British Columbia, Canada. Marine
Ecology Progress Series 574: 113-126. https://doi.org/10.3354/meps12202.
Hallacher, L.E. and D.A. Roberts. 1985. Differential utilization of space and food by the inshore rockfishes
(Scorpaenidate: Sebastes) of Carmel Bay, California. Environmental Biology of Fishes 12: 91-110.
Kashef, N.S., S.M. Sogard, R. Fisher and J.L. Largier. 2014. Ontogeny of critical swimming speeds for
larval and pelagic juvenile rockfishes (Sebastes spp., family Scorpaenidea). Marine Ecology Progress
Series 500: 231-243.
Laidig, T.E., J.R. Chess and D.F. Howard. 2007. Relationship between abundance of juvenile rockfishes
(Sebastes spp.) and environmental variables documented off northern California and potential
mechanisms for covatiation. Fishery Bulletin 105: 39-48
LeClair, L.L., R. Pacunski, L. Hillier, J. Blaine, and D. Lowry. 2018. Summary of findings from periodic
scuba surveys of bottomfish conducted over a sixteen-year period at six nearshore sites in central
Puget Sound. Washington Department of Fish and Wildlife, Olympia, WA. FPT 18-04. 51 pp. + App.
Love, M.S., M. Yoklavich and L.K. Thorsteinson. 2002. The rockfishes of the Northeast Pacific. University
of California Press. 404 pp.
Lowry, D., R. Pacunski, A. Hennings, J. Blaine, T. Tsou, L. Hillier, J. Beam, and E. Wright. 2022. Assessing
bottomfish and select invertebrate occurrence, abundance, and habitat associations in the U.S.
Salish sea with a small, remotely operated vehicle: Results of the 2012-13 systematic survey.
Olympia, WA: Washington Department of Fish and Wildlife. 67 pp.
Markel, R.W., K.E. Lotterhos, and C.L.K. Robinson. 2017. Temporal variability in the environmental and
geographic predictors of spatial-recruitment in nearshore rockfishes. Marine Ecology Progress Series
574: 97-111. https://doi.org/10.3354/meps12120.
Markel, R.W. and J.B. Shurin. 2020. Contrasting effects of coastal upwelling on growth and recruitment
of nearshore Pacific rockfishes (genus Sebastes). Canadian Journal of Fisheries and Aquatic Sciences
77(6): 950-962.
Moser, H.G. and G.W. Boehlert. 1991. Ecology of pelagic larvae and juveniles of the genus Sebastes.
Environmental Biology of Fishes 30: 203-224.
National Marine Fisheries Service (NMFS). 2010. Endangered and Threatened Wildlife and Plants:
Threatened Status for the Puget Sound/Georgia Basin Distinct Population Segments of Yelloweye
and Canary Rockfish and Endangered Status for the Puget Sound/Georgia Basin Distinct Population
Segment of Bocaccio Rockfish. 75 FR 22276: 22276-22290.
NMFS. 2017. Rockfish recovery plan: Puget Sound/Georgia Basin yelloweye rockfish (Sebastes
ruberrimus) and bocaccio (Sebastes paucispinis). National Marine Fisheries Service. Seattle, WA.
National Oceanic and Atmospheric Administration (NOAA). 2021. NOAA citizen science strategy:
applying the power of the crowd. 9 pp.
NOAA. 2022. NOAA Mitigation Policy for Trust Resources. Administrative Order 216-123. Effective date
July 22, 2022. 10 pp.
Obaza, A., A. Bird, J. Selleck, and D. Tonnes. 2021. Results from young-of-the-year rockfish surveys in the
southern Salish Sea 2015-2020. Seattle, WA: National Marine Fisheries Service. 23 pp.
Pacunski, R., D. Lowry, J. Selleck, A. Beam, A. Hennings, E. Wright, L. Hillier, W. Palsson, and T.-S. Tsou.
2020. Quantification of bottomfish populations, and species-specific habitat associations, in the San
28

Juan Islands, WA employing a remotely operated vehicle and a systematic survey design.
Washington Department of Fish and Wildlife, Olympia, WA. FPT 20-07. 35 pp. + App.
Pacunski, R.E., W. Palsson, and H.G. Greene. 2013. Estimating fish abundance and community
composition on rocky habitats in the San Juan Islands using a small remotely operated vehicle.
Olympia, WA: Washington Department of Fish and Wildlife. No. FPT 13-02. 57 pp.
Palsson, W.A., T.S. Tsou, G.G. Bargmann, R.M. Buckley, J.E. West, M.L. Mills, Y.W. Cheng and R.E.
Pacunski. 2009. The biology and assessment of rockfishes in Puget Sound. Washington Department
of Fish and Wildlife, Fish Management Division. Olympia, Washington.
Pietsch, T. W. and J.W. Orr. 2015. Fishes of the Salish Sea: a compilation and distributional analysis.
NOAA Professional Paper NMFS 18. 106 pp. doi: 10.755/PP.18.
R Core Team. 2020 R: A language and environment for statistical computing. R Foundation for Statistical
Computing, Vienna, Austria. URL https://www.R-project.org/.
Stan Development Team. 2020. RStan: the R interface to Stan. R package version 2.21.2. http://mcstan.org/.
Ralston, S., K.M. Sakuma, and J.C. Field. 2013. Interannual variation in pelagic juvenile rockfish (Sebastes
spp.) abundance – going with the flow. Fisheries Oceanography 22(4): 288-308
Schroeder, I.D., J.A. Santora, S.J. Bograd, E.L. Hazen, K.M. Sakuma, A.M. Moore, C.A. Edwards, B.K. Wells
and A.M. Moore. 2019. Source water variability as a driver of rockfish recruitment in the California
Current Ecosystem: implications for climate change and fisheries management. Canadian Journal of
Fisheries and Aquatic Sciences 76(6): 950-960.
Stachura, M.M., T.E. Essington, N.J. Mantua, A.B. Hollowed, M.A. Haltuch, P.D. Spencer, T.A. Branch and
M.J. Doyle. 2014. Linking northeast Pacific recruitment synchrony to environmental variability.
Fisheries Oceanography 23(5): 389-408.
Tolimieri, N., E.E. Holmes, G.D. Williams, R.P. Pacunski, and D. Lowry. 2017. Population assessment using
multivariate time-series analysis: a case study of rockfishes in Puget Sound. Ecology and Evolution
7(8): 2846-2860.
Tonnes, D., M. Bhuthimethee, J. Sawchuk, N. Tolimieri, K. Andrews, and K. Nichols. 2016. Yelloweye
rockfish (Sebastes ruberrimus), canary rockfish (Sebastes pinniger), and bocaccio (Sebastes
paucispinis) of the Puget Sound/Georgia Basin 5-Year Review: Summary and Evaluation. NOAA’s
NMFS West Coast Region, Seattle, WA. 72 pp. + App.
Washington Administrative Code (WAC). Accessed June 8, 2021. Title 220, Chapter 220-314, Sections
220-314-010 and -020. https://apps.leg.wa.gov/wac/default.aspx?cite=220-314-010 and
https://apps.leg.wa.gov/wac/default.aspx?cite=220-314-020.
Washington Department of Fish and Wildlife (WDFW). 2010. Fishing in Washington: 2010/2011
sportfishing rules pamphlet. WDFW Fish Program, Olympia, WA. 132 pp.
WDFW. 2011. Final Puget Sound rockfish conservation plan policies, strategies and actions. WDFW Fish
Program, Olympia, WA. 33 pp.
WDFW. 2015. Washington’s State Wildlife Action Plan: 2015 update. Washington Department of Fish
and Wildlife. Olympia, WA. 1095 pp.
West, J.E., R.M. Buckley, and D.C. Doty. 1994. Ecology and habitat use of juvenile rockfishes (Sebastes
spp.) associated with artificial reefs in Puget Sound, Washington. Bulletin of Marine Science 55(23):
344-350.
Williams, G.D., P.S. Levin., and W.A. Palsson. 2010. Rockfish in Puget Sound: An ecological history of
exploitation. Marine Policy 34: 1010-1020.
29

Appendix A. Results from Version 1 Model
The first iteration of the model used data available in October 2021 from ‘professional’ and ‘citizen
science’ timed-roving SCUBA surveys from 2015 to 2020. These data consisted of 473 sampling events
from citizen science timed-roving surveys and 535 sampling events from professional timed-roving
sampling events across a total of 62 sites in 4 of the 5 management basins (did not include Canadian
waters) identified in the Recovery Plan for yelloweye rockfish and bocaccio (NMFS 2017).
Estimates of raw counts of YOY rockfish (all species) per unit of time surveyed summarized across each
‘basin-year’ combination for each of the two management frameworks suggests near-zero indices of
YOY abundance for all basins except the Main Basin in the Four-basins model, and the “Rest of DPSs''
management basin in the Two-basins model (Figure A-1). These estimates provide simple summaries of
the annual status and levels of uncertainty for YOY rockfish abundance without accounting for variation
associated with important factors of interest (e.g., site, basin, year).

Figure A-1. Mean (+/-SD) counts of young-of-the-year rockfish per minute surveyed across the (a) four and (b) two
management basin frameworks.

Four-basins management framework
Examination of model diagnostic figures suggests the model: (1) adequately explored the same region of
parameter space across multiple chains with a very small proportion of divergences, and converged on
parameter values (Figure A-2); and (2) that predicted posterior values adequately captured the observed
variation in the raw data (Figure A-3).

A-1

Figure A-2. Trace plot of iterations 5000 – 15000 for each of the primary parameters of the model across five chains
after a 5000-iteration warm-up period.

Figure A-3. Comparison of observed counts of YOY rockfish to the corresponding posterior predictive mean across
10,000 iterations for (a) professional (n = 535) and (b) citizen-science (n = 473) sampling event. The red line is the
one-to-one line and the blue lines are the 95% confidence interval for which we would expect 95% of the data
points to be included.

Final abundance indices suggest YOY abundance was greatest within the Main Basin of Puget Sound,
with the highest densities observed in 2017 (Figure A-4). We also found near-zero YOY abundance in the
Hood Canal and San Juan Islands management basins across all surveyed years. The log-density plot
allows us to explore variation across years for each basin, while the true scale density plot shows the
magnitude of difference in YOY densities across all basins. We used the year term to quantify year-toyear variation across all basins throughout the surveyed region – this showed a very similar trend as
observed in the Main Basin (Figure A-5). The vast majority of survey effort occurred in this basin and
influences the number of opportunities to observe >0 counts of YOY rockfish.

A-2

Figure A-4. Index of YOY rockfish abundance in each management basin from the four management basin
framework from 2015 to 2020 on (a) log scale and (b) normal scale.

Figure A-5. Year-to-year deviations in the mean density
of YOY rockfish across all sampled basins in the Puget
Sound/Georgia Basin DPSs. The deviations are
multiplicatively scaled to a value of 1 (e.g., YOY
densities were ~1.8 times more abundant in 2017 than
the long-term mean across years.

Spatial variation in YOY rockfish density across all sites showed the highest densities were found in the
Main Basin (Figure A-6a). The top 26 (out of 62) highest-density sites were all in the Main Basin, while 8
of the 9 lowest-density sites were found in the San Juan Islands. Similar to patterns observed in SCUBA
surveys by the Washington Department of Fish and Wildlife in 2006 (a historically-high rockfish
recruitment year), we observed the greatest densities of YOY rockfish at sites along the eastern shores
of the Main Basin of Puget Sound, including Keystone (at the entrance to Puget Sound Proper), Edmonds
Underwater Park, two sites along the Mukilteo shoreline, and within Elliott Bay at Alki Cove #2. This type
of relationship may be correlated with prevailing oceanographic currents, larval supply, and quantity
and quality of habitat along this section of the shoreline. Within each basin (Figure A-6b), the magnitude
of density values varied with general oceanographic and geographic locations. In the Main Basin, the
highest densities were found at the northernmost site (Keystone had 6 times the average density of
other Main Basin sites) and along the eastern shoreline, as observed across all sites. The highest
densities in South Puget Sound were found along the southeastern shoreline of Fox Island, which is
located at the southern end of the Tacoma Narrows, a location of very high currents and turbulent
mixing conditions. In Hood Canal, the highest densities were found at the most southern site, which
tends to have relatively mild rates of current exchange and very long water residence time. Finally, the
highest densities in the San Juan Islands basin were found in two disjunct locations: in the center of the
A-3

main island archipelago off Shaw Island; and outside and to the southeast of the archipelago at the
entrance to Deception Pass and at Smith Island. Importantly, several of the sites with high densities
were sites that have been sampled most frequently and may contribute to higher probabilities of
observing YOY rockfish simply due to sampling intensity and diver familiarity. Additionally, spatial
variation within each basin should be placed in context relative to where sampling has occurred. Each of
these cautionary points have been considered in the development of the sampling design (e.g., using
index sites paired with random additional sites) and should be lessened over time and with more data,
but these will be important considerations of future analyses and modifications to the modeling
framework.

Figure A-6. Variation in (a) mean density (count/min) of YOY rockfish across all sites in all basins (𝛾𝛾𝑏𝑏 + 𝜅𝜅𝑠𝑠𝑠𝑠 ) and (b)
mean deviation among sites within individual basins (𝜅𝜅𝑠𝑠𝑠𝑠 ) in the Puget Sound/Georgia Basin DPSs. Site-to-site
deviations are multiplicatively scaled to the mean density (equal to 1) within each basin (e.g., YOY densities at
Edmonds Underwater Park were 6.4 times more abundant than the grand mean of sites within the Main Basin).

Two-basins management framework
Examination of model diagnostic figures suggests the model: (1) adequately explored the same region of
parameter space across multiple chains with a very small proportion of divergences and converged on

A-4

parameter values (Figure A-7); and (2) that predicted posterior values adequately captured the observed
variation in the raw data (Figure A-8).

Figure A-7. Trace plot of iterations 5000 – 15000 for each of the primary parameters of the model across five chains
after a 5000-iteration warm-up period.

Figure A-8. Comparison of observed counts of YOY rockfish to the corresponding posterior predictive mean across
10,000 iterations for (a) professional (n = 535) and (b) citizen-science (n = 473) sampling event. The red line is the
one-to-one line and the blue lines are the 95% confidence interval for which we would expect 95% of the data
points to be included.

Final abundance indices suggest YOY abundance was greatest within the combined management unit
“Rest of DPSs,” which included the Main Basin, San Juan Islands, and South Puget Sound, with the
highest densities observed in 2017 (Figure A-9). We also found near-zero YOY abundance in the Hood
Canal management basin across all surveyed years, but there was a large amount of variation in the
estimates, particularly for 2016. The log-density plot allows us to explore variation across years for each
basin, while the true scale density plot shows the magnitude of difference in YOY densities across all
basins. We used the year term to quantify year-to-year variation across all basins throughout the
A-5

surveyed region, which showed a trend very similar to what was observed in the ‘Rest of DPS’
management basin (Figure A-10). The vast majority of survey effort occurred in the ‘Rest of DPS’ basin
(and largely within Central Sound, within this category) and influences the number of opportunities to
observe >0 counts of YOY rockfish.

Figure A-9. Index of YOY rockfish abundance in each management basin from 2015 to 2020 on (a) log scale and (b)
normal scale.

Figure A-10. Year-to-year deviations in the mean
density of YOY rockfish across all sampled basins in the
Puget Sound/Georgia Basin DPSs. The deviations are
multiplicatively scaled to a value of 1 (e.g., YOY
densities were ~1.6 times more abundant in 2017 than
the long-term mean across years.

Spatial variation in YOY rockfish density across all sites showed that the highest densities were found in
the ‘Rest of DPSs’ (Figure A-11a). The top 54 (out of 62) highest-density sites were all in the ‘Rest of
DPSs’ basin, while the four Hood Canal sites were among the lowest eight sites. Expectedly, the overall
spatial pattern across all sites is the same as observed in the “Four basins” model. Within each basin
(Figure A-11b), the magnitude of density values varies in their general oceanographic and geographic
locations. In the ‘Rest of DPSs’ basin, the highest densities were primarily found within Puget Sound
proper. YOY mean density at the northern entrance to Puget Sound proper was 22 times the average
density across other ‘Rest of DPSs’ sites) and densities at four other sites along the eastern shoreline
were > 10 times average densities in the basin. In Hood Canal, the highest densities were found at the
most southern site, which tends to have relatively mild rates of current exchange and very long water
residence time. Importantly, several of the sites with high densities were sites that have been sampled
most frequently and may contribute to higher probabilities of observing YOY rockfish simply due to
A-6

sampling intensity and diver familiarity. Additionally, spatial variation within each basin should be placed
in context relative to where sampling has occurred. Each of these cautionary points have been
considered in the development of the sampling design (e.g., using index sites paired with random
additional sites) and should be lessened over time and with more data, but these will be important
considerations of future analyses and modifications to the modeling framework.

Figure A-11. Variation in (a) mean density (count/min) of YOY rockfish across all sites in all basins (𝛾𝛾𝑏𝑏 + 𝜅𝜅𝑠𝑠𝑠𝑠 ) and
(b) mean deviation among sites within individual basins (𝜅𝜅𝑠𝑠𝑠𝑠 ) in the Puget Sound/Georgia Basin DPSs. Site-to-site
deviations are multiplicatively scaled to the mean density (equal to 1) within each basin. For example, YOY densities
at Edmonds Underwater Park were 17.2 times more abundant than the grand mean of sites across the ‘Rest of
DPSs’ basin.

Example of future model
We show below an example of how we foresee a fully-developed YOY rockfish recruitment index model
for the Puget Sound/Georgia Basin DPSs. This model would include four survey methods across all five
management basins (including the Canadian portion of the DPSs) and would add an additional covariate
to account for variation across various habitat types that YOY rockfish are observed.

A-7

Observation model:
𝑍𝑍𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ~ 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁(𝑒𝑒𝑒𝑒𝑒𝑒(𝜆𝜆𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ), 𝜈𝜈𝑚𝑚 )
Process model:

Fixed effects:

Random effects:

Offsets for survey methods:

Priors:

𝜆𝜆𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑋𝑋𝑠𝑠𝑠𝑠𝑠𝑠 + 𝑙𝑙𝑙𝑙𝑙𝑙 𝛦𝛦𝑖𝑖𝑖𝑖 + 𝛼𝛼𝑚𝑚 + 𝛽𝛽𝑚𝑚
𝑋𝑋𝑠𝑠𝑠𝑠𝑠𝑠 = 𝛾𝛾𝑏𝑏 + 𝛿𝛿𝑦𝑦 + 𝜓𝜓𝑏𝑏𝑏𝑏 + 𝜅𝜅𝑠𝑠𝑠𝑠 + 𝜌𝜌ℎ
𝛾𝛾𝑏𝑏 ~ 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁( −4 , 8)
𝜌𝜌ℎ ~ 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁( −4 , 8)
𝛿𝛿𝑦𝑦 ~ 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁( 0 , 𝜎𝜎 2 )
𝜓𝜓𝑏𝑏𝑏𝑏 ~ 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁( 0 , 𝜔𝜔2 )
𝜅𝜅𝑠𝑠𝑠𝑠 ~ 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁(0, 𝜁𝜁 2 )
𝛼𝛼𝑚𝑚=2,4 ~ 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁( −0.5 , 3)
𝛽𝛽𝑚𝑚=3,4 ~ 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁(−3, 0.5)
𝜈𝜈𝑚𝑚 ~ 𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝑎𝑎(5 , 5)
𝜎𝜎 2 ~ 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁(0, 1)
𝜔𝜔2 ~ 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁(0, 1)
𝜁𝜁 2 ~ 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁(0, 1)

A-8